commit d1a54268ff2877eba4c17f97e33d9d4f92f2aadc Author: ModelHub XC Date: Fri Aug 28 19:16:40 2026 +0800 初始化项目，由ModelHub XC社区提供模型 Model: Neelectric/Ministral-8B-Instruct-2410_SFT_mathv00.01 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..f6abbe3 --- /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_mathv00.01 +tags: +- generated_from_trainer +- sft +- open-r1 +- trl +licence: license +--- + +# Model Card for Ministral-8B-Instruct-2410_SFT_mathv00.01 + +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_mathv00.01", 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/b1n282ng) + + + +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..39f3b93 --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 5.7509457374759354e+19, + "train_loss": 0.3138526783479623, + "train_runtime": 47000.1043, + "train_samples": 125770, + "train_samples_per_second": 8.028, + "train_steps_per_second": 0.502 +} \ 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..88c741e --- /dev/null +++ b/config.json @@ -0,0 +1,64 @@ +{ + "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, + "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..f573e90 --- /dev/null +++ b/generation_config.json @@ -0,0 +1,6 @@ +{ + "_from_model_config": true, + "bos_token_id": 1, + "eos_token_id": 0, + "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..6a4fb7b --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba80b44e952333e4d9229aa61812d2d14fd5f891ba3e5e537f5f1e6fe38c97b3 +size 4983007904 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..d9ad45e --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c684296a1d05124ef1c8fa5d861a8e76ec0ca942b0ebb77ee7d59bf0716713f4 +size 4999836776 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..9f81b98 --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68a168a45ab00b48a14b6ff05bcc20230abe64d5801a487b625ec2f18f684395 +size 4983067960 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..bd958c0 --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5dcb8c9f8d734f2f4c45d5f8d334837ffa5afee42cb79591d452ce48d0451bd +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..2cc8166 --- /dev/null +++ b/special_tokens_map.json @@ -0,0 +1,17 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_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..22f00f2 --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,8016 @@ +{ + "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, + "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..39f3b93 --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 5.7509457374759354e+19, + "train_loss": 0.3138526783479623, + "train_runtime": 47000.1043, + "train_samples": 125770, + "train_samples_per_second": 8.028, + "train_steps_per_second": 0.502 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..2c31295 --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,212290 @@ +{ + "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, + "grad_norm": 6.939845085144043, + "learning_rate": 0.0, + "loss": 0.7421, + "mean_token_accuracy": 0.7937126755714417, + "num_tokens": 41694.0, + "step": 1 + }, + { + "epoch": 0.00025442055718102023, + "grad_norm": 7.210134983062744, + "learning_rate": 4.2390843577787196e-10, + "loss": 0.7296, + "mean_token_accuracy": 0.7980299592018127, + "num_tokens": 81859.0, + "step": 2 + }, + { + "epoch": 0.0003816308357715303, + "grad_norm": 6.723892688751221, + "learning_rate": 8.478168715557439e-10, + "loss": 0.717, + "mean_token_accuracy": 0.7978752851486206, + "num_tokens": 123057.0, + "step": 3 + }, + { + "epoch": 0.0005088411143620405, + "grad_norm": 7.512058258056641, + "learning_rate": 1.271725307333616e-09, + "loss": 0.73, + "mean_token_accuracy": 0.7957618236541748, + "num_tokens": 160063.0, + "step": 4 + }, + { + "epoch": 0.0006360513929525506, + "grad_norm": 8.969283103942871, + "learning_rate": 1.6956337431114878e-09, + "loss": 0.76, + "mean_token_accuracy": 0.787570059299469, + "num_tokens": 198507.0, + "step": 5 + }, + { + "epoch": 0.0007632616715430606, + "grad_norm": 9.146666526794434, + "learning_rate": 2.1195421788893596e-09, + "loss": 0.7071, + "mean_token_accuracy": 0.7999285459518433, + "num_tokens": 240347.0, + "step": 6 + }, + { + "epoch": 0.0008904719501335708, + "grad_norm": 6.649759292602539, + "learning_rate": 2.543450614667232e-09, + "loss": 0.7359, + "mean_token_accuracy": 0.7938611507415771, + "num_tokens": 284200.0, + "step": 7 + }, + { + "epoch": 0.001017682228724081, + "grad_norm": 7.42015266418457, + "learning_rate": 2.967359050445104e-09, + "loss": 0.7306, + "mean_token_accuracy": 0.7944688200950623, + "num_tokens": 321432.0, + "step": 8 + }, + { + "epoch": 0.001144892507314591, + "grad_norm": 7.905355453491211, + "learning_rate": 3.3912674862229757e-09, + "loss": 0.726, + "mean_token_accuracy": 0.7974302768707275, + "num_tokens": 365572.0, + "step": 9 + }, + { + "epoch": 0.0012721027859051012, + "grad_norm": 6.974149227142334, + "learning_rate": 3.815175922000847e-09, + "loss": 0.7276, + "mean_token_accuracy": 0.7967630624771118, + "num_tokens": 407668.0, + "step": 10 + }, + { + "epoch": 0.0013993130644956112, + "grad_norm": 7.007250785827637, + "learning_rate": 4.239084357778719e-09, + "loss": 0.7145, + "mean_token_accuracy": 0.7994145154953003, + "num_tokens": 446337.0, + "step": 11 + }, + { + "epoch": 0.0015265233430861213, + "grad_norm": 7.19791841506958, + "learning_rate": 4.662992793556591e-09, + "loss": 0.7507, + "mean_token_accuracy": 0.788848876953125, + "num_tokens": 486025.0, + "step": 12 + }, + { + "epoch": 0.0016537336216766315, + "grad_norm": 6.625011444091797, + "learning_rate": 5.086901229334464e-09, + "loss": 0.7778, + "mean_token_accuracy": 0.7820862531661987, + "num_tokens": 527074.0, + "step": 13 + }, + { + "epoch": 0.0017809439002671415, + "grad_norm": 7.4220662117004395, + "learning_rate": 5.510809665112336e-09, + "loss": 0.7286, + "mean_token_accuracy": 0.79592365026474, + "num_tokens": 570268.0, + "step": 14 + }, + { + "epoch": 0.0019081541788576518, + "grad_norm": 7.055915832519531, + "learning_rate": 5.934718100890208e-09, + "loss": 0.7236, + "mean_token_accuracy": 0.7937523722648621, + "num_tokens": 608117.0, + "step": 15 + }, + { + "epoch": 0.002035364457448162, + "grad_norm": 7.287496566772461, + "learning_rate": 6.3586265366680796e-09, + "loss": 0.7464, + "mean_token_accuracy": 0.7938611507415771, + "num_tokens": 647716.0, + "step": 16 + }, + { + "epoch": 0.002162574736038672, + "grad_norm": 11.438251495361328, + "learning_rate": 6.782534972445951e-09, + "loss": 0.7393, + "mean_token_accuracy": 0.7957236766815186, + "num_tokens": 682142.0, + "step": 17 + }, + { + "epoch": 0.002289785014629182, + "grad_norm": 8.195871353149414, + "learning_rate": 7.206443408223823e-09, + "loss": 0.7518, + "mean_token_accuracy": 0.7909266948699951, + "num_tokens": 711505.0, + "step": 18 + }, + { + "epoch": 0.0024169952932196924, + "grad_norm": 8.23206901550293, + "learning_rate": 7.630351844001695e-09, + "loss": 0.7376, + "mean_token_accuracy": 0.7971176505088806, + "num_tokens": 749696.0, + "step": 19 + }, + { + "epoch": 0.0025442055718102024, + "grad_norm": 7.470865726470947, + "learning_rate": 8.054260279779567e-09, + "loss": 0.7685, + "mean_token_accuracy": 0.7847698926925659, + "num_tokens": 786038.0, + "step": 20 + }, + { + "epoch": 0.0026714158504007124, + "grad_norm": 8.133748054504395, + "learning_rate": 8.478168715557438e-09, + "loss": 0.732, + "mean_token_accuracy": 0.7989915013313293, + "num_tokens": 829069.0, + "step": 21 + }, + { + "epoch": 0.0027986261289912225, + "grad_norm": 7.868557453155518, + "learning_rate": 8.902077151335311e-09, + "loss": 0.7416, + "mean_token_accuracy": 0.7941840887069702, + "num_tokens": 867645.0, + "step": 22 + }, + { + "epoch": 0.0029258364075817325, + "grad_norm": 9.321551322937012, + "learning_rate": 9.325985587113182e-09, + "loss": 0.6603, + "mean_token_accuracy": 0.8157471418380737, + "num_tokens": 907379.0, + "step": 23 + }, + { + "epoch": 0.0030530466861722425, + "grad_norm": 7.711219310760498, + "learning_rate": 9.749894022891054e-09, + "loss": 0.8083, + "mean_token_accuracy": 0.7738453149795532, + "num_tokens": 944296.0, + "step": 24 + }, + { + "epoch": 0.003180256964762753, + "grad_norm": 8.78279972076416, + "learning_rate": 1.0173802458668929e-08, + "loss": 0.7742, + "mean_token_accuracy": 0.7827745676040649, + "num_tokens": 984736.0, + "step": 25 + }, + { + "epoch": 0.003307467243353263, + "grad_norm": 6.88203239440918, + "learning_rate": 1.05977108944468e-08, + "loss": 0.7433, + "mean_token_accuracy": 0.7929452657699585, + "num_tokens": 1026105.0, + "step": 26 + }, + { + "epoch": 0.003434677521943773, + "grad_norm": 9.973653793334961, + "learning_rate": 1.1021619330224672e-08, + "loss": 0.751, + "mean_token_accuracy": 0.7920303344726562, + "num_tokens": 1067778.0, + "step": 27 + }, + { + "epoch": 0.003561887800534283, + "grad_norm": 7.25452184677124, + "learning_rate": 1.1445527766002543e-08, + "loss": 0.6979, + "mean_token_accuracy": 0.805382490158081, + "num_tokens": 1108673.0, + "step": 28 + }, + { + "epoch": 0.003689098079124793, + "grad_norm": 7.296760559082031, + "learning_rate": 1.1869436201780416e-08, + "loss": 0.7715, + "mean_token_accuracy": 0.7825301885604858, + "num_tokens": 1149629.0, + "step": 29 + }, + { + "epoch": 0.0038163083577153036, + "grad_norm": 7.239584922790527, + "learning_rate": 1.2293344637558287e-08, + "loss": 0.7644, + "mean_token_accuracy": 0.7857990264892578, + "num_tokens": 1189114.0, + "step": 30 + }, + { + "epoch": 0.003943518636305814, + "grad_norm": 10.769287109375, + "learning_rate": 1.2717253073336159e-08, + "loss": 0.7295, + "mean_token_accuracy": 0.7978154420852661, + "num_tokens": 1226932.0, + "step": 31 + }, + { + "epoch": 0.004070728914896324, + "grad_norm": 7.206960201263428, + "learning_rate": 1.314116150911403e-08, + "loss": 0.7464, + "mean_token_accuracy": 0.7878912687301636, + "num_tokens": 1265657.0, + "step": 32 + }, + { + "epoch": 0.004197939193486834, + "grad_norm": 8.522723197937012, + "learning_rate": 1.3565069944891903e-08, + "loss": 0.7689, + "mean_token_accuracy": 0.7838711738586426, + "num_tokens": 1302277.0, + "step": 33 + }, + { + "epoch": 0.004325149472077344, + "grad_norm": 7.990650653839111, + "learning_rate": 1.3988978380669775e-08, + "loss": 0.7469, + "mean_token_accuracy": 0.7937228083610535, + "num_tokens": 1344042.0, + "step": 34 + }, + { + "epoch": 0.004452359750667854, + "grad_norm": 6.441711902618408, + "learning_rate": 1.4412886816447646e-08, + "loss": 0.6929, + "mean_token_accuracy": 0.8024111986160278, + "num_tokens": 1383391.0, + "step": 35 + }, + { + "epoch": 0.004579570029258364, + "grad_norm": 6.847570896148682, + "learning_rate": 1.4836795252225519e-08, + "loss": 0.7599, + "mean_token_accuracy": 0.7877001762390137, + "num_tokens": 1419186.0, + "step": 36 + }, + { + "epoch": 0.004706780307848874, + "grad_norm": 7.617924213409424, + "learning_rate": 1.526070368800339e-08, + "loss": 0.7498, + "mean_token_accuracy": 0.7912587523460388, + "num_tokens": 1459720.0, + "step": 37 + }, + { + "epoch": 0.004833990586439385, + "grad_norm": 7.491961479187012, + "learning_rate": 1.5684612123781262e-08, + "loss": 0.753, + "mean_token_accuracy": 0.7925194501876831, + "num_tokens": 1505579.0, + "step": 38 + }, + { + "epoch": 0.004961200865029895, + "grad_norm": 7.396632194519043, + "learning_rate": 1.6108520559559135e-08, + "loss": 0.7455, + "mean_token_accuracy": 0.7942435145378113, + "num_tokens": 1547078.0, + "step": 39 + }, + { + "epoch": 0.005088411143620405, + "grad_norm": 8.55605411529541, + "learning_rate": 1.6532428995337004e-08, + "loss": 0.8175, + "mean_token_accuracy": 0.7778333425521851, + "num_tokens": 1580887.0, + "step": 40 + }, + { + "epoch": 0.005215621422210915, + "grad_norm": 6.366801738739014, + "learning_rate": 1.6956337431114877e-08, + "loss": 0.7173, + "mean_token_accuracy": 0.8055402040481567, + "num_tokens": 1621356.0, + "step": 41 + }, + { + "epoch": 0.005342831700801425, + "grad_norm": 7.797378063201904, + "learning_rate": 1.738024586689275e-08, + "loss": 0.6957, + "mean_token_accuracy": 0.8068737983703613, + "num_tokens": 1660248.0, + "step": 42 + }, + { + "epoch": 0.005470041979391935, + "grad_norm": 7.113061904907227, + "learning_rate": 1.7804154302670622e-08, + "loss": 0.6961, + "mean_token_accuracy": 0.8066599369049072, + "num_tokens": 1699853.0, + "step": 43 + }, + { + "epoch": 0.005597252257982445, + "grad_norm": 11.127191543579102, + "learning_rate": 1.8228062738448494e-08, + "loss": 0.825, + "mean_token_accuracy": 0.771478533744812, + "num_tokens": 1735508.0, + "step": 44 + }, + { + "epoch": 0.005724462536572955, + "grad_norm": 7.124879837036133, + "learning_rate": 1.8651971174226364e-08, + "loss": 0.708, + "mean_token_accuracy": 0.8016505241394043, + "num_tokens": 1771925.0, + "step": 45 + }, + { + "epoch": 0.005851672815163465, + "grad_norm": 7.102363586425781, + "learning_rate": 1.9075879610004236e-08, + "loss": 0.6591, + "mean_token_accuracy": 0.8119369745254517, + "num_tokens": 1807424.0, + "step": 46 + }, + { + "epoch": 0.005978883093753975, + "grad_norm": 8.515996932983398, + "learning_rate": 1.949978804578211e-08, + "loss": 0.7746, + "mean_token_accuracy": 0.7851760387420654, + "num_tokens": 1849843.0, + "step": 47 + }, + { + "epoch": 0.006106093372344485, + "grad_norm": 6.718783855438232, + "learning_rate": 1.9923696481559985e-08, + "loss": 0.7023, + "mean_token_accuracy": 0.8022018074989319, + "num_tokens": 1887236.0, + "step": 48 + }, + { + "epoch": 0.006233303650934996, + "grad_norm": 8.995641708374023, + "learning_rate": 2.0347604917337857e-08, + "loss": 0.762, + "mean_token_accuracy": 0.7906613349914551, + "num_tokens": 1922724.0, + "step": 49 + }, + { + "epoch": 0.006360513929525506, + "grad_norm": 7.340915679931641, + "learning_rate": 2.0771513353115727e-08, + "loss": 0.7073, + "mean_token_accuracy": 0.804272472858429, + "num_tokens": 1961041.0, + "step": 50 + }, + { + "epoch": 0.006487724208116016, + "grad_norm": 7.759606838226318, + "learning_rate": 2.11954217888936e-08, + "loss": 0.7555, + "mean_token_accuracy": 0.7852110266685486, + "num_tokens": 2003182.0, + "step": 51 + }, + { + "epoch": 0.006614934486706526, + "grad_norm": 6.813400745391846, + "learning_rate": 2.1619330224671472e-08, + "loss": 0.6765, + "mean_token_accuracy": 0.8093680143356323, + "num_tokens": 2041191.0, + "step": 52 + }, + { + "epoch": 0.006742144765297036, + "grad_norm": 8.70772933959961, + "learning_rate": 2.2043238660449344e-08, + "loss": 0.7209, + "mean_token_accuracy": 0.7963791489601135, + "num_tokens": 2082627.0, + "step": 53 + }, + { + "epoch": 0.006869355043887546, + "grad_norm": 7.636796951293945, + "learning_rate": 2.2467147096227214e-08, + "loss": 0.7249, + "mean_token_accuracy": 0.7984099984169006, + "num_tokens": 2121659.0, + "step": 54 + }, + { + "epoch": 0.006996565322478056, + "grad_norm": 7.753715991973877, + "learning_rate": 2.2891055532005086e-08, + "loss": 0.7014, + "mean_token_accuracy": 0.8020628690719604, + "num_tokens": 2166374.0, + "step": 55 + }, + { + "epoch": 0.007123775601068566, + "grad_norm": 12.752725601196289, + "learning_rate": 2.331496396778296e-08, + "loss": 0.7144, + "mean_token_accuracy": 0.802880048751831, + "num_tokens": 2200642.0, + "step": 56 + }, + { + "epoch": 0.007250985879659076, + "grad_norm": 6.494483947753906, + "learning_rate": 2.373887240356083e-08, + "loss": 0.7188, + "mean_token_accuracy": 0.7970597743988037, + "num_tokens": 2241850.0, + "step": 57 + }, + { + "epoch": 0.007378196158249586, + "grad_norm": 8.167981147766113, + "learning_rate": 2.4162780839338704e-08, + "loss": 0.7417, + "mean_token_accuracy": 0.7896934747695923, + "num_tokens": 2282243.0, + "step": 58 + }, + { + "epoch": 0.007505406436840096, + "grad_norm": 7.60552978515625, + "learning_rate": 2.4586689275116573e-08, + "loss": 0.7958, + "mean_token_accuracy": 0.7735310196876526, + "num_tokens": 2318864.0, + "step": 59 + }, + { + "epoch": 0.007632616715430607, + "grad_norm": 8.643433570861816, + "learning_rate": 2.5010597710894446e-08, + "loss": 0.7331, + "mean_token_accuracy": 0.7981340885162354, + "num_tokens": 2357981.0, + "step": 60 + }, + { + "epoch": 0.007759826994021117, + "grad_norm": 7.8196282386779785, + "learning_rate": 2.5434506146672318e-08, + "loss": 0.7549, + "mean_token_accuracy": 0.7877918481826782, + "num_tokens": 2392625.0, + "step": 61 + }, + { + "epoch": 0.007887037272611627, + "grad_norm": 7.062349319458008, + "learning_rate": 2.585841458245019e-08, + "loss": 0.711, + "mean_token_accuracy": 0.800778865814209, + "num_tokens": 2433777.0, + "step": 62 + }, + { + "epoch": 0.008014247551202136, + "grad_norm": 8.37081527709961, + "learning_rate": 2.628232301822806e-08, + "loss": 0.7912, + "mean_token_accuracy": 0.7792073488235474, + "num_tokens": 2473733.0, + "step": 63 + }, + { + "epoch": 0.008141457829792647, + "grad_norm": 7.260584354400635, + "learning_rate": 2.6706231454005933e-08, + "loss": 0.7431, + "mean_token_accuracy": 0.7923746109008789, + "num_tokens": 2512561.0, + "step": 64 + }, + { + "epoch": 0.008268668108383158, + "grad_norm": 7.293780326843262, + "learning_rate": 2.7130139889783805e-08, + "loss": 0.7452, + "mean_token_accuracy": 0.7941263318061829, + "num_tokens": 2546247.0, + "step": 65 + }, + { + "epoch": 0.008395878386973667, + "grad_norm": 6.47797155380249, + "learning_rate": 2.7554048325561678e-08, + "loss": 0.6875, + "mean_token_accuracy": 0.8024738430976868, + "num_tokens": 2586745.0, + "step": 66 + }, + { + "epoch": 0.008523088665564178, + "grad_norm": 10.549086570739746, + "learning_rate": 2.797795676133955e-08, + "loss": 0.7762, + "mean_token_accuracy": 0.783586859703064, + "num_tokens": 2625785.0, + "step": 67 + }, + { + "epoch": 0.008650298944154687, + "grad_norm": 7.172214031219482, + "learning_rate": 2.840186519711742e-08, + "loss": 0.6589, + "mean_token_accuracy": 0.8151118159294128, + "num_tokens": 2669534.0, + "step": 68 + }, + { + "epoch": 0.008777509222745198, + "grad_norm": 10.532620429992676, + "learning_rate": 2.8825773632895292e-08, + "loss": 0.7231, + "mean_token_accuracy": 0.8019028306007385, + "num_tokens": 2707295.0, + "step": 69 + }, + { + "epoch": 0.008904719501335707, + "grad_norm": 8.232490539550781, + "learning_rate": 2.9249682068673165e-08, + "loss": 0.7228, + "mean_token_accuracy": 0.8022891879081726, + "num_tokens": 2742451.0, + "step": 70 + }, + { + "epoch": 0.009031929779926218, + "grad_norm": 6.714896202087402, + "learning_rate": 2.9673590504451037e-08, + "loss": 0.6888, + "mean_token_accuracy": 0.8035791516304016, + "num_tokens": 2784767.0, + "step": 71 + }, + { + "epoch": 0.009159140058516728, + "grad_norm": 7.546891212463379, + "learning_rate": 3.0097498940228907e-08, + "loss": 0.7398, + "mean_token_accuracy": 0.7895712852478027, + "num_tokens": 2826757.0, + "step": 72 + }, + { + "epoch": 0.009286350337107238, + "grad_norm": 8.067140579223633, + "learning_rate": 3.052140737600678e-08, + "loss": 0.7189, + "mean_token_accuracy": 0.7961609363555908, + "num_tokens": 2863349.0, + "step": 73 + }, + { + "epoch": 0.009413560615697748, + "grad_norm": 7.412789344787598, + "learning_rate": 3.094531581178465e-08, + "loss": 0.7812, + "mean_token_accuracy": 0.7807825803756714, + "num_tokens": 2900337.0, + "step": 74 + }, + { + "epoch": 0.009540770894288259, + "grad_norm": 7.879541873931885, + "learning_rate": 3.1369224247562524e-08, + "loss": 0.6757, + "mean_token_accuracy": 0.813334584236145, + "num_tokens": 2936596.0, + "step": 75 + }, + { + "epoch": 0.00966798117287877, + "grad_norm": 7.712625980377197, + "learning_rate": 3.17931326833404e-08, + "loss": 0.704, + "mean_token_accuracy": 0.7995464205741882, + "num_tokens": 2975695.0, + "step": 76 + }, + { + "epoch": 0.009795191451469279, + "grad_norm": 8.02437686920166, + "learning_rate": 3.221704111911827e-08, + "loss": 0.8047, + "mean_token_accuracy": 0.7728497982025146, + "num_tokens": 3009644.0, + "step": 77 + }, + { + "epoch": 0.00992240173005979, + "grad_norm": 6.46244478225708, + "learning_rate": 3.264094955489614e-08, + "loss": 0.7084, + "mean_token_accuracy": 0.7989290952682495, + "num_tokens": 3049024.0, + "step": 78 + }, + { + "epoch": 0.010049612008650299, + "grad_norm": 7.212714672088623, + "learning_rate": 3.306485799067401e-08, + "loss": 0.7446, + "mean_token_accuracy": 0.7906610369682312, + "num_tokens": 3092209.0, + "step": 79 + }, + { + "epoch": 0.01017682228724081, + "grad_norm": 8.574557304382324, + "learning_rate": 3.348876642645188e-08, + "loss": 0.7188, + "mean_token_accuracy": 0.797110915184021, + "num_tokens": 3129221.0, + "step": 80 + }, + { + "epoch": 0.010304032565831319, + "grad_norm": 6.770594596862793, + "learning_rate": 3.391267486222975e-08, + "loss": 0.7192, + "mean_token_accuracy": 0.7921282052993774, + "num_tokens": 3171790.0, + "step": 81 + }, + { + "epoch": 0.01043124284442183, + "grad_norm": 6.045897960662842, + "learning_rate": 3.4336583298007626e-08, + "loss": 0.6788, + "mean_token_accuracy": 0.8075970411300659, + "num_tokens": 3214240.0, + "step": 82 + }, + { + "epoch": 0.010558453123012339, + "grad_norm": 8.279779434204102, + "learning_rate": 3.47604917337855e-08, + "loss": 0.73, + "mean_token_accuracy": 0.7998474836349487, + "num_tokens": 3246494.0, + "step": 83 + }, + { + "epoch": 0.01068566340160285, + "grad_norm": 5.834073543548584, + "learning_rate": 3.518440016956337e-08, + "loss": 0.6899, + "mean_token_accuracy": 0.8047090768814087, + "num_tokens": 3285758.0, + "step": 84 + }, + { + "epoch": 0.010812873680193359, + "grad_norm": 6.606335163116455, + "learning_rate": 3.5608308605341244e-08, + "loss": 0.7326, + "mean_token_accuracy": 0.7940068244934082, + "num_tokens": 3324598.0, + "step": 85 + }, + { + "epoch": 0.01094008395878387, + "grad_norm": 5.96427059173584, + "learning_rate": 3.6032217041119116e-08, + "loss": 0.7423, + "mean_token_accuracy": 0.7878717184066772, + "num_tokens": 3368878.0, + "step": 86 + }, + { + "epoch": 0.01106729423737438, + "grad_norm": 6.681045055389404, + "learning_rate": 3.645612547689699e-08, + "loss": 0.7613, + "mean_token_accuracy": 0.7866742610931396, + "num_tokens": 3405960.0, + "step": 87 + }, + { + "epoch": 0.01119450451596489, + "grad_norm": 8.129051208496094, + "learning_rate": 3.6880033912674855e-08, + "loss": 0.7087, + "mean_token_accuracy": 0.801109254360199, + "num_tokens": 3443335.0, + "step": 88 + }, + { + "epoch": 0.0113217147945554, + "grad_norm": 8.261344909667969, + "learning_rate": 3.730394234845273e-08, + "loss": 0.7508, + "mean_token_accuracy": 0.7907965779304504, + "num_tokens": 3470878.0, + "step": 89 + }, + { + "epoch": 0.01144892507314591, + "grad_norm": 6.316943168640137, + "learning_rate": 3.77278507842306e-08, + "loss": 0.7949, + "mean_token_accuracy": 0.77655428647995, + "num_tokens": 3510081.0, + "step": 90 + }, + { + "epoch": 0.01157613535173642, + "grad_norm": 6.997718334197998, + "learning_rate": 3.815175922000847e-08, + "loss": 0.7913, + "mean_token_accuracy": 0.7764245271682739, + "num_tokens": 3551498.0, + "step": 91 + }, + { + "epoch": 0.01170334563032693, + "grad_norm": 8.201485633850098, + "learning_rate": 3.8575667655786345e-08, + "loss": 0.8431, + "mean_token_accuracy": 0.7646610736846924, + "num_tokens": 3584454.0, + "step": 92 + }, + { + "epoch": 0.01183055590891744, + "grad_norm": 6.846035480499268, + "learning_rate": 3.899957609156422e-08, + "loss": 0.7268, + "mean_token_accuracy": 0.7949182987213135, + "num_tokens": 3617436.0, + "step": 93 + }, + { + "epoch": 0.01195776618750795, + "grad_norm": 6.006182670593262, + "learning_rate": 3.94234845273421e-08, + "loss": 0.6881, + "mean_token_accuracy": 0.8074988126754761, + "num_tokens": 3659576.0, + "step": 94 + }, + { + "epoch": 0.012084976466098461, + "grad_norm": 7.8984222412109375, + "learning_rate": 3.984739296311997e-08, + "loss": 0.7574, + "mean_token_accuracy": 0.7837415337562561, + "num_tokens": 3687929.0, + "step": 95 + }, + { + "epoch": 0.01221218674468897, + "grad_norm": 6.704063892364502, + "learning_rate": 4.027130139889784e-08, + "loss": 0.7397, + "mean_token_accuracy": 0.7919744253158569, + "num_tokens": 3724677.0, + "step": 96 + }, + { + "epoch": 0.012339397023279481, + "grad_norm": 6.4213457107543945, + "learning_rate": 4.0695209834675715e-08, + "loss": 0.7316, + "mean_token_accuracy": 0.794518232345581, + "num_tokens": 3760015.0, + "step": 97 + }, + { + "epoch": 0.012466607301869992, + "grad_norm": 6.695804119110107, + "learning_rate": 4.111911827045358e-08, + "loss": 0.692, + "mean_token_accuracy": 0.8057355284690857, + "num_tokens": 3794511.0, + "step": 98 + }, + { + "epoch": 0.012593817580460501, + "grad_norm": 6.316235065460205, + "learning_rate": 4.154302670623145e-08, + "loss": 0.7828, + "mean_token_accuracy": 0.7787878513336182, + "num_tokens": 3833404.0, + "step": 99 + }, + { + "epoch": 0.012721027859051012, + "grad_norm": 5.777181148529053, + "learning_rate": 4.1966935142009326e-08, + "loss": 0.7298, + "mean_token_accuracy": 0.7976862788200378, + "num_tokens": 3871675.0, + "step": 100 + }, + { + "epoch": 0.012848238137641521, + "grad_norm": 5.682585716247559, + "learning_rate": 4.23908435777872e-08, + "loss": 0.6595, + "mean_token_accuracy": 0.8113763928413391, + "num_tokens": 3909539.0, + "step": 101 + }, + { + "epoch": 0.012975448416232032, + "grad_norm": 5.888823986053467, + "learning_rate": 4.281475201356507e-08, + "loss": 0.7651, + "mean_token_accuracy": 0.7802224159240723, + "num_tokens": 3951238.0, + "step": 102 + }, + { + "epoch": 0.013102658694822541, + "grad_norm": 5.879027366638184, + "learning_rate": 4.3238660449342943e-08, + "loss": 0.7059, + "mean_token_accuracy": 0.8008573055267334, + "num_tokens": 3990063.0, + "step": 103 + }, + { + "epoch": 0.013229868973413052, + "grad_norm": 6.258179664611816, + "learning_rate": 4.3662568885120816e-08, + "loss": 0.7744, + "mean_token_accuracy": 0.7796226739883423, + "num_tokens": 4024919.0, + "step": 104 + }, + { + "epoch": 0.013357079252003561, + "grad_norm": 5.3372297286987305, + "learning_rate": 4.408647732089869e-08, + "loss": 0.6705, + "mean_token_accuracy": 0.8072141408920288, + "num_tokens": 4064991.0, + "step": 105 + }, + { + "epoch": 0.013484289530594072, + "grad_norm": 5.391239166259766, + "learning_rate": 4.451038575667656e-08, + "loss": 0.721, + "mean_token_accuracy": 0.794502317905426, + "num_tokens": 4104374.0, + "step": 106 + }, + { + "epoch": 0.013611499809184581, + "grad_norm": 5.50892448425293, + "learning_rate": 4.493429419245443e-08, + "loss": 0.7068, + "mean_token_accuracy": 0.7947548627853394, + "num_tokens": 4146973.0, + "step": 107 + }, + { + "epoch": 0.013738710087775092, + "grad_norm": 5.386080265045166, + "learning_rate": 4.53582026282323e-08, + "loss": 0.7016, + "mean_token_accuracy": 0.7990014553070068, + "num_tokens": 4188544.0, + "step": 108 + }, + { + "epoch": 0.013865920366365603, + "grad_norm": 5.705214977264404, + "learning_rate": 4.578211106401017e-08, + "loss": 0.6949, + "mean_token_accuracy": 0.8044143319129944, + "num_tokens": 4225165.0, + "step": 109 + }, + { + "epoch": 0.013993130644956112, + "grad_norm": 5.327526092529297, + "learning_rate": 4.6206019499788045e-08, + "loss": 0.6491, + "mean_token_accuracy": 0.811558723449707, + "num_tokens": 4261503.0, + "step": 110 + }, + { + "epoch": 0.014120340923546623, + "grad_norm": 4.6242241859436035, + "learning_rate": 4.662992793556592e-08, + "loss": 0.6867, + "mean_token_accuracy": 0.8006488680839539, + "num_tokens": 4306795.0, + "step": 111 + }, + { + "epoch": 0.014247551202137132, + "grad_norm": 5.997996807098389, + "learning_rate": 4.705383637134379e-08, + "loss": 0.7316, + "mean_token_accuracy": 0.7873029708862305, + "num_tokens": 4341618.0, + "step": 112 + }, + { + "epoch": 0.014374761480727643, + "grad_norm": 6.479738235473633, + "learning_rate": 4.747774480712166e-08, + "loss": 0.709, + "mean_token_accuracy": 0.7986541986465454, + "num_tokens": 4375919.0, + "step": 113 + }, + { + "epoch": 0.014501971759318152, + "grad_norm": 6.095708847045898, + "learning_rate": 4.7901653242899535e-08, + "loss": 0.6615, + "mean_token_accuracy": 0.8119421005249023, + "num_tokens": 4410900.0, + "step": 114 + }, + { + "epoch": 0.014629182037908663, + "grad_norm": 5.308213233947754, + "learning_rate": 4.832556167867741e-08, + "loss": 0.637, + "mean_token_accuracy": 0.8143362998962402, + "num_tokens": 4448858.0, + "step": 115 + }, + { + "epoch": 0.014756392316499172, + "grad_norm": 4.874562740325928, + "learning_rate": 4.8749470114455274e-08, + "loss": 0.6552, + "mean_token_accuracy": 0.8090206980705261, + "num_tokens": 4489308.0, + "step": 116 + }, + { + "epoch": 0.014883602595089683, + "grad_norm": 4.618898391723633, + "learning_rate": 4.9173378550233146e-08, + "loss": 0.6708, + "mean_token_accuracy": 0.8067784309387207, + "num_tokens": 4533708.0, + "step": 117 + }, + { + "epoch": 0.015010812873680193, + "grad_norm": 5.68588924407959, + "learning_rate": 4.959728698601102e-08, + "loss": 0.6688, + "mean_token_accuracy": 0.8076982498168945, + "num_tokens": 4567233.0, + "step": 118 + }, + { + "epoch": 0.015138023152270703, + "grad_norm": 5.007695198059082, + "learning_rate": 5.002119542178889e-08, + "loss": 0.6302, + "mean_token_accuracy": 0.8159814476966858, + "num_tokens": 4607474.0, + "step": 119 + }, + { + "epoch": 0.015265233430861214, + "grad_norm": 4.634917259216309, + "learning_rate": 5.0445103857566764e-08, + "loss": 0.5834, + "mean_token_accuracy": 0.8324447870254517, + "num_tokens": 4644518.0, + "step": 120 + }, + { + "epoch": 0.015392443709451724, + "grad_norm": 5.373696327209473, + "learning_rate": 5.0869012293344637e-08, + "loss": 0.6686, + "mean_token_accuracy": 0.8058242201805115, + "num_tokens": 4682020.0, + "step": 121 + }, + { + "epoch": 0.015519653988042234, + "grad_norm": 5.399783611297607, + "learning_rate": 5.129292072912251e-08, + "loss": 0.6385, + "mean_token_accuracy": 0.8146382570266724, + "num_tokens": 4722243.0, + "step": 122 + }, + { + "epoch": 0.015646864266632744, + "grad_norm": 5.774118900299072, + "learning_rate": 5.171682916490038e-08, + "loss": 0.6463, + "mean_token_accuracy": 0.8152012825012207, + "num_tokens": 4759537.0, + "step": 123 + }, + { + "epoch": 0.015774074545223254, + "grad_norm": 4.963928699493408, + "learning_rate": 5.2140737600678254e-08, + "loss": 0.6588, + "mean_token_accuracy": 0.8081181049346924, + "num_tokens": 4794279.0, + "step": 124 + }, + { + "epoch": 0.015901284823813765, + "grad_norm": 5.378013610839844, + "learning_rate": 5.256464603645612e-08, + "loss": 0.7235, + "mean_token_accuracy": 0.7918686866760254, + "num_tokens": 4829017.0, + "step": 125 + }, + { + "epoch": 0.016028495102404273, + "grad_norm": 4.689215183258057, + "learning_rate": 5.298855447223399e-08, + "loss": 0.5932, + "mean_token_accuracy": 0.8213663101196289, + "num_tokens": 4868697.0, + "step": 126 + }, + { + "epoch": 0.016155705380994784, + "grad_norm": 4.911618232727051, + "learning_rate": 5.3412462908011865e-08, + "loss": 0.6142, + "mean_token_accuracy": 0.8221992254257202, + "num_tokens": 4903357.0, + "step": 127 + }, + { + "epoch": 0.016282915659585295, + "grad_norm": 5.3893842697143555, + "learning_rate": 5.383637134378974e-08, + "loss": 0.709, + "mean_token_accuracy": 0.7919142246246338, + "num_tokens": 4940529.0, + "step": 128 + }, + { + "epoch": 0.016410125938175806, + "grad_norm": 5.173316478729248, + "learning_rate": 5.426027977956761e-08, + "loss": 0.746, + "mean_token_accuracy": 0.7877398729324341, + "num_tokens": 4976307.0, + "step": 129 + }, + { + "epoch": 0.016537336216766316, + "grad_norm": 5.234180450439453, + "learning_rate": 5.468418821534548e-08, + "loss": 0.6166, + "mean_token_accuracy": 0.822358250617981, + "num_tokens": 5012438.0, + "step": 130 + }, + { + "epoch": 0.016664546495356824, + "grad_norm": 4.960449695587158, + "learning_rate": 5.5108096651123356e-08, + "loss": 0.6439, + "mean_token_accuracy": 0.8138243556022644, + "num_tokens": 5049380.0, + "step": 131 + }, + { + "epoch": 0.016791756773947335, + "grad_norm": 4.3664116859436035, + "learning_rate": 5.553200508690123e-08, + "loss": 0.645, + "mean_token_accuracy": 0.8117821216583252, + "num_tokens": 5092559.0, + "step": 132 + }, + { + "epoch": 0.016918967052537846, + "grad_norm": 4.374181747436523, + "learning_rate": 5.59559135226791e-08, + "loss": 0.6753, + "mean_token_accuracy": 0.8042807579040527, + "num_tokens": 5133988.0, + "step": 133 + }, + { + "epoch": 0.017046177331128357, + "grad_norm": 4.041244029998779, + "learning_rate": 5.637982195845697e-08, + "loss": 0.5999, + "mean_token_accuracy": 0.8234436511993408, + "num_tokens": 5178937.0, + "step": 134 + }, + { + "epoch": 0.017173387609718864, + "grad_norm": 4.421737194061279, + "learning_rate": 5.680373039423484e-08, + "loss": 0.6796, + "mean_token_accuracy": 0.7997405529022217, + "num_tokens": 5220810.0, + "step": 135 + }, + { + "epoch": 0.017300597888309375, + "grad_norm": 4.191774368286133, + "learning_rate": 5.722763883001271e-08, + "loss": 0.6162, + "mean_token_accuracy": 0.8158925771713257, + "num_tokens": 5259494.0, + "step": 136 + }, + { + "epoch": 0.017427808166899886, + "grad_norm": 5.012508392333984, + "learning_rate": 5.7651547265790585e-08, + "loss": 0.6832, + "mean_token_accuracy": 0.8076955080032349, + "num_tokens": 5298593.0, + "step": 137 + }, + { + "epoch": 0.017555018445490397, + "grad_norm": 4.712111949920654, + "learning_rate": 5.807545570156846e-08, + "loss": 0.7015, + "mean_token_accuracy": 0.7970151901245117, + "num_tokens": 5343536.0, + "step": 138 + }, + { + "epoch": 0.017682228724080904, + "grad_norm": 3.7535243034362793, + "learning_rate": 5.849936413734633e-08, + "loss": 0.6164, + "mean_token_accuracy": 0.8185596466064453, + "num_tokens": 5387758.0, + "step": 139 + }, + { + "epoch": 0.017809439002671415, + "grad_norm": 3.8617444038391113, + "learning_rate": 5.89232725731242e-08, + "loss": 0.6704, + "mean_token_accuracy": 0.8031830787658691, + "num_tokens": 5431120.0, + "step": 140 + }, + { + "epoch": 0.017936649281261926, + "grad_norm": 4.130899429321289, + "learning_rate": 5.9347181008902075e-08, + "loss": 0.6344, + "mean_token_accuracy": 0.8101925849914551, + "num_tokens": 5463432.0, + "step": 141 + }, + { + "epoch": 0.018063859559852437, + "grad_norm": 3.775186538696289, + "learning_rate": 5.977108944467995e-08, + "loss": 0.6505, + "mean_token_accuracy": 0.8079653382301331, + "num_tokens": 5499199.0, + "step": 142 + }, + { + "epoch": 0.018191069838442948, + "grad_norm": 4.056355953216553, + "learning_rate": 6.019499788045781e-08, + "loss": 0.6172, + "mean_token_accuracy": 0.8151942491531372, + "num_tokens": 5532617.0, + "step": 143 + }, + { + "epoch": 0.018318280117033455, + "grad_norm": 4.348077774047852, + "learning_rate": 6.061890631623569e-08, + "loss": 0.6219, + "mean_token_accuracy": 0.8179042339324951, + "num_tokens": 5573577.0, + "step": 144 + }, + { + "epoch": 0.018445490395623966, + "grad_norm": 3.7537612915039062, + "learning_rate": 6.104281475201356e-08, + "loss": 0.6553, + "mean_token_accuracy": 0.8080774545669556, + "num_tokens": 5615107.0, + "step": 145 + }, + { + "epoch": 0.018572700674214477, + "grad_norm": 3.4495368003845215, + "learning_rate": 6.146672318779143e-08, + "loss": 0.599, + "mean_token_accuracy": 0.8234204053878784, + "num_tokens": 5656172.0, + "step": 146 + }, + { + "epoch": 0.018699910952804988, + "grad_norm": 3.6327810287475586, + "learning_rate": 6.18906316235693e-08, + "loss": 0.6204, + "mean_token_accuracy": 0.8171635866165161, + "num_tokens": 5695498.0, + "step": 147 + }, + { + "epoch": 0.018827121231395495, + "grad_norm": 3.5510196685791016, + "learning_rate": 6.231454005934718e-08, + "loss": 0.608, + "mean_token_accuracy": 0.8199845552444458, + "num_tokens": 5735533.0, + "step": 148 + }, + { + "epoch": 0.018954331509986006, + "grad_norm": 3.4512293338775635, + "learning_rate": 6.273844849512505e-08, + "loss": 0.635, + "mean_token_accuracy": 0.8147587776184082, + "num_tokens": 5778232.0, + "step": 149 + }, + { + "epoch": 0.019081541788576517, + "grad_norm": 3.6618924140930176, + "learning_rate": 6.316235693090292e-08, + "loss": 0.647, + "mean_token_accuracy": 0.8046040534973145, + "num_tokens": 5818498.0, + "step": 150 + }, + { + "epoch": 0.019208752067167028, + "grad_norm": 3.810884475708008, + "learning_rate": 6.35862653666808e-08, + "loss": 0.6326, + "mean_token_accuracy": 0.8113516569137573, + "num_tokens": 5854629.0, + "step": 151 + }, + { + "epoch": 0.01933596234575754, + "grad_norm": 3.583634376525879, + "learning_rate": 6.401017380245867e-08, + "loss": 0.6129, + "mean_token_accuracy": 0.8158899545669556, + "num_tokens": 5893210.0, + "step": 152 + }, + { + "epoch": 0.019463172624348046, + "grad_norm": 3.9543447494506836, + "learning_rate": 6.443408223823654e-08, + "loss": 0.6027, + "mean_token_accuracy": 0.8189770579338074, + "num_tokens": 5927023.0, + "step": 153 + }, + { + "epoch": 0.019590382902938557, + "grad_norm": 3.4902234077453613, + "learning_rate": 6.485799067401441e-08, + "loss": 0.6644, + "mean_token_accuracy": 0.8030644655227661, + "num_tokens": 5969856.0, + "step": 154 + }, + { + "epoch": 0.019717593181529068, + "grad_norm": 3.54006290435791, + "learning_rate": 6.528189910979228e-08, + "loss": 0.6341, + "mean_token_accuracy": 0.8143118619918823, + "num_tokens": 6009093.0, + "step": 155 + }, + { + "epoch": 0.01984480346011958, + "grad_norm": 3.2057533264160156, + "learning_rate": 6.570580754557016e-08, + "loss": 0.5603, + "mean_token_accuracy": 0.8308569192886353, + "num_tokens": 6052335.0, + "step": 156 + }, + { + "epoch": 0.019972013738710086, + "grad_norm": 3.4221343994140625, + "learning_rate": 6.612971598134802e-08, + "loss": 0.6664, + "mean_token_accuracy": 0.7994051575660706, + "num_tokens": 6095044.0, + "step": 157 + }, + { + "epoch": 0.020099224017300597, + "grad_norm": 3.462714195251465, + "learning_rate": 6.655362441712589e-08, + "loss": 0.5917, + "mean_token_accuracy": 0.8255913257598877, + "num_tokens": 6134256.0, + "step": 158 + }, + { + "epoch": 0.020226434295891108, + "grad_norm": 3.7882113456726074, + "learning_rate": 6.697753285290376e-08, + "loss": 0.7031, + "mean_token_accuracy": 0.7922717332839966, + "num_tokens": 6170272.0, + "step": 159 + }, + { + "epoch": 0.02035364457448162, + "grad_norm": 3.2305262088775635, + "learning_rate": 6.740144128868163e-08, + "loss": 0.5899, + "mean_token_accuracy": 0.8186209201812744, + "num_tokens": 6215550.0, + "step": 160 + }, + { + "epoch": 0.020480854853072127, + "grad_norm": 3.2921950817108154, + "learning_rate": 6.78253497244595e-08, + "loss": 0.5704, + "mean_token_accuracy": 0.8267449736595154, + "num_tokens": 6254137.0, + "step": 161 + }, + { + "epoch": 0.020608065131662637, + "grad_norm": 3.212261438369751, + "learning_rate": 6.824925816023738e-08, + "loss": 0.5391, + "mean_token_accuracy": 0.835530161857605, + "num_tokens": 6290035.0, + "step": 162 + }, + { + "epoch": 0.02073527541025315, + "grad_norm": 3.1222097873687744, + "learning_rate": 6.867316659601525e-08, + "loss": 0.6115, + "mean_token_accuracy": 0.8223940134048462, + "num_tokens": 6330507.0, + "step": 163 + }, + { + "epoch": 0.02086248568884366, + "grad_norm": 3.2338979244232178, + "learning_rate": 6.909707503179312e-08, + "loss": 0.6064, + "mean_token_accuracy": 0.8166545033454895, + "num_tokens": 6370701.0, + "step": 164 + }, + { + "epoch": 0.02098969596743417, + "grad_norm": 3.2645821571350098, + "learning_rate": 6.9520983467571e-08, + "loss": 0.6154, + "mean_token_accuracy": 0.8136652708053589, + "num_tokens": 6410237.0, + "step": 165 + }, + { + "epoch": 0.021116906246024678, + "grad_norm": 3.012397527694702, + "learning_rate": 6.994489190334887e-08, + "loss": 0.6072, + "mean_token_accuracy": 0.8145463466644287, + "num_tokens": 6455981.0, + "step": 166 + }, + { + "epoch": 0.02124411652461519, + "grad_norm": 3.3187620639801025, + "learning_rate": 7.036880033912674e-08, + "loss": 0.5899, + "mean_token_accuracy": 0.8187699317932129, + "num_tokens": 6491059.0, + "step": 167 + }, + { + "epoch": 0.0213713268032057, + "grad_norm": 3.3305752277374268, + "learning_rate": 7.079270877490461e-08, + "loss": 0.6279, + "mean_token_accuracy": 0.8109997510910034, + "num_tokens": 6533001.0, + "step": 168 + }, + { + "epoch": 0.02149853708179621, + "grad_norm": 3.472684144973755, + "learning_rate": 7.121661721068249e-08, + "loss": 0.6465, + "mean_token_accuracy": 0.8053048253059387, + "num_tokens": 6570527.0, + "step": 169 + }, + { + "epoch": 0.021625747360386718, + "grad_norm": 3.1682357788085938, + "learning_rate": 7.164052564646036e-08, + "loss": 0.5996, + "mean_token_accuracy": 0.8204360604286194, + "num_tokens": 6611847.0, + "step": 170 + }, + { + "epoch": 0.02175295763897723, + "grad_norm": 3.6940042972564697, + "learning_rate": 7.206443408223823e-08, + "loss": 0.6485, + "mean_token_accuracy": 0.8074818253517151, + "num_tokens": 6644581.0, + "step": 171 + }, + { + "epoch": 0.02188016791756774, + "grad_norm": 3.0305182933807373, + "learning_rate": 7.24883425180161e-08, + "loss": 0.564, + "mean_token_accuracy": 0.8308360576629639, + "num_tokens": 6685713.0, + "step": 172 + }, + { + "epoch": 0.02200737819615825, + "grad_norm": 3.2632603645324707, + "learning_rate": 7.291225095379398e-08, + "loss": 0.65, + "mean_token_accuracy": 0.800949215888977, + "num_tokens": 6725129.0, + "step": 173 + }, + { + "epoch": 0.02213458847474876, + "grad_norm": 3.293576240539551, + "learning_rate": 7.333615938957185e-08, + "loss": 0.5971, + "mean_token_accuracy": 0.8189734220504761, + "num_tokens": 6761101.0, + "step": 174 + }, + { + "epoch": 0.02226179875333927, + "grad_norm": 3.36242413520813, + "learning_rate": 7.376006782534971e-08, + "loss": 0.573, + "mean_token_accuracy": 0.826209306716919, + "num_tokens": 6794045.0, + "step": 175 + }, + { + "epoch": 0.02238900903192978, + "grad_norm": 3.0980119705200195, + "learning_rate": 7.418397626112758e-08, + "loss": 0.5808, + "mean_token_accuracy": 0.8226041197776794, + "num_tokens": 6833458.0, + "step": 176 + }, + { + "epoch": 0.02251621931052029, + "grad_norm": 3.4342682361602783, + "learning_rate": 7.460788469690545e-08, + "loss": 0.6131, + "mean_token_accuracy": 0.8134138584136963, + "num_tokens": 6873898.0, + "step": 177 + }, + { + "epoch": 0.0226434295891108, + "grad_norm": 3.2408671379089355, + "learning_rate": 7.503179313268333e-08, + "loss": 0.601, + "mean_token_accuracy": 0.8155239820480347, + "num_tokens": 6912155.0, + "step": 178 + }, + { + "epoch": 0.02277063986770131, + "grad_norm": 3.205305337905884, + "learning_rate": 7.54557015684612e-08, + "loss": 0.5829, + "mean_token_accuracy": 0.8220143914222717, + "num_tokens": 6947324.0, + "step": 179 + }, + { + "epoch": 0.02289785014629182, + "grad_norm": 3.215634346008301, + "learning_rate": 7.587961000423907e-08, + "loss": 0.6152, + "mean_token_accuracy": 0.8138933181762695, + "num_tokens": 6986163.0, + "step": 180 + }, + { + "epoch": 0.02302506042488233, + "grad_norm": 3.003775119781494, + "learning_rate": 7.630351844001694e-08, + "loss": 0.5592, + "mean_token_accuracy": 0.8301811218261719, + "num_tokens": 7027268.0, + "step": 181 + }, + { + "epoch": 0.02315227070347284, + "grad_norm": 3.2323217391967773, + "learning_rate": 7.672742687579482e-08, + "loss": 0.6053, + "mean_token_accuracy": 0.8155161142349243, + "num_tokens": 7065709.0, + "step": 182 + }, + { + "epoch": 0.02327948098206335, + "grad_norm": 2.8990225791931152, + "learning_rate": 7.715133531157269e-08, + "loss": 0.5875, + "mean_token_accuracy": 0.8211745619773865, + "num_tokens": 7108124.0, + "step": 183 + }, + { + "epoch": 0.02340669126065386, + "grad_norm": 2.9445018768310547, + "learning_rate": 7.757524374735056e-08, + "loss": 0.5761, + "mean_token_accuracy": 0.8259589672088623, + "num_tokens": 7155007.0, + "step": 184 + }, + { + "epoch": 0.02353390153924437, + "grad_norm": 2.950753927230835, + "learning_rate": 7.799915218312844e-08, + "loss": 0.5962, + "mean_token_accuracy": 0.8186321258544922, + "num_tokens": 7197010.0, + "step": 185 + }, + { + "epoch": 0.02366111181783488, + "grad_norm": 2.797281503677368, + "learning_rate": 7.842306061890631e-08, + "loss": 0.5936, + "mean_token_accuracy": 0.8191157579421997, + "num_tokens": 7244586.0, + "step": 186 + }, + { + "epoch": 0.023788322096425393, + "grad_norm": 2.973834276199341, + "learning_rate": 7.88469690546842e-08, + "loss": 0.5968, + "mean_token_accuracy": 0.8208262920379639, + "num_tokens": 7283721.0, + "step": 187 + }, + { + "epoch": 0.0239155323750159, + "grad_norm": 3.0963833332061768, + "learning_rate": 7.927087749046207e-08, + "loss": 0.5978, + "mean_token_accuracy": 0.8144268989562988, + "num_tokens": 7320132.0, + "step": 188 + }, + { + "epoch": 0.02404274265360641, + "grad_norm": 2.9068198204040527, + "learning_rate": 7.969478592623994e-08, + "loss": 0.5773, + "mean_token_accuracy": 0.8213964104652405, + "num_tokens": 7366009.0, + "step": 189 + }, + { + "epoch": 0.024169952932196922, + "grad_norm": 2.7777082920074463, + "learning_rate": 8.011869436201781e-08, + "loss": 0.5735, + "mean_token_accuracy": 0.826187014579773, + "num_tokens": 7407514.0, + "step": 190 + }, + { + "epoch": 0.024297163210787433, + "grad_norm": 3.044638156890869, + "learning_rate": 8.054260279779568e-08, + "loss": 0.5956, + "mean_token_accuracy": 0.818964958190918, + "num_tokens": 7445638.0, + "step": 191 + }, + { + "epoch": 0.02442437348937794, + "grad_norm": 3.252720832824707, + "learning_rate": 8.096651123357356e-08, + "loss": 0.5446, + "mean_token_accuracy": 0.8333422541618347, + "num_tokens": 7477495.0, + "step": 192 + }, + { + "epoch": 0.02455158376796845, + "grad_norm": 3.0787339210510254, + "learning_rate": 8.139041966935143e-08, + "loss": 0.6068, + "mean_token_accuracy": 0.8144227862358093, + "num_tokens": 7514861.0, + "step": 193 + }, + { + "epoch": 0.024678794046558962, + "grad_norm": 2.7172224521636963, + "learning_rate": 8.181432810512929e-08, + "loss": 0.5404, + "mean_token_accuracy": 0.8326760530471802, + "num_tokens": 7556949.0, + "step": 194 + }, + { + "epoch": 0.024806004325149473, + "grad_norm": 2.9573259353637695, + "learning_rate": 8.223823654090716e-08, + "loss": 0.5569, + "mean_token_accuracy": 0.8286169767379761, + "num_tokens": 7597129.0, + "step": 195 + }, + { + "epoch": 0.024933214603739984, + "grad_norm": 2.749664068222046, + "learning_rate": 8.266214497668503e-08, + "loss": 0.5891, + "mean_token_accuracy": 0.8204962611198425, + "num_tokens": 7643102.0, + "step": 196 + }, + { + "epoch": 0.02506042488233049, + "grad_norm": 3.275874614715576, + "learning_rate": 8.30860534124629e-08, + "loss": 0.5536, + "mean_token_accuracy": 0.828984260559082, + "num_tokens": 7675847.0, + "step": 197 + }, + { + "epoch": 0.025187635160921002, + "grad_norm": 2.8897199630737305, + "learning_rate": 8.350996184824078e-08, + "loss": 0.5875, + "mean_token_accuracy": 0.8224993944168091, + "num_tokens": 7715196.0, + "step": 198 + }, + { + "epoch": 0.025314845439511513, + "grad_norm": 2.8773486614227295, + "learning_rate": 8.393387028401865e-08, + "loss": 0.5706, + "mean_token_accuracy": 0.8258112668991089, + "num_tokens": 7754587.0, + "step": 199 + }, + { + "epoch": 0.025442055718102024, + "grad_norm": 2.744950771331787, + "learning_rate": 8.435777871979652e-08, + "loss": 0.5011, + "mean_token_accuracy": 0.844011664390564, + "num_tokens": 7790707.0, + "step": 200 + }, + { + "epoch": 0.02556926599669253, + "grad_norm": 2.816615581512451, + "learning_rate": 8.47816871555744e-08, + "loss": 0.5709, + "mean_token_accuracy": 0.8240379691123962, + "num_tokens": 7832245.0, + "step": 201 + }, + { + "epoch": 0.025696476275283042, + "grad_norm": 3.2710864543914795, + "learning_rate": 8.520559559135227e-08, + "loss": 0.6514, + "mean_token_accuracy": 0.8069828748703003, + "num_tokens": 7867415.0, + "step": 202 + }, + { + "epoch": 0.025823686553873553, + "grad_norm": 2.677642822265625, + "learning_rate": 8.562950402713014e-08, + "loss": 0.5595, + "mean_token_accuracy": 0.823906660079956, + "num_tokens": 7910924.0, + "step": 203 + }, + { + "epoch": 0.025950896832464064, + "grad_norm": 3.0577399730682373, + "learning_rate": 8.605341246290801e-08, + "loss": 0.538, + "mean_token_accuracy": 0.831372857093811, + "num_tokens": 7943466.0, + "step": 204 + }, + { + "epoch": 0.026078107111054575, + "grad_norm": 2.921208620071411, + "learning_rate": 8.647732089868589e-08, + "loss": 0.5054, + "mean_token_accuracy": 0.8407784104347229, + "num_tokens": 7976653.0, + "step": 205 + }, + { + "epoch": 0.026205317389645082, + "grad_norm": 2.893279552459717, + "learning_rate": 8.690122933446376e-08, + "loss": 0.5961, + "mean_token_accuracy": 0.815682053565979, + "num_tokens": 8014236.0, + "step": 206 + }, + { + "epoch": 0.026332527668235593, + "grad_norm": 2.977954864501953, + "learning_rate": 8.732513777024163e-08, + "loss": 0.5538, + "mean_token_accuracy": 0.8290812969207764, + "num_tokens": 8051969.0, + "step": 207 + }, + { + "epoch": 0.026459737946826104, + "grad_norm": 2.597805976867676, + "learning_rate": 8.77490462060195e-08, + "loss": 0.4776, + "mean_token_accuracy": 0.8507893085479736, + "num_tokens": 8091479.0, + "step": 208 + }, + { + "epoch": 0.026586948225416615, + "grad_norm": 2.9047293663024902, + "learning_rate": 8.817295464179738e-08, + "loss": 0.5736, + "mean_token_accuracy": 0.8223727941513062, + "num_tokens": 8128287.0, + "step": 209 + }, + { + "epoch": 0.026714158504007122, + "grad_norm": 2.7059309482574463, + "learning_rate": 8.859686307757525e-08, + "loss": 0.546, + "mean_token_accuracy": 0.8341982364654541, + "num_tokens": 8166716.0, + "step": 210 + }, + { + "epoch": 0.026841368782597633, + "grad_norm": 2.9333786964416504, + "learning_rate": 8.902077151335312e-08, + "loss": 0.576, + "mean_token_accuracy": 0.8231337666511536, + "num_tokens": 8204535.0, + "step": 211 + }, + { + "epoch": 0.026968579061188144, + "grad_norm": 2.911827325820923, + "learning_rate": 8.944467994913098e-08, + "loss": 0.524, + "mean_token_accuracy": 0.8389162421226501, + "num_tokens": 8238777.0, + "step": 212 + }, + { + "epoch": 0.027095789339778655, + "grad_norm": 2.714789628982544, + "learning_rate": 8.986858838490885e-08, + "loss": 0.5432, + "mean_token_accuracy": 0.8335452079772949, + "num_tokens": 8280762.0, + "step": 213 + }, + { + "epoch": 0.027222999618369163, + "grad_norm": 2.8334712982177734, + "learning_rate": 9.029249682068673e-08, + "loss": 0.5498, + "mean_token_accuracy": 0.8278334736824036, + "num_tokens": 8318719.0, + "step": 214 + }, + { + "epoch": 0.027350209896959674, + "grad_norm": 2.600674867630005, + "learning_rate": 9.07164052564646e-08, + "loss": 0.5211, + "mean_token_accuracy": 0.8358433246612549, + "num_tokens": 8360034.0, + "step": 215 + }, + { + "epoch": 0.027477420175550184, + "grad_norm": 2.914384603500366, + "learning_rate": 9.114031369224247e-08, + "loss": 0.5756, + "mean_token_accuracy": 0.8246141672134399, + "num_tokens": 8395509.0, + "step": 216 + }, + { + "epoch": 0.027604630454140695, + "grad_norm": 2.7520804405212402, + "learning_rate": 9.156422212802034e-08, + "loss": 0.5514, + "mean_token_accuracy": 0.8303922414779663, + "num_tokens": 8434463.0, + "step": 217 + }, + { + "epoch": 0.027731840732731206, + "grad_norm": 2.6964664459228516, + "learning_rate": 9.198813056379822e-08, + "loss": 0.5516, + "mean_token_accuracy": 0.8283290863037109, + "num_tokens": 8475875.0, + "step": 218 + }, + { + "epoch": 0.027859051011321714, + "grad_norm": 2.6858632564544678, + "learning_rate": 9.241203899957609e-08, + "loss": 0.561, + "mean_token_accuracy": 0.826911211013794, + "num_tokens": 8518328.0, + "step": 219 + }, + { + "epoch": 0.027986261289912225, + "grad_norm": 2.7084357738494873, + "learning_rate": 9.283594743535396e-08, + "loss": 0.5369, + "mean_token_accuracy": 0.8295474052429199, + "num_tokens": 8560669.0, + "step": 220 + }, + { + "epoch": 0.028113471568502735, + "grad_norm": 2.7272887229919434, + "learning_rate": 9.325985587113183e-08, + "loss": 0.5722, + "mean_token_accuracy": 0.8215881586074829, + "num_tokens": 8603357.0, + "step": 221 + }, + { + "epoch": 0.028240681847093246, + "grad_norm": 2.670517683029175, + "learning_rate": 9.368376430690971e-08, + "loss": 0.5006, + "mean_token_accuracy": 0.8404536247253418, + "num_tokens": 8642109.0, + "step": 222 + }, + { + "epoch": 0.028367892125683754, + "grad_norm": 2.702662706375122, + "learning_rate": 9.410767274268758e-08, + "loss": 0.5009, + "mean_token_accuracy": 0.8424524068832397, + "num_tokens": 8682180.0, + "step": 223 + }, + { + "epoch": 0.028495102404274265, + "grad_norm": 2.854450225830078, + "learning_rate": 9.453158117846545e-08, + "loss": 0.514, + "mean_token_accuracy": 0.8396710157394409, + "num_tokens": 8717631.0, + "step": 224 + }, + { + "epoch": 0.028622312682864776, + "grad_norm": 2.763122320175171, + "learning_rate": 9.495548961424333e-08, + "loss": 0.5232, + "mean_token_accuracy": 0.8353489637374878, + "num_tokens": 8757388.0, + "step": 225 + }, + { + "epoch": 0.028749522961455286, + "grad_norm": 2.993499517440796, + "learning_rate": 9.53793980500212e-08, + "loss": 0.5395, + "mean_token_accuracy": 0.8335427045822144, + "num_tokens": 8793250.0, + "step": 226 + }, + { + "epoch": 0.028876733240045797, + "grad_norm": 2.8263401985168457, + "learning_rate": 9.580330648579907e-08, + "loss": 0.5752, + "mean_token_accuracy": 0.8210756182670593, + "num_tokens": 8833888.0, + "step": 227 + }, + { + "epoch": 0.029003943518636305, + "grad_norm": 2.678356409072876, + "learning_rate": 9.622721492157694e-08, + "loss": 0.5284, + "mean_token_accuracy": 0.8351384401321411, + "num_tokens": 8874136.0, + "step": 228 + }, + { + "epoch": 0.029131153797226816, + "grad_norm": 2.58211350440979, + "learning_rate": 9.665112335735482e-08, + "loss": 0.5364, + "mean_token_accuracy": 0.8311750888824463, + "num_tokens": 8917266.0, + "step": 229 + }, + { + "epoch": 0.029258364075817327, + "grad_norm": 2.7190933227539062, + "learning_rate": 9.707503179313267e-08, + "loss": 0.4928, + "mean_token_accuracy": 0.8428329825401306, + "num_tokens": 8952584.0, + "step": 230 + }, + { + "epoch": 0.029385574354407838, + "grad_norm": 2.6344869136810303, + "learning_rate": 9.749894022891055e-08, + "loss": 0.5395, + "mean_token_accuracy": 0.8313517570495605, + "num_tokens": 8995144.0, + "step": 231 + }, + { + "epoch": 0.029512784632998345, + "grad_norm": 2.8428642749786377, + "learning_rate": 9.792284866468842e-08, + "loss": 0.5434, + "mean_token_accuracy": 0.8330553770065308, + "num_tokens": 9033098.0, + "step": 232 + }, + { + "epoch": 0.029639994911588856, + "grad_norm": 2.8560545444488525, + "learning_rate": 9.834675710046629e-08, + "loss": 0.5605, + "mean_token_accuracy": 0.8251652717590332, + "num_tokens": 9070409.0, + "step": 233 + }, + { + "epoch": 0.029767205190179367, + "grad_norm": 2.5736701488494873, + "learning_rate": 9.877066553624416e-08, + "loss": 0.5343, + "mean_token_accuracy": 0.8343055248260498, + "num_tokens": 9112725.0, + "step": 234 + }, + { + "epoch": 0.029894415468769878, + "grad_norm": 2.8470346927642822, + "learning_rate": 9.919457397202204e-08, + "loss": 0.5439, + "mean_token_accuracy": 0.8314188718795776, + "num_tokens": 9148191.0, + "step": 235 + }, + { + "epoch": 0.030021625747360385, + "grad_norm": 2.645294427871704, + "learning_rate": 9.961848240779991e-08, + "loss": 0.5154, + "mean_token_accuracy": 0.8380269408226013, + "num_tokens": 9189613.0, + "step": 236 + }, + { + "epoch": 0.030148836025950896, + "grad_norm": 2.776836395263672, + "learning_rate": 1.0004239084357778e-07, + "loss": 0.5131, + "mean_token_accuracy": 0.8362679481506348, + "num_tokens": 9225198.0, + "step": 237 + }, + { + "epoch": 0.030276046304541407, + "grad_norm": 3.0032336711883545, + "learning_rate": 1.0046629927935566e-07, + "loss": 0.5075, + "mean_token_accuracy": 0.8377412557601929, + "num_tokens": 9256780.0, + "step": 238 + }, + { + "epoch": 0.030403256583131918, + "grad_norm": 2.961663246154785, + "learning_rate": 1.0089020771513353e-07, + "loss": 0.5829, + "mean_token_accuracy": 0.8172580003738403, + "num_tokens": 9295993.0, + "step": 239 + }, + { + "epoch": 0.03053046686172243, + "grad_norm": 2.7104859352111816, + "learning_rate": 1.013141161509114e-07, + "loss": 0.5413, + "mean_token_accuracy": 0.8314675092697144, + "num_tokens": 9337985.0, + "step": 240 + }, + { + "epoch": 0.030657677140312936, + "grad_norm": 2.76869797706604, + "learning_rate": 1.0173802458668927e-07, + "loss": 0.5875, + "mean_token_accuracy": 0.8171294331550598, + "num_tokens": 9378334.0, + "step": 241 + }, + { + "epoch": 0.030784887418903447, + "grad_norm": 2.514705181121826, + "learning_rate": 1.0216193302246715e-07, + "loss": 0.4708, + "mean_token_accuracy": 0.851851224899292, + "num_tokens": 9419520.0, + "step": 242 + }, + { + "epoch": 0.030912097697493958, + "grad_norm": 2.4706883430480957, + "learning_rate": 1.0258584145824502e-07, + "loss": 0.467, + "mean_token_accuracy": 0.8483937978744507, + "num_tokens": 9461664.0, + "step": 243 + }, + { + "epoch": 0.03103930797608447, + "grad_norm": 2.6395270824432373, + "learning_rate": 1.0300974989402289e-07, + "loss": 0.5135, + "mean_token_accuracy": 0.8398679494857788, + "num_tokens": 9501851.0, + "step": 244 + }, + { + "epoch": 0.031166518254674976, + "grad_norm": 2.763108968734741, + "learning_rate": 1.0343365832980076e-07, + "loss": 0.5268, + "mean_token_accuracy": 0.832537055015564, + "num_tokens": 9540821.0, + "step": 245 + }, + { + "epoch": 0.03129372853326549, + "grad_norm": 2.5326266288757324, + "learning_rate": 1.0385756676557864e-07, + "loss": 0.4834, + "mean_token_accuracy": 0.8458501696586609, + "num_tokens": 9581786.0, + "step": 246 + }, + { + "epoch": 0.031420938811855995, + "grad_norm": 2.5277812480926514, + "learning_rate": 1.0428147520135651e-07, + "loss": 0.4765, + "mean_token_accuracy": 0.8505228161811829, + "num_tokens": 9621988.0, + "step": 247 + }, + { + "epoch": 0.03154814909044651, + "grad_norm": 2.583372116088867, + "learning_rate": 1.0470538363713437e-07, + "loss": 0.5065, + "mean_token_accuracy": 0.838394284248352, + "num_tokens": 9664054.0, + "step": 248 + }, + { + "epoch": 0.031675359369037016, + "grad_norm": 2.7590513229370117, + "learning_rate": 1.0512929207291224e-07, + "loss": 0.5087, + "mean_token_accuracy": 0.8383837938308716, + "num_tokens": 9701994.0, + "step": 249 + }, + { + "epoch": 0.03180256964762753, + "grad_norm": 2.903249979019165, + "learning_rate": 1.0555320050869011e-07, + "loss": 0.5672, + "mean_token_accuracy": 0.8176788687705994, + "num_tokens": 9738880.0, + "step": 250 + }, + { + "epoch": 0.03192977992621804, + "grad_norm": 2.874760627746582, + "learning_rate": 1.0597710894446799e-07, + "loss": 0.4699, + "mean_token_accuracy": 0.848349928855896, + "num_tokens": 9773749.0, + "step": 251 + }, + { + "epoch": 0.032056990204808546, + "grad_norm": 2.9139389991760254, + "learning_rate": 1.0640101738024586e-07, + "loss": 0.5296, + "mean_token_accuracy": 0.8322674036026001, + "num_tokens": 9808526.0, + "step": 252 + }, + { + "epoch": 0.03218420048339906, + "grad_norm": 2.634732961654663, + "learning_rate": 1.0682492581602373e-07, + "loss": 0.4661, + "mean_token_accuracy": 0.8525162935256958, + "num_tokens": 9847078.0, + "step": 253 + }, + { + "epoch": 0.03231141076198957, + "grad_norm": 2.729233980178833, + "learning_rate": 1.072488342518016e-07, + "loss": 0.5236, + "mean_token_accuracy": 0.8306689858436584, + "num_tokens": 9884587.0, + "step": 254 + }, + { + "epoch": 0.03243862104058008, + "grad_norm": 2.659259796142578, + "learning_rate": 1.0767274268757948e-07, + "loss": 0.5392, + "mean_token_accuracy": 0.8283737301826477, + "num_tokens": 9926789.0, + "step": 255 + }, + { + "epoch": 0.03256583131917059, + "grad_norm": 2.99153995513916, + "learning_rate": 1.0809665112335735e-07, + "loss": 0.4671, + "mean_token_accuracy": 0.8503957986831665, + "num_tokens": 9959506.0, + "step": 256 + }, + { + "epoch": 0.0326930415977611, + "grad_norm": 2.6605782508850098, + "learning_rate": 1.0852055955913522e-07, + "loss": 0.5205, + "mean_token_accuracy": 0.8345227837562561, + "num_tokens": 10000749.0, + "step": 257 + }, + { + "epoch": 0.03282025187635161, + "grad_norm": 2.851363182067871, + "learning_rate": 1.089444679949131e-07, + "loss": 0.5576, + "mean_token_accuracy": 0.8235474228858948, + "num_tokens": 10038379.0, + "step": 258 + }, + { + "epoch": 0.03294746215494212, + "grad_norm": 2.9840691089630127, + "learning_rate": 1.0936837643069097e-07, + "loss": 0.5291, + "mean_token_accuracy": 0.8296732902526855, + "num_tokens": 10073079.0, + "step": 259 + }, + { + "epoch": 0.03307467243353263, + "grad_norm": 2.6161251068115234, + "learning_rate": 1.0979228486646884e-07, + "loss": 0.5362, + "mean_token_accuracy": 0.8320180773735046, + "num_tokens": 10116956.0, + "step": 260 + }, + { + "epoch": 0.03320188271212314, + "grad_norm": 2.6735689640045166, + "learning_rate": 1.1021619330224671e-07, + "loss": 0.4949, + "mean_token_accuracy": 0.8428924679756165, + "num_tokens": 10157174.0, + "step": 261 + }, + { + "epoch": 0.03332909299071365, + "grad_norm": 2.7623183727264404, + "learning_rate": 1.1064010173802458e-07, + "loss": 0.5059, + "mean_token_accuracy": 0.8391542434692383, + "num_tokens": 10195933.0, + "step": 262 + }, + { + "epoch": 0.03345630326930416, + "grad_norm": 2.7110817432403564, + "learning_rate": 1.1106401017380246e-07, + "loss": 0.5704, + "mean_token_accuracy": 0.8234541416168213, + "num_tokens": 10239075.0, + "step": 263 + }, + { + "epoch": 0.03358351354789467, + "grad_norm": 3.1123502254486084, + "learning_rate": 1.1148791860958033e-07, + "loss": 0.545, + "mean_token_accuracy": 0.824012279510498, + "num_tokens": 10271794.0, + "step": 264 + }, + { + "epoch": 0.03371072382648518, + "grad_norm": 2.635190486907959, + "learning_rate": 1.119118270453582e-07, + "loss": 0.484, + "mean_token_accuracy": 0.8437820076942444, + "num_tokens": 10310078.0, + "step": 265 + }, + { + "epoch": 0.03383793410507569, + "grad_norm": 2.925550699234009, + "learning_rate": 1.1233573548113607e-07, + "loss": 0.5139, + "mean_token_accuracy": 0.8401834964752197, + "num_tokens": 10342739.0, + "step": 266 + }, + { + "epoch": 0.0339651443836662, + "grad_norm": 2.9712910652160645, + "learning_rate": 1.1275964391691393e-07, + "loss": 0.5584, + "mean_token_accuracy": 0.8239673376083374, + "num_tokens": 10377636.0, + "step": 267 + }, + { + "epoch": 0.03409235466225671, + "grad_norm": 2.552680253982544, + "learning_rate": 1.131835523526918e-07, + "loss": 0.5128, + "mean_token_accuracy": 0.8352646827697754, + "num_tokens": 10424632.0, + "step": 268 + }, + { + "epoch": 0.03421956494084722, + "grad_norm": 2.6857810020446777, + "learning_rate": 1.1360746078846968e-07, + "loss": 0.5243, + "mean_token_accuracy": 0.8344918489456177, + "num_tokens": 10466755.0, + "step": 269 + }, + { + "epoch": 0.03434677521943773, + "grad_norm": 2.8616340160369873, + "learning_rate": 1.1403136922424755e-07, + "loss": 0.4863, + "mean_token_accuracy": 0.8450928330421448, + "num_tokens": 10503200.0, + "step": 270 + }, + { + "epoch": 0.03447398549802824, + "grad_norm": 2.5440430641174316, + "learning_rate": 1.1445527766002542e-07, + "loss": 0.5198, + "mean_token_accuracy": 0.8353347182273865, + "num_tokens": 10549743.0, + "step": 271 + }, + { + "epoch": 0.03460119577661875, + "grad_norm": 2.5168211460113525, + "learning_rate": 1.148791860958033e-07, + "loss": 0.4715, + "mean_token_accuracy": 0.8503294587135315, + "num_tokens": 10592043.0, + "step": 272 + }, + { + "epoch": 0.034728406055209264, + "grad_norm": 2.5754234790802, + "learning_rate": 1.1530309453158117e-07, + "loss": 0.5176, + "mean_token_accuracy": 0.8370569944381714, + "num_tokens": 10633863.0, + "step": 273 + }, + { + "epoch": 0.03485561633379977, + "grad_norm": 2.7636091709136963, + "learning_rate": 1.1572700296735904e-07, + "loss": 0.5472, + "mean_token_accuracy": 0.8220945596694946, + "num_tokens": 10673167.0, + "step": 274 + }, + { + "epoch": 0.03498282661239028, + "grad_norm": 2.8022072315216064, + "learning_rate": 1.1615091140313691e-07, + "loss": 0.5271, + "mean_token_accuracy": 0.8312933444976807, + "num_tokens": 10711119.0, + "step": 275 + }, + { + "epoch": 0.03511003689098079, + "grad_norm": 2.6058740615844727, + "learning_rate": 1.1657481983891479e-07, + "loss": 0.5085, + "mean_token_accuracy": 0.8390834331512451, + "num_tokens": 10754802.0, + "step": 276 + }, + { + "epoch": 0.0352372471695713, + "grad_norm": 2.964907646179199, + "learning_rate": 1.1699872827469266e-07, + "loss": 0.5181, + "mean_token_accuracy": 0.833001971244812, + "num_tokens": 10789779.0, + "step": 277 + }, + { + "epoch": 0.03536445744816181, + "grad_norm": 2.6113970279693604, + "learning_rate": 1.1742263671047053e-07, + "loss": 0.5082, + "mean_token_accuracy": 0.8359619975090027, + "num_tokens": 10836291.0, + "step": 278 + }, + { + "epoch": 0.03549166772675232, + "grad_norm": 2.7503762245178223, + "learning_rate": 1.178465451462484e-07, + "loss": 0.5342, + "mean_token_accuracy": 0.8303400874137878, + "num_tokens": 10874927.0, + "step": 279 + }, + { + "epoch": 0.03561887800534283, + "grad_norm": 2.765123128890991, + "learning_rate": 1.1827045358202628e-07, + "loss": 0.5352, + "mean_token_accuracy": 0.8272196054458618, + "num_tokens": 10915388.0, + "step": 280 + }, + { + "epoch": 0.035746088283933344, + "grad_norm": 2.9648609161376953, + "learning_rate": 1.1869436201780415e-07, + "loss": 0.5539, + "mean_token_accuracy": 0.8247078657150269, + "num_tokens": 10952907.0, + "step": 281 + }, + { + "epoch": 0.03587329856252385, + "grad_norm": 2.9234397411346436, + "learning_rate": 1.1911827045358202e-07, + "loss": 0.4643, + "mean_token_accuracy": 0.8548400402069092, + "num_tokens": 10987915.0, + "step": 282 + }, + { + "epoch": 0.03600050884111436, + "grad_norm": 2.8612587451934814, + "learning_rate": 1.195421788893599e-07, + "loss": 0.5425, + "mean_token_accuracy": 0.825972318649292, + "num_tokens": 11026781.0, + "step": 283 + }, + { + "epoch": 0.036127719119704874, + "grad_norm": 2.640324115753174, + "learning_rate": 1.1996608732513778e-07, + "loss": 0.4861, + "mean_token_accuracy": 0.8444494009017944, + "num_tokens": 11068539.0, + "step": 284 + }, + { + "epoch": 0.03625492939829538, + "grad_norm": 2.5774221420288086, + "learning_rate": 1.2038999576091563e-07, + "loss": 0.4575, + "mean_token_accuracy": 0.8523976802825928, + "num_tokens": 11109397.0, + "step": 285 + }, + { + "epoch": 0.036382139676885895, + "grad_norm": 2.553209066390991, + "learning_rate": 1.208139041966935e-07, + "loss": 0.5043, + "mean_token_accuracy": 0.8368252515792847, + "num_tokens": 11151955.0, + "step": 286 + }, + { + "epoch": 0.0365093499554764, + "grad_norm": 3.081850290298462, + "learning_rate": 1.2123781263247137e-07, + "loss": 0.5189, + "mean_token_accuracy": 0.8349445462226868, + "num_tokens": 11183834.0, + "step": 287 + }, + { + "epoch": 0.03663656023406691, + "grad_norm": 2.662644147872925, + "learning_rate": 1.2166172106824924e-07, + "loss": 0.5459, + "mean_token_accuracy": 0.8245487213134766, + "num_tokens": 11225388.0, + "step": 288 + }, + { + "epoch": 0.036763770512657425, + "grad_norm": 2.874358892440796, + "learning_rate": 1.2208562950402712e-07, + "loss": 0.5355, + "mean_token_accuracy": 0.8308872580528259, + "num_tokens": 11264874.0, + "step": 289 + }, + { + "epoch": 0.03689098079124793, + "grad_norm": 2.6561014652252197, + "learning_rate": 1.22509537939805e-07, + "loss": 0.4793, + "mean_token_accuracy": 0.8484276533126831, + "num_tokens": 11305140.0, + "step": 290 + }, + { + "epoch": 0.03701819106983844, + "grad_norm": 2.8010923862457275, + "learning_rate": 1.2293344637558286e-07, + "loss": 0.5028, + "mean_token_accuracy": 0.8379591703414917, + "num_tokens": 11341287.0, + "step": 291 + }, + { + "epoch": 0.037145401348428954, + "grad_norm": 2.780644416809082, + "learning_rate": 1.2335735481136073e-07, + "loss": 0.4611, + "mean_token_accuracy": 0.8539893627166748, + "num_tokens": 11378515.0, + "step": 292 + }, + { + "epoch": 0.03727261162701946, + "grad_norm": 2.8670308589935303, + "learning_rate": 1.237812632471386e-07, + "loss": 0.5326, + "mean_token_accuracy": 0.8285370469093323, + "num_tokens": 11417890.0, + "step": 293 + }, + { + "epoch": 0.037399821905609976, + "grad_norm": 3.134093761444092, + "learning_rate": 1.2420517168291648e-07, + "loss": 0.5295, + "mean_token_accuracy": 0.8300285339355469, + "num_tokens": 11451299.0, + "step": 294 + }, + { + "epoch": 0.03752703218420048, + "grad_norm": 2.7301037311553955, + "learning_rate": 1.2462908011869435e-07, + "loss": 0.5072, + "mean_token_accuracy": 0.8394430875778198, + "num_tokens": 11491798.0, + "step": 295 + }, + { + "epoch": 0.03765424246279099, + "grad_norm": 2.701005458831787, + "learning_rate": 1.2505298855447223e-07, + "loss": 0.485, + "mean_token_accuracy": 0.8446699380874634, + "num_tokens": 11528664.0, + "step": 296 + }, + { + "epoch": 0.037781452741381505, + "grad_norm": 2.8868582248687744, + "learning_rate": 1.254768969902501e-07, + "loss": 0.5103, + "mean_token_accuracy": 0.8370888233184814, + "num_tokens": 11564258.0, + "step": 297 + }, + { + "epoch": 0.03790866301997201, + "grad_norm": 2.6449155807495117, + "learning_rate": 1.2590080542602797e-07, + "loss": 0.5005, + "mean_token_accuracy": 0.8397849798202515, + "num_tokens": 11605344.0, + "step": 298 + }, + { + "epoch": 0.03803587329856253, + "grad_norm": 2.498579740524292, + "learning_rate": 1.2632471386180584e-07, + "loss": 0.4417, + "mean_token_accuracy": 0.8588099479675293, + "num_tokens": 11648156.0, + "step": 299 + }, + { + "epoch": 0.038163083577153034, + "grad_norm": 2.586977481842041, + "learning_rate": 1.2674862229758372e-07, + "loss": 0.4488, + "mean_token_accuracy": 0.8536894917488098, + "num_tokens": 11686740.0, + "step": 300 + }, + { + "epoch": 0.03829029385574354, + "grad_norm": 2.508023977279663, + "learning_rate": 1.271725307333616e-07, + "loss": 0.463, + "mean_token_accuracy": 0.8500285148620605, + "num_tokens": 11729498.0, + "step": 301 + }, + { + "epoch": 0.038417504134334056, + "grad_norm": 2.9358267784118652, + "learning_rate": 1.2759643916913946e-07, + "loss": 0.4964, + "mean_token_accuracy": 0.8376650214195251, + "num_tokens": 11764925.0, + "step": 302 + }, + { + "epoch": 0.03854471441292456, + "grad_norm": 2.671342372894287, + "learning_rate": 1.2802034760491733e-07, + "loss": 0.4866, + "mean_token_accuracy": 0.8438529372215271, + "num_tokens": 11804423.0, + "step": 303 + }, + { + "epoch": 0.03867192469151508, + "grad_norm": 2.6527810096740723, + "learning_rate": 1.284442560406952e-07, + "loss": 0.4703, + "mean_token_accuracy": 0.8496280908584595, + "num_tokens": 11846034.0, + "step": 304 + }, + { + "epoch": 0.038799134970105585, + "grad_norm": 2.8308908939361572, + "learning_rate": 1.2886816447647308e-07, + "loss": 0.5165, + "mean_token_accuracy": 0.8339725732803345, + "num_tokens": 11883326.0, + "step": 305 + }, + { + "epoch": 0.03892634524869609, + "grad_norm": 2.7964813709259033, + "learning_rate": 1.2929207291225095e-07, + "loss": 0.528, + "mean_token_accuracy": 0.8330367803573608, + "num_tokens": 11924257.0, + "step": 306 + }, + { + "epoch": 0.03905355552728661, + "grad_norm": 2.9818246364593506, + "learning_rate": 1.2971598134802882e-07, + "loss": 0.4864, + "mean_token_accuracy": 0.8450630307197571, + "num_tokens": 11957018.0, + "step": 307 + }, + { + "epoch": 0.039180765805877114, + "grad_norm": 2.9151196479797363, + "learning_rate": 1.301398897838067e-07, + "loss": 0.5434, + "mean_token_accuracy": 0.8282420635223389, + "num_tokens": 11995694.0, + "step": 308 + }, + { + "epoch": 0.03930797608446762, + "grad_norm": 2.714268922805786, + "learning_rate": 1.3056379821958457e-07, + "loss": 0.4832, + "mean_token_accuracy": 0.843132495880127, + "num_tokens": 12036492.0, + "step": 309 + }, + { + "epoch": 0.039435186363058136, + "grad_norm": 2.7821884155273438, + "learning_rate": 1.3098770665536244e-07, + "loss": 0.5178, + "mean_token_accuracy": 0.8324293494224548, + "num_tokens": 12075932.0, + "step": 310 + }, + { + "epoch": 0.039562396641648644, + "grad_norm": 2.711829900741577, + "learning_rate": 1.3141161509114031e-07, + "loss": 0.4701, + "mean_token_accuracy": 0.8495362401008606, + "num_tokens": 12118194.0, + "step": 311 + }, + { + "epoch": 0.03968960692023916, + "grad_norm": 2.7677206993103027, + "learning_rate": 1.3183552352691819e-07, + "loss": 0.4881, + "mean_token_accuracy": 0.8427667617797852, + "num_tokens": 12160386.0, + "step": 312 + }, + { + "epoch": 0.039816817198829665, + "grad_norm": 2.77972149848938, + "learning_rate": 1.3225943196269603e-07, + "loss": 0.4949, + "mean_token_accuracy": 0.8395684957504272, + "num_tokens": 12203287.0, + "step": 313 + }, + { + "epoch": 0.03994402747742017, + "grad_norm": 3.0107500553131104, + "learning_rate": 1.3268334039847393e-07, + "loss": 0.5114, + "mean_token_accuracy": 0.8395240306854248, + "num_tokens": 12237390.0, + "step": 314 + }, + { + "epoch": 0.04007123775601069, + "grad_norm": 2.7467312812805176, + "learning_rate": 1.3310724883425178e-07, + "loss": 0.4567, + "mean_token_accuracy": 0.851729154586792, + "num_tokens": 12277552.0, + "step": 315 + }, + { + "epoch": 0.040198448034601195, + "grad_norm": 2.8073911666870117, + "learning_rate": 1.3353115727002968e-07, + "loss": 0.4994, + "mean_token_accuracy": 0.8394796252250671, + "num_tokens": 12315385.0, + "step": 316 + }, + { + "epoch": 0.04032565831319171, + "grad_norm": 2.8918569087982178, + "learning_rate": 1.3395506570580752e-07, + "loss": 0.5078, + "mean_token_accuracy": 0.8364330530166626, + "num_tokens": 12352582.0, + "step": 317 + }, + { + "epoch": 0.040452868591782216, + "grad_norm": 2.830057382583618, + "learning_rate": 1.3437897414158542e-07, + "loss": 0.5455, + "mean_token_accuracy": 0.8251610994338989, + "num_tokens": 12396279.0, + "step": 318 + }, + { + "epoch": 0.040580078870372724, + "grad_norm": 2.984844923019409, + "learning_rate": 1.3480288257736327e-07, + "loss": 0.5254, + "mean_token_accuracy": 0.8327940702438354, + "num_tokens": 12431010.0, + "step": 319 + }, + { + "epoch": 0.04070728914896324, + "grad_norm": 2.9617347717285156, + "learning_rate": 1.3522679101314117e-07, + "loss": 0.4697, + "mean_token_accuracy": 0.8453373312950134, + "num_tokens": 12463902.0, + "step": 320 + }, + { + "epoch": 0.040834499427553746, + "grad_norm": 2.7550442218780518, + "learning_rate": 1.35650699448919e-07, + "loss": 0.5351, + "mean_token_accuracy": 0.8233837485313416, + "num_tokens": 12507598.0, + "step": 321 + }, + { + "epoch": 0.04096170970614425, + "grad_norm": 2.855734348297119, + "learning_rate": 1.360746078846969e-07, + "loss": 0.5213, + "mean_token_accuracy": 0.8250024318695068, + "num_tokens": 12547455.0, + "step": 322 + }, + { + "epoch": 0.04108891998473477, + "grad_norm": 2.6593096256256104, + "learning_rate": 1.3649851632047476e-07, + "loss": 0.4757, + "mean_token_accuracy": 0.8432743549346924, + "num_tokens": 12587783.0, + "step": 323 + }, + { + "epoch": 0.041216130263325275, + "grad_norm": 2.7577693462371826, + "learning_rate": 1.3692242475625266e-07, + "loss": 0.5541, + "mean_token_accuracy": 0.824157178401947, + "num_tokens": 12630845.0, + "step": 324 + }, + { + "epoch": 0.04134334054191579, + "grad_norm": 2.5320379734039307, + "learning_rate": 1.373463331920305e-07, + "loss": 0.4465, + "mean_token_accuracy": 0.8559125065803528, + "num_tokens": 12671945.0, + "step": 325 + }, + { + "epoch": 0.0414705508205063, + "grad_norm": 2.5837254524230957, + "learning_rate": 1.377702416278084e-07, + "loss": 0.4741, + "mean_token_accuracy": 0.8458970785140991, + "num_tokens": 12715776.0, + "step": 326 + }, + { + "epoch": 0.041597761099096804, + "grad_norm": 2.7290642261505127, + "learning_rate": 1.3819415006358625e-07, + "loss": 0.4355, + "mean_token_accuracy": 0.856819748878479, + "num_tokens": 12751562.0, + "step": 327 + }, + { + "epoch": 0.04172497137768732, + "grad_norm": 2.886525869369507, + "learning_rate": 1.3861805849936415e-07, + "loss": 0.5329, + "mean_token_accuracy": 0.8297366499900818, + "num_tokens": 12789255.0, + "step": 328 + }, + { + "epoch": 0.041852181656277826, + "grad_norm": 2.8604371547698975, + "learning_rate": 1.39041966935142e-07, + "loss": 0.454, + "mean_token_accuracy": 0.8530953526496887, + "num_tokens": 12824278.0, + "step": 329 + }, + { + "epoch": 0.04197939193486834, + "grad_norm": 2.7607998847961426, + "learning_rate": 1.394658753709199e-07, + "loss": 0.4675, + "mean_token_accuracy": 0.8505017757415771, + "num_tokens": 12862520.0, + "step": 330 + }, + { + "epoch": 0.04210660221345885, + "grad_norm": 2.6804022789001465, + "learning_rate": 1.3988978380669774e-07, + "loss": 0.4411, + "mean_token_accuracy": 0.8554638624191284, + "num_tokens": 12901412.0, + "step": 331 + }, + { + "epoch": 0.042233812492049355, + "grad_norm": 2.4775431156158447, + "learning_rate": 1.403136922424756e-07, + "loss": 0.4399, + "mean_token_accuracy": 0.8566654324531555, + "num_tokens": 12945940.0, + "step": 332 + }, + { + "epoch": 0.04236102277063987, + "grad_norm": 2.65472149848938, + "learning_rate": 1.4073760067825348e-07, + "loss": 0.5065, + "mean_token_accuracy": 0.8408644199371338, + "num_tokens": 12988948.0, + "step": 333 + }, + { + "epoch": 0.04248823304923038, + "grad_norm": 2.750460386276245, + "learning_rate": 1.4116150911403136e-07, + "loss": 0.4613, + "mean_token_accuracy": 0.8493434190750122, + "num_tokens": 13024980.0, + "step": 334 + }, + { + "epoch": 0.04261544332782089, + "grad_norm": 2.757948398590088, + "learning_rate": 1.4158541754980923e-07, + "loss": 0.4749, + "mean_token_accuracy": 0.8454369306564331, + "num_tokens": 13062066.0, + "step": 335 + }, + { + "epoch": 0.0427426536064114, + "grad_norm": 2.630983352661133, + "learning_rate": 1.420093259855871e-07, + "loss": 0.5212, + "mean_token_accuracy": 0.8324450254440308, + "num_tokens": 13104953.0, + "step": 336 + }, + { + "epoch": 0.042869863885001906, + "grad_norm": 2.88224196434021, + "learning_rate": 1.4243323442136497e-07, + "loss": 0.4903, + "mean_token_accuracy": 0.8410887718200684, + "num_tokens": 13142061.0, + "step": 337 + }, + { + "epoch": 0.04299707416359242, + "grad_norm": 2.8240621089935303, + "learning_rate": 1.4285714285714285e-07, + "loss": 0.5134, + "mean_token_accuracy": 0.8340415954589844, + "num_tokens": 13180387.0, + "step": 338 + }, + { + "epoch": 0.04312428444218293, + "grad_norm": 2.909027099609375, + "learning_rate": 1.4328105129292072e-07, + "loss": 0.4995, + "mean_token_accuracy": 0.8410443663597107, + "num_tokens": 13216380.0, + "step": 339 + }, + { + "epoch": 0.043251494720773435, + "grad_norm": 2.603198289871216, + "learning_rate": 1.437049597286986e-07, + "loss": 0.4728, + "mean_token_accuracy": 0.8482728004455566, + "num_tokens": 13258006.0, + "step": 340 + }, + { + "epoch": 0.04337870499936395, + "grad_norm": 2.7620737552642822, + "learning_rate": 1.4412886816447646e-07, + "loss": 0.4426, + "mean_token_accuracy": 0.8550286293029785, + "num_tokens": 13294282.0, + "step": 341 + }, + { + "epoch": 0.04350591527795446, + "grad_norm": 2.590381383895874, + "learning_rate": 1.4455277660025434e-07, + "loss": 0.4416, + "mean_token_accuracy": 0.8537005186080933, + "num_tokens": 13335988.0, + "step": 342 + }, + { + "epoch": 0.04363312555654497, + "grad_norm": 2.6899235248565674, + "learning_rate": 1.449766850360322e-07, + "loss": 0.4582, + "mean_token_accuracy": 0.849513828754425, + "num_tokens": 13372984.0, + "step": 343 + }, + { + "epoch": 0.04376033583513548, + "grad_norm": 2.7348268032073975, + "learning_rate": 1.4540059347181008e-07, + "loss": 0.5054, + "mean_token_accuracy": 0.8373873233795166, + "num_tokens": 13414611.0, + "step": 344 + }, + { + "epoch": 0.043887546113725986, + "grad_norm": 2.877351999282837, + "learning_rate": 1.4582450190758795e-07, + "loss": 0.4705, + "mean_token_accuracy": 0.848885178565979, + "num_tokens": 13450016.0, + "step": 345 + }, + { + "epoch": 0.0440147563923165, + "grad_norm": 2.7454617023468018, + "learning_rate": 1.4624841034336583e-07, + "loss": 0.4499, + "mean_token_accuracy": 0.8519750833511353, + "num_tokens": 13486150.0, + "step": 346 + }, + { + "epoch": 0.04414196667090701, + "grad_norm": 2.7150473594665527, + "learning_rate": 1.466723187791437e-07, + "loss": 0.4614, + "mean_token_accuracy": 0.851051390171051, + "num_tokens": 13524325.0, + "step": 347 + }, + { + "epoch": 0.04426917694949752, + "grad_norm": 2.9243268966674805, + "learning_rate": 1.4709622721492157e-07, + "loss": 0.483, + "mean_token_accuracy": 0.8412215113639832, + "num_tokens": 13561152.0, + "step": 348 + }, + { + "epoch": 0.04439638722808803, + "grad_norm": 3.080989360809326, + "learning_rate": 1.4752013565069942e-07, + "loss": 0.4488, + "mean_token_accuracy": 0.8496854305267334, + "num_tokens": 13593068.0, + "step": 349 + }, + { + "epoch": 0.04452359750667854, + "grad_norm": 3.014843463897705, + "learning_rate": 1.4794404408647732e-07, + "loss": 0.5046, + "mean_token_accuracy": 0.8334959149360657, + "num_tokens": 13627511.0, + "step": 350 + }, + { + "epoch": 0.04465080778526905, + "grad_norm": 2.471921920776367, + "learning_rate": 1.4836795252225516e-07, + "loss": 0.467, + "mean_token_accuracy": 0.8499644994735718, + "num_tokens": 13672274.0, + "step": 351 + }, + { + "epoch": 0.04477801806385956, + "grad_norm": 2.7758195400238037, + "learning_rate": 1.4879186095803306e-07, + "loss": 0.4466, + "mean_token_accuracy": 0.8552318811416626, + "num_tokens": 13711078.0, + "step": 352 + }, + { + "epoch": 0.04490522834245007, + "grad_norm": 2.6332733631134033, + "learning_rate": 1.492157693938109e-07, + "loss": 0.4642, + "mean_token_accuracy": 0.8502146005630493, + "num_tokens": 13753054.0, + "step": 353 + }, + { + "epoch": 0.04503243862104058, + "grad_norm": 2.885486125946045, + "learning_rate": 1.496396778295888e-07, + "loss": 0.4562, + "mean_token_accuracy": 0.8505034446716309, + "num_tokens": 13787573.0, + "step": 354 + }, + { + "epoch": 0.04515964889963109, + "grad_norm": 2.864490270614624, + "learning_rate": 1.5006358626536665e-07, + "loss": 0.47, + "mean_token_accuracy": 0.8485416173934937, + "num_tokens": 13825973.0, + "step": 355 + }, + { + "epoch": 0.0452868591782216, + "grad_norm": 2.9414737224578857, + "learning_rate": 1.5048749470114455e-07, + "loss": 0.5107, + "mean_token_accuracy": 0.8378837704658508, + "num_tokens": 13863527.0, + "step": 356 + }, + { + "epoch": 0.04541406945681211, + "grad_norm": 2.8023176193237305, + "learning_rate": 1.509114031369224e-07, + "loss": 0.5004, + "mean_token_accuracy": 0.8375387191772461, + "num_tokens": 13906798.0, + "step": 357 + }, + { + "epoch": 0.04554127973540262, + "grad_norm": 2.6622655391693115, + "learning_rate": 1.513353115727003e-07, + "loss": 0.4636, + "mean_token_accuracy": 0.8484257459640503, + "num_tokens": 13946432.0, + "step": 358 + }, + { + "epoch": 0.04566849001399313, + "grad_norm": 2.6676177978515625, + "learning_rate": 1.5175922000847814e-07, + "loss": 0.4858, + "mean_token_accuracy": 0.8466971516609192, + "num_tokens": 13989620.0, + "step": 359 + }, + { + "epoch": 0.04579570029258364, + "grad_norm": 2.873718738555908, + "learning_rate": 1.5218312844425604e-07, + "loss": 0.4697, + "mean_token_accuracy": 0.8458429574966431, + "num_tokens": 14026237.0, + "step": 360 + }, + { + "epoch": 0.045922910571174154, + "grad_norm": 2.604780435562134, + "learning_rate": 1.526070368800339e-07, + "loss": 0.4667, + "mean_token_accuracy": 0.8457894325256348, + "num_tokens": 14066855.0, + "step": 361 + }, + { + "epoch": 0.04605012084976466, + "grad_norm": 3.075986623764038, + "learning_rate": 1.530309453158118e-07, + "loss": 0.489, + "mean_token_accuracy": 0.839026927947998, + "num_tokens": 14100105.0, + "step": 362 + }, + { + "epoch": 0.04617733112835517, + "grad_norm": 3.05810546875, + "learning_rate": 1.5345485375158963e-07, + "loss": 0.467, + "mean_token_accuracy": 0.8453057408332825, + "num_tokens": 14134288.0, + "step": 363 + }, + { + "epoch": 0.04630454140694568, + "grad_norm": 2.8740580081939697, + "learning_rate": 1.5387876218736753e-07, + "loss": 0.5236, + "mean_token_accuracy": 0.8306488990783691, + "num_tokens": 14172657.0, + "step": 364 + }, + { + "epoch": 0.04643175168553619, + "grad_norm": 2.632493495941162, + "learning_rate": 1.5430267062314538e-07, + "loss": 0.4539, + "mean_token_accuracy": 0.8529391884803772, + "num_tokens": 14214766.0, + "step": 365 + }, + { + "epoch": 0.0465589619641267, + "grad_norm": 4.486075401306152, + "learning_rate": 1.5472657905892328e-07, + "loss": 0.4556, + "mean_token_accuracy": 0.8534196019172668, + "num_tokens": 14260607.0, + "step": 366 + }, + { + "epoch": 0.04668617224271721, + "grad_norm": 2.6022109985351562, + "learning_rate": 1.5515048749470113e-07, + "loss": 0.4436, + "mean_token_accuracy": 0.8518645763397217, + "num_tokens": 14304645.0, + "step": 367 + }, + { + "epoch": 0.04681338252130772, + "grad_norm": 2.623616933822632, + "learning_rate": 1.55574395930479e-07, + "loss": 0.4592, + "mean_token_accuracy": 0.8510429263114929, + "num_tokens": 14344021.0, + "step": 368 + }, + { + "epoch": 0.046940592799898234, + "grad_norm": 2.6043176651000977, + "learning_rate": 1.5599830436625687e-07, + "loss": 0.4705, + "mean_token_accuracy": 0.8472155332565308, + "num_tokens": 14387740.0, + "step": 369 + }, + { + "epoch": 0.04706780307848874, + "grad_norm": 3.019803285598755, + "learning_rate": 1.5642221280203474e-07, + "loss": 0.4975, + "mean_token_accuracy": 0.8417560458183289, + "num_tokens": 14433265.0, + "step": 370 + }, + { + "epoch": 0.04719501335707925, + "grad_norm": 2.9781131744384766, + "learning_rate": 1.5684612123781262e-07, + "loss": 0.4973, + "mean_token_accuracy": 0.8379999995231628, + "num_tokens": 14472686.0, + "step": 371 + }, + { + "epoch": 0.04732222363566976, + "grad_norm": 2.853092670440674, + "learning_rate": 1.572700296735905e-07, + "loss": 0.4553, + "mean_token_accuracy": 0.8542243242263794, + "num_tokens": 14508565.0, + "step": 372 + }, + { + "epoch": 0.04744943391426027, + "grad_norm": 2.733144521713257, + "learning_rate": 1.576939381093684e-07, + "loss": 0.5032, + "mean_token_accuracy": 0.8392324447631836, + "num_tokens": 14549812.0, + "step": 373 + }, + { + "epoch": 0.047576644192850785, + "grad_norm": 2.8902101516723633, + "learning_rate": 1.5811784654514623e-07, + "loss": 0.4681, + "mean_token_accuracy": 0.8439010381698608, + "num_tokens": 14588694.0, + "step": 374 + }, + { + "epoch": 0.04770385447144129, + "grad_norm": 2.463256597518921, + "learning_rate": 1.5854175498092413e-07, + "loss": 0.3918, + "mean_token_accuracy": 0.8704352378845215, + "num_tokens": 14627263.0, + "step": 375 + }, + { + "epoch": 0.0478310647500318, + "grad_norm": 2.898787498474121, + "learning_rate": 1.5896566341670198e-07, + "loss": 0.4487, + "mean_token_accuracy": 0.8511735796928406, + "num_tokens": 14663478.0, + "step": 376 + }, + { + "epoch": 0.047958275028622314, + "grad_norm": 2.8990390300750732, + "learning_rate": 1.5938957185247988e-07, + "loss": 0.5083, + "mean_token_accuracy": 0.8388836979866028, + "num_tokens": 14701475.0, + "step": 377 + }, + { + "epoch": 0.04808548530721282, + "grad_norm": 2.7646262645721436, + "learning_rate": 1.5981348028825772e-07, + "loss": 0.5759, + "mean_token_accuracy": 0.8153548836708069, + "num_tokens": 14746786.0, + "step": 378 + }, + { + "epoch": 0.048212695585803336, + "grad_norm": 2.919665575027466, + "learning_rate": 1.6023738872403562e-07, + "loss": 0.5251, + "mean_token_accuracy": 0.8318307995796204, + "num_tokens": 14782970.0, + "step": 379 + }, + { + "epoch": 0.048339905864393844, + "grad_norm": 2.8658318519592285, + "learning_rate": 1.6066129715981347e-07, + "loss": 0.4489, + "mean_token_accuracy": 0.8529468178749084, + "num_tokens": 14818930.0, + "step": 380 + }, + { + "epoch": 0.04846711614298435, + "grad_norm": 2.7125988006591797, + "learning_rate": 1.6108520559559137e-07, + "loss": 0.4647, + "mean_token_accuracy": 0.8506039381027222, + "num_tokens": 14858317.0, + "step": 381 + }, + { + "epoch": 0.048594326421574865, + "grad_norm": 2.83148455619812, + "learning_rate": 1.6150911403136921e-07, + "loss": 0.5079, + "mean_token_accuracy": 0.8363046050071716, + "num_tokens": 14895904.0, + "step": 382 + }, + { + "epoch": 0.04872153670016537, + "grad_norm": 2.7429258823394775, + "learning_rate": 1.619330224671471e-07, + "loss": 0.4351, + "mean_token_accuracy": 0.8542770743370056, + "num_tokens": 14935774.0, + "step": 383 + }, + { + "epoch": 0.04884874697875588, + "grad_norm": 2.80892014503479, + "learning_rate": 1.6235693090292496e-07, + "loss": 0.4488, + "mean_token_accuracy": 0.8526856899261475, + "num_tokens": 14972717.0, + "step": 384 + }, + { + "epoch": 0.048975957257346395, + "grad_norm": 3.006998062133789, + "learning_rate": 1.6278083933870286e-07, + "loss": 0.5297, + "mean_token_accuracy": 0.8293236494064331, + "num_tokens": 15009854.0, + "step": 385 + }, + { + "epoch": 0.0491031675359369, + "grad_norm": 2.7395570278167725, + "learning_rate": 1.632047477744807e-07, + "loss": 0.4931, + "mean_token_accuracy": 0.8405119180679321, + "num_tokens": 15051101.0, + "step": 386 + }, + { + "epoch": 0.049230377814527417, + "grad_norm": 2.807176351547241, + "learning_rate": 1.6362865621025858e-07, + "loss": 0.4638, + "mean_token_accuracy": 0.8528451919555664, + "num_tokens": 15089482.0, + "step": 387 + }, + { + "epoch": 0.049357588093117924, + "grad_norm": 2.9286367893218994, + "learning_rate": 1.6405256464603645e-07, + "loss": 0.4679, + "mean_token_accuracy": 0.8479813933372498, + "num_tokens": 15125062.0, + "step": 388 + }, + { + "epoch": 0.04948479837170843, + "grad_norm": 2.824570894241333, + "learning_rate": 1.6447647308181432e-07, + "loss": 0.513, + "mean_token_accuracy": 0.8361847400665283, + "num_tokens": 15166069.0, + "step": 389 + }, + { + "epoch": 0.049612008650298946, + "grad_norm": 3.148928642272949, + "learning_rate": 1.649003815175922e-07, + "loss": 0.5027, + "mean_token_accuracy": 0.8381463885307312, + "num_tokens": 15200484.0, + "step": 390 + }, + { + "epoch": 0.04973921892888945, + "grad_norm": 2.6620264053344727, + "learning_rate": 1.6532428995337007e-07, + "loss": 0.4551, + "mean_token_accuracy": 0.8535119295120239, + "num_tokens": 15240275.0, + "step": 391 + }, + { + "epoch": 0.04986642920747997, + "grad_norm": 3.0769472122192383, + "learning_rate": 1.6574819838914794e-07, + "loss": 0.5128, + "mean_token_accuracy": 0.8336455821990967, + "num_tokens": 15275781.0, + "step": 392 + }, + { + "epoch": 0.049993639486070475, + "grad_norm": 2.9100425243377686, + "learning_rate": 1.661721068249258e-07, + "loss": 0.4852, + "mean_token_accuracy": 0.8424484133720398, + "num_tokens": 15311139.0, + "step": 393 + }, + { + "epoch": 0.05012084976466098, + "grad_norm": 2.957288980484009, + "learning_rate": 1.6659601526070368e-07, + "loss": 0.5099, + "mean_token_accuracy": 0.8385380506515503, + "num_tokens": 15348288.0, + "step": 394 + }, + { + "epoch": 0.0502480600432515, + "grad_norm": 2.7945470809936523, + "learning_rate": 1.6701992369648156e-07, + "loss": 0.4784, + "mean_token_accuracy": 0.8448497653007507, + "num_tokens": 15386215.0, + "step": 395 + }, + { + "epoch": 0.050375270321842004, + "grad_norm": 2.869891405105591, + "learning_rate": 1.6744383213225943e-07, + "loss": 0.4345, + "mean_token_accuracy": 0.8578327894210815, + "num_tokens": 15421470.0, + "step": 396 + }, + { + "epoch": 0.05050248060043251, + "grad_norm": 2.961404800415039, + "learning_rate": 1.678677405680373e-07, + "loss": 0.4414, + "mean_token_accuracy": 0.8542053699493408, + "num_tokens": 15456192.0, + "step": 397 + }, + { + "epoch": 0.050629690879023026, + "grad_norm": 2.8163681030273438, + "learning_rate": 1.6829164900381518e-07, + "loss": 0.4834, + "mean_token_accuracy": 0.8458978533744812, + "num_tokens": 15498201.0, + "step": 398 + }, + { + "epoch": 0.05075690115761353, + "grad_norm": 2.9781839847564697, + "learning_rate": 1.6871555743959305e-07, + "loss": 0.488, + "mean_token_accuracy": 0.8428382873535156, + "num_tokens": 15535326.0, + "step": 399 + }, + { + "epoch": 0.05088411143620405, + "grad_norm": 2.824734926223755, + "learning_rate": 1.6913946587537092e-07, + "loss": 0.5086, + "mean_token_accuracy": 0.8358142375946045, + "num_tokens": 15574251.0, + "step": 400 + }, + { + "epoch": 0.051011321714794555, + "grad_norm": 2.6947181224823, + "learning_rate": 1.695633743111488e-07, + "loss": 0.4215, + "mean_token_accuracy": 0.8614190220832825, + "num_tokens": 15614047.0, + "step": 401 + }, + { + "epoch": 0.05113853199338506, + "grad_norm": 3.0086052417755127, + "learning_rate": 1.6998728274692667e-07, + "loss": 0.4692, + "mean_token_accuracy": 0.8465609550476074, + "num_tokens": 15649211.0, + "step": 402 + }, + { + "epoch": 0.05126574227197558, + "grad_norm": 2.965179204940796, + "learning_rate": 1.7041119118270454e-07, + "loss": 0.4516, + "mean_token_accuracy": 0.8508898615837097, + "num_tokens": 15683603.0, + "step": 403 + }, + { + "epoch": 0.051392952550566084, + "grad_norm": 2.7848305702209473, + "learning_rate": 1.7083509961848238e-07, + "loss": 0.442, + "mean_token_accuracy": 0.8571701645851135, + "num_tokens": 15721079.0, + "step": 404 + }, + { + "epoch": 0.0515201628291566, + "grad_norm": 2.8024418354034424, + "learning_rate": 1.7125900805426028e-07, + "loss": 0.4829, + "mean_token_accuracy": 0.8423936367034912, + "num_tokens": 15757622.0, + "step": 405 + }, + { + "epoch": 0.051647373107747106, + "grad_norm": 2.822634696960449, + "learning_rate": 1.7168291649003813e-07, + "loss": 0.4726, + "mean_token_accuracy": 0.8436732292175293, + "num_tokens": 15792691.0, + "step": 406 + }, + { + "epoch": 0.051774583386337614, + "grad_norm": 2.615941286087036, + "learning_rate": 1.7210682492581603e-07, + "loss": 0.4336, + "mean_token_accuracy": 0.8581984043121338, + "num_tokens": 15831893.0, + "step": 407 + }, + { + "epoch": 0.05190179366492813, + "grad_norm": 2.8585944175720215, + "learning_rate": 1.7253073336159387e-07, + "loss": 0.492, + "mean_token_accuracy": 0.8391366600990295, + "num_tokens": 15866958.0, + "step": 408 + }, + { + "epoch": 0.052029003943518635, + "grad_norm": 2.604957103729248, + "learning_rate": 1.7295464179737177e-07, + "loss": 0.4889, + "mean_token_accuracy": 0.8452004194259644, + "num_tokens": 15910338.0, + "step": 409 + }, + { + "epoch": 0.05215621422210915, + "grad_norm": 3.011876106262207, + "learning_rate": 1.7337855023314962e-07, + "loss": 0.504, + "mean_token_accuracy": 0.8349363803863525, + "num_tokens": 15945082.0, + "step": 410 + }, + { + "epoch": 0.05228342450069966, + "grad_norm": 2.747854232788086, + "learning_rate": 1.7380245866892752e-07, + "loss": 0.4337, + "mean_token_accuracy": 0.8590865135192871, + "num_tokens": 15983518.0, + "step": 411 + }, + { + "epoch": 0.052410634779290165, + "grad_norm": 2.7817296981811523, + "learning_rate": 1.7422636710470536e-07, + "loss": 0.5095, + "mean_token_accuracy": 0.8389014601707458, + "num_tokens": 16024604.0, + "step": 412 + }, + { + "epoch": 0.05253784505788068, + "grad_norm": 2.6929080486297607, + "learning_rate": 1.7465027554048326e-07, + "loss": 0.4529, + "mean_token_accuracy": 0.8529760837554932, + "num_tokens": 16064475.0, + "step": 413 + }, + { + "epoch": 0.05266505533647119, + "grad_norm": 2.7596795558929443, + "learning_rate": 1.750741839762611e-07, + "loss": 0.4569, + "mean_token_accuracy": 0.8512924909591675, + "num_tokens": 16100715.0, + "step": 414 + }, + { + "epoch": 0.052792265615061694, + "grad_norm": 2.726365089416504, + "learning_rate": 1.75498092412039e-07, + "loss": 0.4611, + "mean_token_accuracy": 0.8492721319198608, + "num_tokens": 16142627.0, + "step": 415 + }, + { + "epoch": 0.05291947589365221, + "grad_norm": 2.383010149002075, + "learning_rate": 1.7592200084781686e-07, + "loss": 0.4317, + "mean_token_accuracy": 0.859312891960144, + "num_tokens": 16190882.0, + "step": 416 + }, + { + "epoch": 0.053046686172242716, + "grad_norm": 3.1917049884796143, + "learning_rate": 1.7634590928359475e-07, + "loss": 0.519, + "mean_token_accuracy": 0.8362757563591003, + "num_tokens": 16225847.0, + "step": 417 + }, + { + "epoch": 0.05317389645083323, + "grad_norm": 2.945664167404175, + "learning_rate": 1.767698177193726e-07, + "loss": 0.4915, + "mean_token_accuracy": 0.843975841999054, + "num_tokens": 16262172.0, + "step": 418 + }, + { + "epoch": 0.05330110672942374, + "grad_norm": 2.6461546421051025, + "learning_rate": 1.771937261551505e-07, + "loss": 0.4479, + "mean_token_accuracy": 0.8539419174194336, + "num_tokens": 16301858.0, + "step": 419 + }, + { + "epoch": 0.053428317008014245, + "grad_norm": 2.7596962451934814, + "learning_rate": 1.7761763459092835e-07, + "loss": 0.5215, + "mean_token_accuracy": 0.8333034515380859, + "num_tokens": 16345346.0, + "step": 420 + }, + { + "epoch": 0.05355552728660476, + "grad_norm": 2.9948019981384277, + "learning_rate": 1.7804154302670624e-07, + "loss": 0.4901, + "mean_token_accuracy": 0.8426105380058289, + "num_tokens": 16381292.0, + "step": 421 + }, + { + "epoch": 0.05368273756519527, + "grad_norm": 2.716113805770874, + "learning_rate": 1.784654514624841e-07, + "loss": 0.4298, + "mean_token_accuracy": 0.8580721616744995, + "num_tokens": 16418052.0, + "step": 422 + }, + { + "epoch": 0.05380994784378578, + "grad_norm": 2.625749111175537, + "learning_rate": 1.7888935989826196e-07, + "loss": 0.4474, + "mean_token_accuracy": 0.8533780574798584, + "num_tokens": 16461647.0, + "step": 423 + }, + { + "epoch": 0.05393715812237629, + "grad_norm": 2.682250738143921, + "learning_rate": 1.7931326833403984e-07, + "loss": 0.4855, + "mean_token_accuracy": 0.8451309204101562, + "num_tokens": 16504133.0, + "step": 424 + }, + { + "epoch": 0.054064368400966796, + "grad_norm": 2.709491729736328, + "learning_rate": 1.797371767698177e-07, + "loss": 0.4497, + "mean_token_accuracy": 0.8529797196388245, + "num_tokens": 16542785.0, + "step": 425 + }, + { + "epoch": 0.05419157867955731, + "grad_norm": 2.762908697128296, + "learning_rate": 1.8016108520559558e-07, + "loss": 0.4584, + "mean_token_accuracy": 0.8527382612228394, + "num_tokens": 16581300.0, + "step": 426 + }, + { + "epoch": 0.05431878895814782, + "grad_norm": 2.594954252243042, + "learning_rate": 1.8058499364137345e-07, + "loss": 0.4337, + "mean_token_accuracy": 0.8552451133728027, + "num_tokens": 16624072.0, + "step": 427 + }, + { + "epoch": 0.054445999236738325, + "grad_norm": 2.642185926437378, + "learning_rate": 1.8100890207715133e-07, + "loss": 0.4821, + "mean_token_accuracy": 0.8430174589157104, + "num_tokens": 16671593.0, + "step": 428 + }, + { + "epoch": 0.05457320951532884, + "grad_norm": 2.760733127593994, + "learning_rate": 1.814328105129292e-07, + "loss": 0.486, + "mean_token_accuracy": 0.8466005325317383, + "num_tokens": 16711833.0, + "step": 429 + }, + { + "epoch": 0.05470041979391935, + "grad_norm": 2.7072737216949463, + "learning_rate": 1.8185671894870707e-07, + "loss": 0.453, + "mean_token_accuracy": 0.8523850440979004, + "num_tokens": 16751418.0, + "step": 430 + }, + { + "epoch": 0.05482763007250986, + "grad_norm": 3.018636465072632, + "learning_rate": 1.8228062738448494e-07, + "loss": 0.4429, + "mean_token_accuracy": 0.8564432859420776, + "num_tokens": 16784117.0, + "step": 431 + }, + { + "epoch": 0.05495484035110037, + "grad_norm": 2.829690456390381, + "learning_rate": 1.8270453582026282e-07, + "loss": 0.4798, + "mean_token_accuracy": 0.839066207408905, + "num_tokens": 16823617.0, + "step": 432 + }, + { + "epoch": 0.055082050629690876, + "grad_norm": 2.7565059661865234, + "learning_rate": 1.831284442560407e-07, + "loss": 0.4295, + "mean_token_accuracy": 0.8596795797348022, + "num_tokens": 16860120.0, + "step": 433 + }, + { + "epoch": 0.05520926090828139, + "grad_norm": 3.2705862522125244, + "learning_rate": 1.8355235269181856e-07, + "loss": 0.4422, + "mean_token_accuracy": 0.856634259223938, + "num_tokens": 16891938.0, + "step": 434 + }, + { + "epoch": 0.0553364711868719, + "grad_norm": 2.901871919631958, + "learning_rate": 1.8397626112759643e-07, + "loss": 0.4282, + "mean_token_accuracy": 0.8587947487831116, + "num_tokens": 16927022.0, + "step": 435 + }, + { + "epoch": 0.05546368146546241, + "grad_norm": 3.0686912536621094, + "learning_rate": 1.844001695633743e-07, + "loss": 0.5236, + "mean_token_accuracy": 0.8303632736206055, + "num_tokens": 16965843.0, + "step": 436 + }, + { + "epoch": 0.05559089174405292, + "grad_norm": 2.657994031906128, + "learning_rate": 1.8482407799915218e-07, + "loss": 0.4818, + "mean_token_accuracy": 0.8435001969337463, + "num_tokens": 17006708.0, + "step": 437 + }, + { + "epoch": 0.05571810202264343, + "grad_norm": 2.804105281829834, + "learning_rate": 1.8524798643493005e-07, + "loss": 0.5004, + "mean_token_accuracy": 0.8376768827438354, + "num_tokens": 17048565.0, + "step": 438 + }, + { + "epoch": 0.05584531230123394, + "grad_norm": 2.951634407043457, + "learning_rate": 1.8567189487070792e-07, + "loss": 0.4742, + "mean_token_accuracy": 0.847723662853241, + "num_tokens": 17088969.0, + "step": 439 + }, + { + "epoch": 0.05597252257982445, + "grad_norm": 3.0625174045562744, + "learning_rate": 1.8609580330648577e-07, + "loss": 0.4537, + "mean_token_accuracy": 0.852565348148346, + "num_tokens": 17123690.0, + "step": 440 + }, + { + "epoch": 0.05609973285841496, + "grad_norm": 2.8689358234405518, + "learning_rate": 1.8651971174226367e-07, + "loss": 0.4675, + "mean_token_accuracy": 0.8479344248771667, + "num_tokens": 17160488.0, + "step": 441 + }, + { + "epoch": 0.05622694313700547, + "grad_norm": 2.9512217044830322, + "learning_rate": 1.8694362017804152e-07, + "loss": 0.4548, + "mean_token_accuracy": 0.8504582047462463, + "num_tokens": 17198691.0, + "step": 442 + }, + { + "epoch": 0.05635415341559598, + "grad_norm": 2.9429588317871094, + "learning_rate": 1.8736752861381941e-07, + "loss": 0.4581, + "mean_token_accuracy": 0.8488172888755798, + "num_tokens": 17238648.0, + "step": 443 + }, + { + "epoch": 0.05648136369418649, + "grad_norm": 2.6748712062835693, + "learning_rate": 1.8779143704959726e-07, + "loss": 0.4757, + "mean_token_accuracy": 0.8458440899848938, + "num_tokens": 17281064.0, + "step": 444 + }, + { + "epoch": 0.056608573972777, + "grad_norm": 3.210848808288574, + "learning_rate": 1.8821534548537516e-07, + "loss": 0.441, + "mean_token_accuracy": 0.8515665531158447, + "num_tokens": 17312402.0, + "step": 445 + }, + { + "epoch": 0.05673578425136751, + "grad_norm": 3.1427717208862305, + "learning_rate": 1.88639253921153e-07, + "loss": 0.5737, + "mean_token_accuracy": 0.8141032457351685, + "num_tokens": 17349318.0, + "step": 446 + }, + { + "epoch": 0.05686299452995802, + "grad_norm": 2.536829948425293, + "learning_rate": 1.890631623569309e-07, + "loss": 0.4289, + "mean_token_accuracy": 0.8573183417320251, + "num_tokens": 17391069.0, + "step": 447 + }, + { + "epoch": 0.05699020480854853, + "grad_norm": 2.6771676540374756, + "learning_rate": 1.8948707079270875e-07, + "loss": 0.4812, + "mean_token_accuracy": 0.8411345481872559, + "num_tokens": 17434550.0, + "step": 448 + }, + { + "epoch": 0.057117415087139044, + "grad_norm": 2.6422080993652344, + "learning_rate": 1.8991097922848665e-07, + "loss": 0.4504, + "mean_token_accuracy": 0.8530607223510742, + "num_tokens": 17477021.0, + "step": 449 + }, + { + "epoch": 0.05724462536572955, + "grad_norm": 3.3117358684539795, + "learning_rate": 1.903348876642645e-07, + "loss": 0.4826, + "mean_token_accuracy": 0.8420759439468384, + "num_tokens": 17509624.0, + "step": 450 + }, + { + "epoch": 0.05737183564432006, + "grad_norm": 3.0059893131256104, + "learning_rate": 1.907587961000424e-07, + "loss": 0.4749, + "mean_token_accuracy": 0.8461908102035522, + "num_tokens": 17545027.0, + "step": 451 + }, + { + "epoch": 0.05749904592291057, + "grad_norm": 2.672466993331909, + "learning_rate": 1.9118270453582024e-07, + "loss": 0.441, + "mean_token_accuracy": 0.8529188632965088, + "num_tokens": 17585410.0, + "step": 452 + }, + { + "epoch": 0.05762625620150108, + "grad_norm": 2.6237905025482178, + "learning_rate": 1.9160661297159814e-07, + "loss": 0.4519, + "mean_token_accuracy": 0.8518626689910889, + "num_tokens": 17626822.0, + "step": 453 + }, + { + "epoch": 0.057753466480091595, + "grad_norm": 3.0087685585021973, + "learning_rate": 1.9203052140737599e-07, + "loss": 0.4514, + "mean_token_accuracy": 0.8523250818252563, + "num_tokens": 17662567.0, + "step": 454 + }, + { + "epoch": 0.0578806767586821, + "grad_norm": 2.893582344055176, + "learning_rate": 1.9245442984315389e-07, + "loss": 0.4655, + "mean_token_accuracy": 0.8469316959381104, + "num_tokens": 17701860.0, + "step": 455 + }, + { + "epoch": 0.05800788703727261, + "grad_norm": 2.6529011726379395, + "learning_rate": 1.9287833827893173e-07, + "loss": 0.48, + "mean_token_accuracy": 0.8442368507385254, + "num_tokens": 17744903.0, + "step": 456 + }, + { + "epoch": 0.058135097315863124, + "grad_norm": 3.116649866104126, + "learning_rate": 1.9330224671470963e-07, + "loss": 0.4467, + "mean_token_accuracy": 0.8522716760635376, + "num_tokens": 17779311.0, + "step": 457 + }, + { + "epoch": 0.05826230759445363, + "grad_norm": 2.6879069805145264, + "learning_rate": 1.9372615515048748e-07, + "loss": 0.4616, + "mean_token_accuracy": 0.851274847984314, + "num_tokens": 17821965.0, + "step": 458 + }, + { + "epoch": 0.05838951787304414, + "grad_norm": 2.861629009246826, + "learning_rate": 1.9415006358626535e-07, + "loss": 0.4462, + "mean_token_accuracy": 0.8554117679595947, + "num_tokens": 17859995.0, + "step": 459 + }, + { + "epoch": 0.05851672815163465, + "grad_norm": 2.59212327003479, + "learning_rate": 1.9457397202204322e-07, + "loss": 0.5051, + "mean_token_accuracy": 0.8370145559310913, + "num_tokens": 17909827.0, + "step": 460 + }, + { + "epoch": 0.05864393843022516, + "grad_norm": 2.9530258178710938, + "learning_rate": 1.949978804578211e-07, + "loss": 0.4364, + "mean_token_accuracy": 0.8537482023239136, + "num_tokens": 17945870.0, + "step": 461 + }, + { + "epoch": 0.058771148708815675, + "grad_norm": 2.8606696128845215, + "learning_rate": 1.9542178889359897e-07, + "loss": 0.4864, + "mean_token_accuracy": 0.8386588096618652, + "num_tokens": 17988571.0, + "step": 462 + }, + { + "epoch": 0.05889835898740618, + "grad_norm": 2.887407064437866, + "learning_rate": 1.9584569732937684e-07, + "loss": 0.4311, + "mean_token_accuracy": 0.8562655448913574, + "num_tokens": 18025392.0, + "step": 463 + }, + { + "epoch": 0.05902556926599669, + "grad_norm": 3.3175721168518066, + "learning_rate": 1.962696057651547e-07, + "loss": 0.496, + "mean_token_accuracy": 0.8347387313842773, + "num_tokens": 18060218.0, + "step": 464 + }, + { + "epoch": 0.059152779544587204, + "grad_norm": 2.834130048751831, + "learning_rate": 1.9669351420093258e-07, + "loss": 0.4812, + "mean_token_accuracy": 0.8455079793930054, + "num_tokens": 18099113.0, + "step": 465 + }, + { + "epoch": 0.05927998982317771, + "grad_norm": 2.5061142444610596, + "learning_rate": 1.9711742263671046e-07, + "loss": 0.4478, + "mean_token_accuracy": 0.8495842218399048, + "num_tokens": 18144697.0, + "step": 466 + }, + { + "epoch": 0.059407200101768226, + "grad_norm": 2.9198877811431885, + "learning_rate": 1.9754133107248833e-07, + "loss": 0.4718, + "mean_token_accuracy": 0.8466806411743164, + "num_tokens": 18181581.0, + "step": 467 + }, + { + "epoch": 0.059534410380358734, + "grad_norm": 2.909285068511963, + "learning_rate": 1.979652395082662e-07, + "loss": 0.4738, + "mean_token_accuracy": 0.8419240713119507, + "num_tokens": 18218695.0, + "step": 468 + }, + { + "epoch": 0.05966162065894924, + "grad_norm": 2.836836099624634, + "learning_rate": 1.9838914794404408e-07, + "loss": 0.4733, + "mean_token_accuracy": 0.8459504842758179, + "num_tokens": 18259689.0, + "step": 469 + }, + { + "epoch": 0.059788830937539755, + "grad_norm": 2.7259256839752197, + "learning_rate": 1.9881305637982195e-07, + "loss": 0.4453, + "mean_token_accuracy": 0.8556722402572632, + "num_tokens": 18298961.0, + "step": 470 + }, + { + "epoch": 0.05991604121613026, + "grad_norm": 2.804755926132202, + "learning_rate": 1.9923696481559982e-07, + "loss": 0.4337, + "mean_token_accuracy": 0.8568069338798523, + "num_tokens": 18338143.0, + "step": 471 + }, + { + "epoch": 0.06004325149472077, + "grad_norm": 3.186272144317627, + "learning_rate": 1.996608732513777e-07, + "loss": 0.4776, + "mean_token_accuracy": 0.84787917137146, + "num_tokens": 18372421.0, + "step": 472 + }, + { + "epoch": 0.060170461773311285, + "grad_norm": 2.995964288711548, + "learning_rate": 2.0008478168715557e-07, + "loss": 0.4785, + "mean_token_accuracy": 0.8503095507621765, + "num_tokens": 18408553.0, + "step": 473 + }, + { + "epoch": 0.06029767205190179, + "grad_norm": 2.778333902359009, + "learning_rate": 2.0050869012293344e-07, + "loss": 0.4758, + "mean_token_accuracy": 0.8463001251220703, + "num_tokens": 18448978.0, + "step": 474 + }, + { + "epoch": 0.060424882330492306, + "grad_norm": 2.6796507835388184, + "learning_rate": 2.009325985587113e-07, + "loss": 0.4495, + "mean_token_accuracy": 0.8479381799697876, + "num_tokens": 18492525.0, + "step": 475 + }, + { + "epoch": 0.060552092609082814, + "grad_norm": 2.720040798187256, + "learning_rate": 2.0135650699448918e-07, + "loss": 0.4463, + "mean_token_accuracy": 0.8542692065238953, + "num_tokens": 18537958.0, + "step": 476 + }, + { + "epoch": 0.06067930288767332, + "grad_norm": 2.67600417137146, + "learning_rate": 2.0178041543026706e-07, + "loss": 0.4125, + "mean_token_accuracy": 0.8653149604797363, + "num_tokens": 18580700.0, + "step": 477 + }, + { + "epoch": 0.060806513166263836, + "grad_norm": 2.831871509552002, + "learning_rate": 2.022043238660449e-07, + "loss": 0.4663, + "mean_token_accuracy": 0.8476271629333496, + "num_tokens": 18619703.0, + "step": 478 + }, + { + "epoch": 0.06093372344485434, + "grad_norm": 2.8948259353637695, + "learning_rate": 2.026282323018228e-07, + "loss": 0.452, + "mean_token_accuracy": 0.8531346321105957, + "num_tokens": 18656562.0, + "step": 479 + }, + { + "epoch": 0.06106093372344486, + "grad_norm": 2.652796745300293, + "learning_rate": 2.0305214073760065e-07, + "loss": 0.4431, + "mean_token_accuracy": 0.855027437210083, + "num_tokens": 18699388.0, + "step": 480 + }, + { + "epoch": 0.061188144002035365, + "grad_norm": 3.1541619300842285, + "learning_rate": 2.0347604917337855e-07, + "loss": 0.473, + "mean_token_accuracy": 0.8437752723693848, + "num_tokens": 18734431.0, + "step": 481 + }, + { + "epoch": 0.06131535428062587, + "grad_norm": 2.7541468143463135, + "learning_rate": 2.038999576091564e-07, + "loss": 0.4148, + "mean_token_accuracy": 0.8663238286972046, + "num_tokens": 18771553.0, + "step": 482 + }, + { + "epoch": 0.06144256455921639, + "grad_norm": 2.7138466835021973, + "learning_rate": 2.043238660449343e-07, + "loss": 0.4112, + "mean_token_accuracy": 0.8652635216712952, + "num_tokens": 18811070.0, + "step": 483 + }, + { + "epoch": 0.061569774837806894, + "grad_norm": 2.8288285732269287, + "learning_rate": 2.0474777448071214e-07, + "loss": 0.4114, + "mean_token_accuracy": 0.8637441396713257, + "num_tokens": 18845249.0, + "step": 484 + }, + { + "epoch": 0.0616969851163974, + "grad_norm": 2.895512342453003, + "learning_rate": 2.0517168291649004e-07, + "loss": 0.5059, + "mean_token_accuracy": 0.8381316661834717, + "num_tokens": 18885037.0, + "step": 485 + }, + { + "epoch": 0.061824195394987916, + "grad_norm": 2.8456220626831055, + "learning_rate": 2.0559559135226788e-07, + "loss": 0.4403, + "mean_token_accuracy": 0.8543533086776733, + "num_tokens": 18921518.0, + "step": 486 + }, + { + "epoch": 0.06195140567357842, + "grad_norm": 2.7323052883148193, + "learning_rate": 2.0601949978804578e-07, + "loss": 0.4663, + "mean_token_accuracy": 0.8457828164100647, + "num_tokens": 18965168.0, + "step": 487 + }, + { + "epoch": 0.06207861595216894, + "grad_norm": 2.8460710048675537, + "learning_rate": 2.0644340822382363e-07, + "loss": 0.4839, + "mean_token_accuracy": 0.83931565284729, + "num_tokens": 19002433.0, + "step": 488 + }, + { + "epoch": 0.062205826230759445, + "grad_norm": 3.0122432708740234, + "learning_rate": 2.0686731665960153e-07, + "loss": 0.489, + "mean_token_accuracy": 0.8396191596984863, + "num_tokens": 19038336.0, + "step": 489 + }, + { + "epoch": 0.06233303650934995, + "grad_norm": 2.8891842365264893, + "learning_rate": 2.0729122509537937e-07, + "loss": 0.4297, + "mean_token_accuracy": 0.8569645881652832, + "num_tokens": 19074764.0, + "step": 490 + }, + { + "epoch": 0.06246024678794047, + "grad_norm": 2.9272031784057617, + "learning_rate": 2.0771513353115727e-07, + "loss": 0.4862, + "mean_token_accuracy": 0.8400980830192566, + "num_tokens": 19112261.0, + "step": 491 + }, + { + "epoch": 0.06258745706653097, + "grad_norm": 2.662693500518799, + "learning_rate": 2.0813904196693512e-07, + "loss": 0.4297, + "mean_token_accuracy": 0.8585199117660522, + "num_tokens": 19150455.0, + "step": 492 + }, + { + "epoch": 0.06271466734512149, + "grad_norm": 2.895585298538208, + "learning_rate": 2.0856295040271302e-07, + "loss": 0.4989, + "mean_token_accuracy": 0.8368469476699829, + "num_tokens": 19187788.0, + "step": 493 + }, + { + "epoch": 0.06284187762371199, + "grad_norm": 2.8317506313323975, + "learning_rate": 2.0898685883849086e-07, + "loss": 0.4234, + "mean_token_accuracy": 0.8582879304885864, + "num_tokens": 19225200.0, + "step": 494 + }, + { + "epoch": 0.0629690879023025, + "grad_norm": 3.099520206451416, + "learning_rate": 2.0941076727426874e-07, + "loss": 0.4868, + "mean_token_accuracy": 0.8426961898803711, + "num_tokens": 19260406.0, + "step": 495 + }, + { + "epoch": 0.06309629818089302, + "grad_norm": 2.7343010902404785, + "learning_rate": 2.098346757100466e-07, + "loss": 0.4411, + "mean_token_accuracy": 0.8528043031692505, + "num_tokens": 19298749.0, + "step": 496 + }, + { + "epoch": 0.06322350845948353, + "grad_norm": 2.6079673767089844, + "learning_rate": 2.1025858414582448e-07, + "loss": 0.4164, + "mean_token_accuracy": 0.8623991012573242, + "num_tokens": 19340677.0, + "step": 497 + }, + { + "epoch": 0.06335071873807403, + "grad_norm": 2.7460343837738037, + "learning_rate": 2.1068249258160238e-07, + "loss": 0.4486, + "mean_token_accuracy": 0.8532446026802063, + "num_tokens": 19382362.0, + "step": 498 + }, + { + "epoch": 0.06347792901666455, + "grad_norm": 2.614227056503296, + "learning_rate": 2.1110640101738023e-07, + "loss": 0.4633, + "mean_token_accuracy": 0.8465572595596313, + "num_tokens": 19426346.0, + "step": 499 + }, + { + "epoch": 0.06360513929525506, + "grad_norm": 2.530939817428589, + "learning_rate": 2.1153030945315813e-07, + "loss": 0.4565, + "mean_token_accuracy": 0.8477694392204285, + "num_tokens": 19471959.0, + "step": 500 + }, + { + "epoch": 0.06373234957384556, + "grad_norm": 3.0573911666870117, + "learning_rate": 2.1195421788893597e-07, + "loss": 0.4678, + "mean_token_accuracy": 0.8455621004104614, + "num_tokens": 19508500.0, + "step": 501 + }, + { + "epoch": 0.06385955985243608, + "grad_norm": 2.9048142433166504, + "learning_rate": 2.1237812632471387e-07, + "loss": 0.4003, + "mean_token_accuracy": 0.8645986318588257, + "num_tokens": 19545025.0, + "step": 502 + }, + { + "epoch": 0.06398677013102659, + "grad_norm": 2.544025182723999, + "learning_rate": 2.1280203476049172e-07, + "loss": 0.4454, + "mean_token_accuracy": 0.8544672727584839, + "num_tokens": 19588379.0, + "step": 503 + }, + { + "epoch": 0.06411398040961709, + "grad_norm": 2.5840067863464355, + "learning_rate": 2.1322594319626962e-07, + "loss": 0.4032, + "mean_token_accuracy": 0.8696613907814026, + "num_tokens": 19630327.0, + "step": 504 + }, + { + "epoch": 0.0642411906882076, + "grad_norm": 2.7231063842773438, + "learning_rate": 2.1364985163204746e-07, + "loss": 0.3993, + "mean_token_accuracy": 0.8687847852706909, + "num_tokens": 19669802.0, + "step": 505 + }, + { + "epoch": 0.06436840096679812, + "grad_norm": 3.0020995140075684, + "learning_rate": 2.1407376006782536e-07, + "loss": 0.4701, + "mean_token_accuracy": 0.8495665788650513, + "num_tokens": 19705430.0, + "step": 506 + }, + { + "epoch": 0.06449561124538863, + "grad_norm": 3.129161834716797, + "learning_rate": 2.144976685036032e-07, + "loss": 0.4948, + "mean_token_accuracy": 0.8414561748504639, + "num_tokens": 19746299.0, + "step": 507 + }, + { + "epoch": 0.06462282152397913, + "grad_norm": 3.0135834217071533, + "learning_rate": 2.149215769393811e-07, + "loss": 0.4947, + "mean_token_accuracy": 0.8377445340156555, + "num_tokens": 19783393.0, + "step": 508 + }, + { + "epoch": 0.06475003180256965, + "grad_norm": 2.9665040969848633, + "learning_rate": 2.1534548537515895e-07, + "loss": 0.414, + "mean_token_accuracy": 0.8606566190719604, + "num_tokens": 19819235.0, + "step": 509 + }, + { + "epoch": 0.06487724208116016, + "grad_norm": 3.0163090229034424, + "learning_rate": 2.1576939381093685e-07, + "loss": 0.46, + "mean_token_accuracy": 0.8515697717666626, + "num_tokens": 19854506.0, + "step": 510 + }, + { + "epoch": 0.06500445235975066, + "grad_norm": 2.946307897567749, + "learning_rate": 2.161933022467147e-07, + "loss": 0.5201, + "mean_token_accuracy": 0.8326457738876343, + "num_tokens": 19895675.0, + "step": 511 + }, + { + "epoch": 0.06513166263834118, + "grad_norm": 2.3598921298980713, + "learning_rate": 2.166172106824926e-07, + "loss": 0.433, + "mean_token_accuracy": 0.8564255237579346, + "num_tokens": 19945929.0, + "step": 512 + }, + { + "epoch": 0.06525887291693169, + "grad_norm": 2.9137935638427734, + "learning_rate": 2.1704111911827044e-07, + "loss": 0.4479, + "mean_token_accuracy": 0.8547487854957581, + "num_tokens": 19985372.0, + "step": 513 + }, + { + "epoch": 0.0653860831955222, + "grad_norm": 2.7163124084472656, + "learning_rate": 2.1746502755404831e-07, + "loss": 0.414, + "mean_token_accuracy": 0.8597089648246765, + "num_tokens": 20023896.0, + "step": 514 + }, + { + "epoch": 0.06551329347411271, + "grad_norm": 2.998279571533203, + "learning_rate": 2.178889359898262e-07, + "loss": 0.4477, + "mean_token_accuracy": 0.8496125936508179, + "num_tokens": 20060094.0, + "step": 515 + }, + { + "epoch": 0.06564050375270322, + "grad_norm": 2.732935667037964, + "learning_rate": 2.1831284442560406e-07, + "loss": 0.4441, + "mean_token_accuracy": 0.8562928438186646, + "num_tokens": 20098540.0, + "step": 516 + }, + { + "epoch": 0.06576771403129372, + "grad_norm": 3.011193037033081, + "learning_rate": 2.1873675286138193e-07, + "loss": 0.4426, + "mean_token_accuracy": 0.8536100387573242, + "num_tokens": 20133401.0, + "step": 517 + }, + { + "epoch": 0.06589492430988424, + "grad_norm": 2.565321445465088, + "learning_rate": 2.191606612971598e-07, + "loss": 0.4603, + "mean_token_accuracy": 0.849397599697113, + "num_tokens": 20179906.0, + "step": 518 + }, + { + "epoch": 0.06602213458847475, + "grad_norm": 2.9231135845184326, + "learning_rate": 2.1958456973293768e-07, + "loss": 0.4086, + "mean_token_accuracy": 0.8669577836990356, + "num_tokens": 20219469.0, + "step": 519 + }, + { + "epoch": 0.06614934486706527, + "grad_norm": 3.0525412559509277, + "learning_rate": 2.2000847816871555e-07, + "loss": 0.4488, + "mean_token_accuracy": 0.8543969988822937, + "num_tokens": 20256520.0, + "step": 520 + }, + { + "epoch": 0.06627655514565577, + "grad_norm": 3.0408449172973633, + "learning_rate": 2.2043238660449342e-07, + "loss": 0.4633, + "mean_token_accuracy": 0.8522530794143677, + "num_tokens": 20290271.0, + "step": 521 + }, + { + "epoch": 0.06640376542424628, + "grad_norm": 2.7235782146453857, + "learning_rate": 2.208562950402713e-07, + "loss": 0.4458, + "mean_token_accuracy": 0.8548436164855957, + "num_tokens": 20334776.0, + "step": 522 + }, + { + "epoch": 0.0665309757028368, + "grad_norm": 2.817035436630249, + "learning_rate": 2.2128020347604917e-07, + "loss": 0.4366, + "mean_token_accuracy": 0.8579902648925781, + "num_tokens": 20374454.0, + "step": 523 + }, + { + "epoch": 0.0666581859814273, + "grad_norm": 3.161566972732544, + "learning_rate": 2.2170411191182704e-07, + "loss": 0.5085, + "mean_token_accuracy": 0.8426688313484192, + "num_tokens": 20412244.0, + "step": 524 + }, + { + "epoch": 0.06678539626001781, + "grad_norm": 2.988189220428467, + "learning_rate": 2.221280203476049e-07, + "loss": 0.4634, + "mean_token_accuracy": 0.8462725877761841, + "num_tokens": 20449676.0, + "step": 525 + }, + { + "epoch": 0.06691260653860832, + "grad_norm": 2.6524434089660645, + "learning_rate": 2.2255192878338279e-07, + "loss": 0.4505, + "mean_token_accuracy": 0.852535605430603, + "num_tokens": 20490922.0, + "step": 526 + }, + { + "epoch": 0.06703981681719882, + "grad_norm": 2.622208595275879, + "learning_rate": 2.2297583721916066e-07, + "loss": 0.4203, + "mean_token_accuracy": 0.8615023493766785, + "num_tokens": 20532610.0, + "step": 527 + }, + { + "epoch": 0.06716702709578934, + "grad_norm": 2.6304595470428467, + "learning_rate": 2.2339974565493853e-07, + "loss": 0.4075, + "mean_token_accuracy": 0.8671175241470337, + "num_tokens": 20569526.0, + "step": 528 + }, + { + "epoch": 0.06729423737437985, + "grad_norm": 2.5685489177703857, + "learning_rate": 2.238236540907164e-07, + "loss": 0.4307, + "mean_token_accuracy": 0.8577561378479004, + "num_tokens": 20613478.0, + "step": 529 + }, + { + "epoch": 0.06742144765297035, + "grad_norm": 2.7586376667022705, + "learning_rate": 2.2424756252649428e-07, + "loss": 0.434, + "mean_token_accuracy": 0.855843186378479, + "num_tokens": 20653599.0, + "step": 530 + }, + { + "epoch": 0.06754865793156087, + "grad_norm": 2.851386308670044, + "learning_rate": 2.2467147096227215e-07, + "loss": 0.4481, + "mean_token_accuracy": 0.8491859436035156, + "num_tokens": 20689923.0, + "step": 531 + }, + { + "epoch": 0.06767586821015138, + "grad_norm": 2.8507416248321533, + "learning_rate": 2.2509537939805002e-07, + "loss": 0.412, + "mean_token_accuracy": 0.8629186153411865, + "num_tokens": 20725168.0, + "step": 532 + }, + { + "epoch": 0.0678030784887419, + "grad_norm": 2.7175917625427246, + "learning_rate": 2.2551928783382787e-07, + "loss": 0.4457, + "mean_token_accuracy": 0.8537073135375977, + "num_tokens": 20765865.0, + "step": 533 + }, + { + "epoch": 0.0679302887673324, + "grad_norm": 2.849252939224243, + "learning_rate": 2.2594319626960577e-07, + "loss": 0.4488, + "mean_token_accuracy": 0.8522699475288391, + "num_tokens": 20803989.0, + "step": 534 + }, + { + "epoch": 0.06805749904592291, + "grad_norm": 2.8596997261047363, + "learning_rate": 2.263671047053836e-07, + "loss": 0.4843, + "mean_token_accuracy": 0.8421770334243774, + "num_tokens": 20843339.0, + "step": 535 + }, + { + "epoch": 0.06818470932451343, + "grad_norm": 2.908940076828003, + "learning_rate": 2.267910131411615e-07, + "loss": 0.444, + "mean_token_accuracy": 0.8506212830543518, + "num_tokens": 20880790.0, + "step": 536 + }, + { + "epoch": 0.06831191960310393, + "grad_norm": 2.9677438735961914, + "learning_rate": 2.2721492157693936e-07, + "loss": 0.4539, + "mean_token_accuracy": 0.852723240852356, + "num_tokens": 20918493.0, + "step": 537 + }, + { + "epoch": 0.06843912988169444, + "grad_norm": 2.8824238777160645, + "learning_rate": 2.2763883001271726e-07, + "loss": 0.4505, + "mean_token_accuracy": 0.8516265153884888, + "num_tokens": 20957439.0, + "step": 538 + }, + { + "epoch": 0.06856634016028496, + "grad_norm": 2.6589515209198, + "learning_rate": 2.280627384484951e-07, + "loss": 0.429, + "mean_token_accuracy": 0.8564094305038452, + "num_tokens": 20998435.0, + "step": 539 + }, + { + "epoch": 0.06869355043887546, + "grad_norm": 2.5871105194091797, + "learning_rate": 2.28486646884273e-07, + "loss": 0.4247, + "mean_token_accuracy": 0.8583810925483704, + "num_tokens": 21041816.0, + "step": 540 + }, + { + "epoch": 0.06882076071746597, + "grad_norm": 2.852968215942383, + "learning_rate": 2.2891055532005085e-07, + "loss": 0.4137, + "mean_token_accuracy": 0.859168529510498, + "num_tokens": 21077258.0, + "step": 541 + }, + { + "epoch": 0.06894797099605648, + "grad_norm": 2.720400333404541, + "learning_rate": 2.2933446375582875e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8630061149597168, + "num_tokens": 21116307.0, + "step": 542 + }, + { + "epoch": 0.06907518127464699, + "grad_norm": 3.2675089836120605, + "learning_rate": 2.297583721916066e-07, + "loss": 0.4796, + "mean_token_accuracy": 0.8440357446670532, + "num_tokens": 21149992.0, + "step": 543 + }, + { + "epoch": 0.0692023915532375, + "grad_norm": 2.7286887168884277, + "learning_rate": 2.301822806273845e-07, + "loss": 0.4592, + "mean_token_accuracy": 0.8459148406982422, + "num_tokens": 21191300.0, + "step": 544 + }, + { + "epoch": 0.06932960183182801, + "grad_norm": 3.2074403762817383, + "learning_rate": 2.3060618906316234e-07, + "loss": 0.4906, + "mean_token_accuracy": 0.8373256325721741, + "num_tokens": 21226901.0, + "step": 545 + }, + { + "epoch": 0.06945681211041853, + "grad_norm": 3.177316427230835, + "learning_rate": 2.3103009749894024e-07, + "loss": 0.4344, + "mean_token_accuracy": 0.857670247554779, + "num_tokens": 21260394.0, + "step": 546 + }, + { + "epoch": 0.06958402238900903, + "grad_norm": 3.1402814388275146, + "learning_rate": 2.3145400593471808e-07, + "loss": 0.5336, + "mean_token_accuracy": 0.8268373012542725, + "num_tokens": 21298102.0, + "step": 547 + }, + { + "epoch": 0.06971123266759954, + "grad_norm": 3.1266980171203613, + "learning_rate": 2.3187791437049598e-07, + "loss": 0.4874, + "mean_token_accuracy": 0.8413441777229309, + "num_tokens": 21334874.0, + "step": 548 + }, + { + "epoch": 0.06983844294619006, + "grad_norm": 3.1702513694763184, + "learning_rate": 2.3230182280627383e-07, + "loss": 0.4888, + "mean_token_accuracy": 0.8403713703155518, + "num_tokens": 21372429.0, + "step": 549 + }, + { + "epoch": 0.06996565322478056, + "grad_norm": 2.7317864894866943, + "learning_rate": 2.327257312420517e-07, + "loss": 0.4024, + "mean_token_accuracy": 0.8641936779022217, + "num_tokens": 21408679.0, + "step": 550 + }, + { + "epoch": 0.07009286350337107, + "grad_norm": 2.7416892051696777, + "learning_rate": 2.3314963967782957e-07, + "loss": 0.4599, + "mean_token_accuracy": 0.8478851318359375, + "num_tokens": 21451894.0, + "step": 551 + }, + { + "epoch": 0.07022007378196159, + "grad_norm": 3.2256486415863037, + "learning_rate": 2.3357354811360745e-07, + "loss": 0.4495, + "mean_token_accuracy": 0.8481927514076233, + "num_tokens": 21483790.0, + "step": 552 + }, + { + "epoch": 0.07034728406055209, + "grad_norm": 2.802316188812256, + "learning_rate": 2.3399745654938532e-07, + "loss": 0.4556, + "mean_token_accuracy": 0.8515995740890503, + "num_tokens": 21528030.0, + "step": 553 + }, + { + "epoch": 0.0704744943391426, + "grad_norm": 3.157693862915039, + "learning_rate": 2.344213649851632e-07, + "loss": 0.4703, + "mean_token_accuracy": 0.8459073305130005, + "num_tokens": 21559989.0, + "step": 554 + }, + { + "epoch": 0.07060170461773312, + "grad_norm": 2.926149845123291, + "learning_rate": 2.3484527342094106e-07, + "loss": 0.4397, + "mean_token_accuracy": 0.8545874357223511, + "num_tokens": 21600916.0, + "step": 555 + }, + { + "epoch": 0.07072891489632362, + "grad_norm": 3.0131022930145264, + "learning_rate": 2.3526918185671894e-07, + "loss": 0.4581, + "mean_token_accuracy": 0.8495944142341614, + "num_tokens": 21636418.0, + "step": 556 + }, + { + "epoch": 0.07085612517491413, + "grad_norm": 2.505884885787964, + "learning_rate": 2.356930902924968e-07, + "loss": 0.4102, + "mean_token_accuracy": 0.8640081286430359, + "num_tokens": 21679814.0, + "step": 557 + }, + { + "epoch": 0.07098333545350465, + "grad_norm": 3.067599296569824, + "learning_rate": 2.3611699872827468e-07, + "loss": 0.4485, + "mean_token_accuracy": 0.8564081192016602, + "num_tokens": 21712491.0, + "step": 558 + }, + { + "epoch": 0.07111054573209516, + "grad_norm": 2.8843672275543213, + "learning_rate": 2.3654090716405255e-07, + "loss": 0.4261, + "mean_token_accuracy": 0.8617040514945984, + "num_tokens": 21750129.0, + "step": 559 + }, + { + "epoch": 0.07123775601068566, + "grad_norm": 2.7740437984466553, + "learning_rate": 2.3696481559983043e-07, + "loss": 0.4391, + "mean_token_accuracy": 0.8555909991264343, + "num_tokens": 21787939.0, + "step": 560 + }, + { + "epoch": 0.07136496628927617, + "grad_norm": 2.89886474609375, + "learning_rate": 2.373887240356083e-07, + "loss": 0.4291, + "mean_token_accuracy": 0.8555764555931091, + "num_tokens": 21823383.0, + "step": 561 + }, + { + "epoch": 0.07149217656786669, + "grad_norm": 2.952481985092163, + "learning_rate": 2.3781263247138617e-07, + "loss": 0.442, + "mean_token_accuracy": 0.8588563799858093, + "num_tokens": 21860774.0, + "step": 562 + }, + { + "epoch": 0.07161938684645719, + "grad_norm": 3.470534324645996, + "learning_rate": 2.3823654090716404e-07, + "loss": 0.5299, + "mean_token_accuracy": 0.8254684209823608, + "num_tokens": 21892139.0, + "step": 563 + }, + { + "epoch": 0.0717465971250477, + "grad_norm": 2.9277701377868652, + "learning_rate": 2.386604493429419e-07, + "loss": 0.5085, + "mean_token_accuracy": 0.8320180177688599, + "num_tokens": 21932471.0, + "step": 564 + }, + { + "epoch": 0.07187380740363822, + "grad_norm": 2.6870386600494385, + "learning_rate": 2.390843577787198e-07, + "loss": 0.3998, + "mean_token_accuracy": 0.8646465539932251, + "num_tokens": 21969786.0, + "step": 565 + }, + { + "epoch": 0.07200101768222872, + "grad_norm": 2.5277304649353027, + "learning_rate": 2.3950826621449766e-07, + "loss": 0.387, + "mean_token_accuracy": 0.868983805179596, + "num_tokens": 22011329.0, + "step": 566 + }, + { + "epoch": 0.07212822796081923, + "grad_norm": 2.378563165664673, + "learning_rate": 2.3993217465027556e-07, + "loss": 0.3842, + "mean_token_accuracy": 0.8725093603134155, + "num_tokens": 22056071.0, + "step": 567 + }, + { + "epoch": 0.07225543823940975, + "grad_norm": 2.968219757080078, + "learning_rate": 2.403560830860534e-07, + "loss": 0.4422, + "mean_token_accuracy": 0.8541086316108704, + "num_tokens": 22092351.0, + "step": 568 + }, + { + "epoch": 0.07238264851800025, + "grad_norm": 2.932393789291382, + "learning_rate": 2.4077999152183125e-07, + "loss": 0.4106, + "mean_token_accuracy": 0.861508309841156, + "num_tokens": 22128143.0, + "step": 569 + }, + { + "epoch": 0.07250985879659076, + "grad_norm": 2.718273162841797, + "learning_rate": 2.4120389995760915e-07, + "loss": 0.465, + "mean_token_accuracy": 0.8507491946220398, + "num_tokens": 22169557.0, + "step": 570 + }, + { + "epoch": 0.07263706907518128, + "grad_norm": 2.809039831161499, + "learning_rate": 2.41627808393387e-07, + "loss": 0.4422, + "mean_token_accuracy": 0.8547127842903137, + "num_tokens": 22207953.0, + "step": 571 + }, + { + "epoch": 0.07276427935377179, + "grad_norm": 2.7563416957855225, + "learning_rate": 2.420517168291649e-07, + "loss": 0.3996, + "mean_token_accuracy": 0.8675310611724854, + "num_tokens": 22246592.0, + "step": 572 + }, + { + "epoch": 0.07289148963236229, + "grad_norm": 2.580327272415161, + "learning_rate": 2.4247562526494274e-07, + "loss": 0.4352, + "mean_token_accuracy": 0.8573297262191772, + "num_tokens": 22288250.0, + "step": 573 + }, + { + "epoch": 0.0730186999109528, + "grad_norm": 2.6500394344329834, + "learning_rate": 2.4289953370072064e-07, + "loss": 0.3846, + "mean_token_accuracy": 0.8674805164337158, + "num_tokens": 22325431.0, + "step": 574 + }, + { + "epoch": 0.07314591018954332, + "grad_norm": 2.6600027084350586, + "learning_rate": 2.433234421364985e-07, + "loss": 0.4178, + "mean_token_accuracy": 0.8599252700805664, + "num_tokens": 22365626.0, + "step": 575 + }, + { + "epoch": 0.07327312046813382, + "grad_norm": 2.6949195861816406, + "learning_rate": 2.437473505722764e-07, + "loss": 0.4133, + "mean_token_accuracy": 0.8624798059463501, + "num_tokens": 22406010.0, + "step": 576 + }, + { + "epoch": 0.07340033074672433, + "grad_norm": 2.702777624130249, + "learning_rate": 2.4417125900805423e-07, + "loss": 0.4426, + "mean_token_accuracy": 0.855394184589386, + "num_tokens": 22445966.0, + "step": 577 + }, + { + "epoch": 0.07352754102531485, + "grad_norm": 2.91587495803833, + "learning_rate": 2.4459516744383213e-07, + "loss": 0.4733, + "mean_token_accuracy": 0.8409011960029602, + "num_tokens": 22486181.0, + "step": 578 + }, + { + "epoch": 0.07365475130390535, + "grad_norm": 2.763625383377075, + "learning_rate": 2.4501907587961e-07, + "loss": 0.4875, + "mean_token_accuracy": 0.8441387414932251, + "num_tokens": 22527205.0, + "step": 579 + }, + { + "epoch": 0.07378196158249586, + "grad_norm": 2.844721555709839, + "learning_rate": 2.454429843153879e-07, + "loss": 0.4744, + "mean_token_accuracy": 0.842616856098175, + "num_tokens": 22569345.0, + "step": 580 + }, + { + "epoch": 0.07390917186108638, + "grad_norm": 2.671900987625122, + "learning_rate": 2.458668927511657e-07, + "loss": 0.4125, + "mean_token_accuracy": 0.864399790763855, + "num_tokens": 22611772.0, + "step": 581 + }, + { + "epoch": 0.07403638213967688, + "grad_norm": 2.9769396781921387, + "learning_rate": 2.462908011869436e-07, + "loss": 0.4249, + "mean_token_accuracy": 0.8583711385726929, + "num_tokens": 22647444.0, + "step": 582 + }, + { + "epoch": 0.0741635924182674, + "grad_norm": 2.779841184616089, + "learning_rate": 2.4671470962272147e-07, + "loss": 0.42, + "mean_token_accuracy": 0.8631848692893982, + "num_tokens": 22686790.0, + "step": 583 + }, + { + "epoch": 0.07429080269685791, + "grad_norm": 2.9031691551208496, + "learning_rate": 2.4713861805849937e-07, + "loss": 0.4656, + "mean_token_accuracy": 0.8438650369644165, + "num_tokens": 22727175.0, + "step": 584 + }, + { + "epoch": 0.07441801297544842, + "grad_norm": 3.15429425239563, + "learning_rate": 2.475625264942772e-07, + "loss": 0.4575, + "mean_token_accuracy": 0.8505262136459351, + "num_tokens": 22760159.0, + "step": 585 + }, + { + "epoch": 0.07454522325403892, + "grad_norm": 2.633694648742676, + "learning_rate": 2.479864349300551e-07, + "loss": 0.3975, + "mean_token_accuracy": 0.8697764873504639, + "num_tokens": 22800390.0, + "step": 586 + }, + { + "epoch": 0.07467243353262944, + "grad_norm": 2.7171456813812256, + "learning_rate": 2.4841034336583296e-07, + "loss": 0.4247, + "mean_token_accuracy": 0.8600481748580933, + "num_tokens": 22841434.0, + "step": 587 + }, + { + "epoch": 0.07479964381121995, + "grad_norm": 2.8326005935668945, + "learning_rate": 2.488342518016108e-07, + "loss": 0.4076, + "mean_token_accuracy": 0.8629134893417358, + "num_tokens": 22875787.0, + "step": 588 + }, + { + "epoch": 0.07492685408981045, + "grad_norm": 2.797940969467163, + "learning_rate": 2.492581602373887e-07, + "loss": 0.4722, + "mean_token_accuracy": 0.8481526970863342, + "num_tokens": 22918099.0, + "step": 589 + }, + { + "epoch": 0.07505406436840097, + "grad_norm": 3.1908178329467773, + "learning_rate": 2.4968206867316655e-07, + "loss": 0.4465, + "mean_token_accuracy": 0.855919599533081, + "num_tokens": 22951955.0, + "step": 590 + }, + { + "epoch": 0.07518127464699148, + "grad_norm": 3.1405506134033203, + "learning_rate": 2.5010597710894445e-07, + "loss": 0.437, + "mean_token_accuracy": 0.8539055585861206, + "num_tokens": 22983520.0, + "step": 591 + }, + { + "epoch": 0.07530848492558198, + "grad_norm": 2.988196611404419, + "learning_rate": 2.505298855447223e-07, + "loss": 0.428, + "mean_token_accuracy": 0.8552442789077759, + "num_tokens": 23017075.0, + "step": 592 + }, + { + "epoch": 0.0754356952041725, + "grad_norm": 3.1062710285186768, + "learning_rate": 2.509537939805002e-07, + "loss": 0.465, + "mean_token_accuracy": 0.8463068008422852, + "num_tokens": 23054803.0, + "step": 593 + }, + { + "epoch": 0.07556290548276301, + "grad_norm": 2.741312265396118, + "learning_rate": 2.513777024162781e-07, + "loss": 0.4727, + "mean_token_accuracy": 0.8448078632354736, + "num_tokens": 23097632.0, + "step": 594 + }, + { + "epoch": 0.07569011576135352, + "grad_norm": 2.7659826278686523, + "learning_rate": 2.5180161085205594e-07, + "loss": 0.3922, + "mean_token_accuracy": 0.8680919408798218, + "num_tokens": 23134737.0, + "step": 595 + }, + { + "epoch": 0.07581732603994402, + "grad_norm": 2.885028123855591, + "learning_rate": 2.522255192878338e-07, + "loss": 0.4573, + "mean_token_accuracy": 0.8485636711120605, + "num_tokens": 23172738.0, + "step": 596 + }, + { + "epoch": 0.07594453631853454, + "grad_norm": 2.911669969558716, + "learning_rate": 2.526494277236117e-07, + "loss": 0.4637, + "mean_token_accuracy": 0.8501964211463928, + "num_tokens": 23212775.0, + "step": 597 + }, + { + "epoch": 0.07607174659712505, + "grad_norm": 2.863741397857666, + "learning_rate": 2.530733361593896e-07, + "loss": 0.4283, + "mean_token_accuracy": 0.8587530851364136, + "num_tokens": 23250577.0, + "step": 598 + }, + { + "epoch": 0.07619895687571555, + "grad_norm": 2.4292802810668945, + "learning_rate": 2.5349724459516743e-07, + "loss": 0.4103, + "mean_token_accuracy": 0.8672951459884644, + "num_tokens": 23299252.0, + "step": 599 + }, + { + "epoch": 0.07632616715430607, + "grad_norm": 2.611942768096924, + "learning_rate": 2.539211530309453e-07, + "loss": 0.4049, + "mean_token_accuracy": 0.8649991750717163, + "num_tokens": 23340700.0, + "step": 600 + }, + { + "epoch": 0.07645337743289658, + "grad_norm": 2.7649614810943604, + "learning_rate": 2.543450614667232e-07, + "loss": 0.4555, + "mean_token_accuracy": 0.8552904725074768, + "num_tokens": 23381206.0, + "step": 601 + }, + { + "epoch": 0.07658058771148708, + "grad_norm": 2.6124210357666016, + "learning_rate": 2.547689699025011e-07, + "loss": 0.4251, + "mean_token_accuracy": 0.8594434261322021, + "num_tokens": 23424261.0, + "step": 602 + }, + { + "epoch": 0.0767077979900776, + "grad_norm": 2.898576021194458, + "learning_rate": 2.551928783382789e-07, + "loss": 0.4636, + "mean_token_accuracy": 0.8491230607032776, + "num_tokens": 23464568.0, + "step": 603 + }, + { + "epoch": 0.07683500826866811, + "grad_norm": 3.0307059288024902, + "learning_rate": 2.5561678677405677e-07, + "loss": 0.4364, + "mean_token_accuracy": 0.8546459674835205, + "num_tokens": 23497533.0, + "step": 604 + }, + { + "epoch": 0.07696221854725861, + "grad_norm": 2.972656011581421, + "learning_rate": 2.5604069520983467e-07, + "loss": 0.4088, + "mean_token_accuracy": 0.864466667175293, + "num_tokens": 23532072.0, + "step": 605 + }, + { + "epoch": 0.07708942882584913, + "grad_norm": 2.6428189277648926, + "learning_rate": 2.564646036456125e-07, + "loss": 0.4052, + "mean_token_accuracy": 0.8660961985588074, + "num_tokens": 23573063.0, + "step": 606 + }, + { + "epoch": 0.07721663910443964, + "grad_norm": 3.046562910079956, + "learning_rate": 2.568885120813904e-07, + "loss": 0.475, + "mean_token_accuracy": 0.8452872037887573, + "num_tokens": 23611183.0, + "step": 607 + }, + { + "epoch": 0.07734384938303016, + "grad_norm": 3.1095080375671387, + "learning_rate": 2.5731242051716826e-07, + "loss": 0.457, + "mean_token_accuracy": 0.8490122556686401, + "num_tokens": 23649691.0, + "step": 608 + }, + { + "epoch": 0.07747105966162066, + "grad_norm": 2.930574655532837, + "learning_rate": 2.5773632895294616e-07, + "loss": 0.4551, + "mean_token_accuracy": 0.8496590852737427, + "num_tokens": 23688861.0, + "step": 609 + }, + { + "epoch": 0.07759826994021117, + "grad_norm": 2.6070072650909424, + "learning_rate": 2.58160237388724e-07, + "loss": 0.4221, + "mean_token_accuracy": 0.8595035076141357, + "num_tokens": 23730504.0, + "step": 610 + }, + { + "epoch": 0.07772548021880168, + "grad_norm": 2.9671378135681152, + "learning_rate": 2.585841458245019e-07, + "loss": 0.4381, + "mean_token_accuracy": 0.8576528429985046, + "num_tokens": 23766488.0, + "step": 611 + }, + { + "epoch": 0.07785269049739219, + "grad_norm": 3.058469295501709, + "learning_rate": 2.5900805426027975e-07, + "loss": 0.4403, + "mean_token_accuracy": 0.8545424938201904, + "num_tokens": 23804726.0, + "step": 612 + }, + { + "epoch": 0.0779799007759827, + "grad_norm": 2.7957584857940674, + "learning_rate": 2.5943196269605765e-07, + "loss": 0.464, + "mean_token_accuracy": 0.8480116128921509, + "num_tokens": 23844152.0, + "step": 613 + }, + { + "epoch": 0.07810711105457321, + "grad_norm": 2.99780011177063, + "learning_rate": 2.598558711318355e-07, + "loss": 0.4186, + "mean_token_accuracy": 0.8605356216430664, + "num_tokens": 23879988.0, + "step": 614 + }, + { + "epoch": 0.07823432133316371, + "grad_norm": 2.806809663772583, + "learning_rate": 2.602797795676134e-07, + "loss": 0.4438, + "mean_token_accuracy": 0.8557107448577881, + "num_tokens": 23919613.0, + "step": 615 + }, + { + "epoch": 0.07836153161175423, + "grad_norm": 2.8081488609313965, + "learning_rate": 2.6070368800339124e-07, + "loss": 0.494, + "mean_token_accuracy": 0.8354904651641846, + "num_tokens": 23961801.0, + "step": 616 + }, + { + "epoch": 0.07848874189034474, + "grad_norm": 2.862614393234253, + "learning_rate": 2.6112759643916914e-07, + "loss": 0.4235, + "mean_token_accuracy": 0.8585066795349121, + "num_tokens": 23998474.0, + "step": 617 + }, + { + "epoch": 0.07861595216893524, + "grad_norm": 3.256460189819336, + "learning_rate": 2.61551504874947e-07, + "loss": 0.5204, + "mean_token_accuracy": 0.8314387798309326, + "num_tokens": 24034583.0, + "step": 618 + }, + { + "epoch": 0.07874316244752576, + "grad_norm": 2.8851752281188965, + "learning_rate": 2.619754133107249e-07, + "loss": 0.4334, + "mean_token_accuracy": 0.857316255569458, + "num_tokens": 24074010.0, + "step": 619 + }, + { + "epoch": 0.07887037272611627, + "grad_norm": 3.0846426486968994, + "learning_rate": 2.623993217465028e-07, + "loss": 0.5137, + "mean_token_accuracy": 0.8370674252510071, + "num_tokens": 24111596.0, + "step": 620 + }, + { + "epoch": 0.07899758300470679, + "grad_norm": 3.0438437461853027, + "learning_rate": 2.6282323018228063e-07, + "loss": 0.4503, + "mean_token_accuracy": 0.8559277653694153, + "num_tokens": 24150835.0, + "step": 621 + }, + { + "epoch": 0.07912479328329729, + "grad_norm": 2.817674160003662, + "learning_rate": 2.632471386180585e-07, + "loss": 0.4452, + "mean_token_accuracy": 0.8528842329978943, + "num_tokens": 24188304.0, + "step": 622 + }, + { + "epoch": 0.0792520035618878, + "grad_norm": 2.5874197483062744, + "learning_rate": 2.6367104705383637e-07, + "loss": 0.3794, + "mean_token_accuracy": 0.8723642230033875, + "num_tokens": 24227857.0, + "step": 623 + }, + { + "epoch": 0.07937921384047832, + "grad_norm": 3.004931926727295, + "learning_rate": 2.6409495548961427e-07, + "loss": 0.4609, + "mean_token_accuracy": 0.8482992649078369, + "num_tokens": 24263928.0, + "step": 624 + }, + { + "epoch": 0.07950642411906882, + "grad_norm": 2.6462419033050537, + "learning_rate": 2.6451886392539206e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8691921830177307, + "num_tokens": 24306738.0, + "step": 625 + }, + { + "epoch": 0.07963363439765933, + "grad_norm": 2.680030107498169, + "learning_rate": 2.6494277236116996e-07, + "loss": 0.4831, + "mean_token_accuracy": 0.8398833274841309, + "num_tokens": 24350910.0, + "step": 626 + }, + { + "epoch": 0.07976084467624985, + "grad_norm": 2.7628772258758545, + "learning_rate": 2.6536668079694786e-07, + "loss": 0.4384, + "mean_token_accuracy": 0.8575575947761536, + "num_tokens": 24390187.0, + "step": 627 + }, + { + "epoch": 0.07988805495484035, + "grad_norm": 2.7575135231018066, + "learning_rate": 2.6579058923272576e-07, + "loss": 0.4815, + "mean_token_accuracy": 0.8433607220649719, + "num_tokens": 24435454.0, + "step": 628 + }, + { + "epoch": 0.08001526523343086, + "grad_norm": 2.7624661922454834, + "learning_rate": 2.6621449766850356e-07, + "loss": 0.4401, + "mean_token_accuracy": 0.8522270321846008, + "num_tokens": 24476299.0, + "step": 629 + }, + { + "epoch": 0.08014247551202137, + "grad_norm": 2.96089243888855, + "learning_rate": 2.6663840610428145e-07, + "loss": 0.468, + "mean_token_accuracy": 0.8453305959701538, + "num_tokens": 24512512.0, + "step": 630 + }, + { + "epoch": 0.08026968579061187, + "grad_norm": 2.7450108528137207, + "learning_rate": 2.6706231454005935e-07, + "loss": 0.4354, + "mean_token_accuracy": 0.8569403886795044, + "num_tokens": 24552518.0, + "step": 631 + }, + { + "epoch": 0.08039689606920239, + "grad_norm": 2.9792370796203613, + "learning_rate": 2.6748622297583725e-07, + "loss": 0.4978, + "mean_token_accuracy": 0.8353633880615234, + "num_tokens": 24588414.0, + "step": 632 + }, + { + "epoch": 0.0805241063477929, + "grad_norm": 2.909437894821167, + "learning_rate": 2.6791013141161505e-07, + "loss": 0.4463, + "mean_token_accuracy": 0.8549051284790039, + "num_tokens": 24628231.0, + "step": 633 + }, + { + "epoch": 0.08065131662638342, + "grad_norm": 3.022686243057251, + "learning_rate": 2.6833403984739294e-07, + "loss": 0.4744, + "mean_token_accuracy": 0.8461493849754333, + "num_tokens": 24665461.0, + "step": 634 + }, + { + "epoch": 0.08077852690497392, + "grad_norm": 3.099097728729248, + "learning_rate": 2.6875794828317084e-07, + "loss": 0.4469, + "mean_token_accuracy": 0.8534034490585327, + "num_tokens": 24702619.0, + "step": 635 + }, + { + "epoch": 0.08090573718356443, + "grad_norm": 2.6700706481933594, + "learning_rate": 2.6918185671894874e-07, + "loss": 0.3935, + "mean_token_accuracy": 0.8677027225494385, + "num_tokens": 24743809.0, + "step": 636 + }, + { + "epoch": 0.08103294746215495, + "grad_norm": 2.7292661666870117, + "learning_rate": 2.6960576515472654e-07, + "loss": 0.4539, + "mean_token_accuracy": 0.8486394286155701, + "num_tokens": 24786299.0, + "step": 637 + }, + { + "epoch": 0.08116015774074545, + "grad_norm": 3.0367324352264404, + "learning_rate": 2.7002967359050443e-07, + "loss": 0.4473, + "mean_token_accuracy": 0.8571505546569824, + "num_tokens": 24823999.0, + "step": 638 + }, + { + "epoch": 0.08128736801933596, + "grad_norm": 2.8190248012542725, + "learning_rate": 2.7045358202628233e-07, + "loss": 0.476, + "mean_token_accuracy": 0.8443806767463684, + "num_tokens": 24864652.0, + "step": 639 + }, + { + "epoch": 0.08141457829792648, + "grad_norm": 3.0818135738372803, + "learning_rate": 2.7087749046206023e-07, + "loss": 0.4566, + "mean_token_accuracy": 0.844626784324646, + "num_tokens": 24900954.0, + "step": 640 + }, + { + "epoch": 0.08154178857651698, + "grad_norm": 3.0326757431030273, + "learning_rate": 2.71301398897838e-07, + "loss": 0.4283, + "mean_token_accuracy": 0.8557688593864441, + "num_tokens": 24933748.0, + "step": 641 + }, + { + "epoch": 0.08166899885510749, + "grad_norm": 2.760464668273926, + "learning_rate": 2.717253073336159e-07, + "loss": 0.4463, + "mean_token_accuracy": 0.8543701171875, + "num_tokens": 24972454.0, + "step": 642 + }, + { + "epoch": 0.081796209133698, + "grad_norm": 3.1963796615600586, + "learning_rate": 2.721492157693938e-07, + "loss": 0.4862, + "mean_token_accuracy": 0.8401336073875427, + "num_tokens": 25008212.0, + "step": 643 + }, + { + "epoch": 0.0819234194122885, + "grad_norm": 2.834345817565918, + "learning_rate": 2.7257312420517167e-07, + "loss": 0.4807, + "mean_token_accuracy": 0.8415329456329346, + "num_tokens": 25049265.0, + "step": 644 + }, + { + "epoch": 0.08205062969087902, + "grad_norm": 2.6343894004821777, + "learning_rate": 2.729970326409495e-07, + "loss": 0.3984, + "mean_token_accuracy": 0.8669523000717163, + "num_tokens": 25091024.0, + "step": 645 + }, + { + "epoch": 0.08217783996946953, + "grad_norm": 3.0728249549865723, + "learning_rate": 2.734209410767274e-07, + "loss": 0.4671, + "mean_token_accuracy": 0.8445784449577332, + "num_tokens": 25126457.0, + "step": 646 + }, + { + "epoch": 0.08230505024806005, + "grad_norm": 2.922171115875244, + "learning_rate": 2.738448495125053e-07, + "loss": 0.4161, + "mean_token_accuracy": 0.8598039150238037, + "num_tokens": 25164100.0, + "step": 647 + }, + { + "epoch": 0.08243226052665055, + "grad_norm": 3.0117502212524414, + "learning_rate": 2.7426875794828316e-07, + "loss": 0.4261, + "mean_token_accuracy": 0.8570291996002197, + "num_tokens": 25200795.0, + "step": 648 + }, + { + "epoch": 0.08255947080524106, + "grad_norm": 2.680408239364624, + "learning_rate": 2.74692666384061e-07, + "loss": 0.4271, + "mean_token_accuracy": 0.856609046459198, + "num_tokens": 25242159.0, + "step": 649 + }, + { + "epoch": 0.08268668108383158, + "grad_norm": 3.188093423843384, + "learning_rate": 2.751165748198389e-07, + "loss": 0.4303, + "mean_token_accuracy": 0.8610950708389282, + "num_tokens": 25276029.0, + "step": 650 + }, + { + "epoch": 0.08281389136242208, + "grad_norm": 2.771190881729126, + "learning_rate": 2.755404832556168e-07, + "loss": 0.4881, + "mean_token_accuracy": 0.8381960391998291, + "num_tokens": 25324520.0, + "step": 651 + }, + { + "epoch": 0.0829411016410126, + "grad_norm": 2.602341413497925, + "learning_rate": 2.7596439169139465e-07, + "loss": 0.4025, + "mean_token_accuracy": 0.8656573295593262, + "num_tokens": 25369295.0, + "step": 652 + }, + { + "epoch": 0.08306831191960311, + "grad_norm": 3.273120641708374, + "learning_rate": 2.763883001271725e-07, + "loss": 0.409, + "mean_token_accuracy": 0.8641707897186279, + "num_tokens": 25400386.0, + "step": 653 + }, + { + "epoch": 0.08319552219819361, + "grad_norm": 2.6290202140808105, + "learning_rate": 2.768122085629504e-07, + "loss": 0.4617, + "mean_token_accuracy": 0.8461084961891174, + "num_tokens": 25444772.0, + "step": 654 + }, + { + "epoch": 0.08332273247678412, + "grad_norm": 2.9897360801696777, + "learning_rate": 2.772361169987283e-07, + "loss": 0.434, + "mean_token_accuracy": 0.8551359176635742, + "num_tokens": 25482680.0, + "step": 655 + }, + { + "epoch": 0.08344994275537464, + "grad_norm": 2.870671272277832, + "learning_rate": 2.7766002543450614e-07, + "loss": 0.4754, + "mean_token_accuracy": 0.842168927192688, + "num_tokens": 25521365.0, + "step": 656 + }, + { + "epoch": 0.08357715303396514, + "grad_norm": 2.6177961826324463, + "learning_rate": 2.78083933870284e-07, + "loss": 0.4148, + "mean_token_accuracy": 0.8574331998825073, + "num_tokens": 25562579.0, + "step": 657 + }, + { + "epoch": 0.08370436331255565, + "grad_norm": 2.9637467861175537, + "learning_rate": 2.785078423060619e-07, + "loss": 0.4246, + "mean_token_accuracy": 0.8598600029945374, + "num_tokens": 25597360.0, + "step": 658 + }, + { + "epoch": 0.08383157359114617, + "grad_norm": 2.8410804271698, + "learning_rate": 2.789317507418398e-07, + "loss": 0.4584, + "mean_token_accuracy": 0.8472881317138672, + "num_tokens": 25636282.0, + "step": 659 + }, + { + "epoch": 0.08395878386973668, + "grad_norm": 2.7645230293273926, + "learning_rate": 2.7935565917761763e-07, + "loss": 0.3998, + "mean_token_accuracy": 0.8688003420829773, + "num_tokens": 25672898.0, + "step": 660 + }, + { + "epoch": 0.08408599414832718, + "grad_norm": 2.8922252655029297, + "learning_rate": 2.797795676133955e-07, + "loss": 0.3838, + "mean_token_accuracy": 0.8717814683914185, + "num_tokens": 25711575.0, + "step": 661 + }, + { + "epoch": 0.0842132044269177, + "grad_norm": 2.9525394439697266, + "learning_rate": 2.802034760491734e-07, + "loss": 0.461, + "mean_token_accuracy": 0.8483638763427734, + "num_tokens": 25750129.0, + "step": 662 + }, + { + "epoch": 0.08434041470550821, + "grad_norm": 2.6568145751953125, + "learning_rate": 2.806273844849512e-07, + "loss": 0.4062, + "mean_token_accuracy": 0.8655024766921997, + "num_tokens": 25789133.0, + "step": 663 + }, + { + "epoch": 0.08446762498409871, + "grad_norm": 2.8373255729675293, + "learning_rate": 2.810512929207291e-07, + "loss": 0.4088, + "mean_token_accuracy": 0.8640605211257935, + "num_tokens": 25828229.0, + "step": 664 + }, + { + "epoch": 0.08459483526268922, + "grad_norm": 2.7642102241516113, + "learning_rate": 2.8147520135650697e-07, + "loss": 0.4085, + "mean_token_accuracy": 0.8634793758392334, + "num_tokens": 25865748.0, + "step": 665 + }, + { + "epoch": 0.08472204554127974, + "grad_norm": 3.0843007564544678, + "learning_rate": 2.8189910979228487e-07, + "loss": 0.4133, + "mean_token_accuracy": 0.8594844341278076, + "num_tokens": 25900778.0, + "step": 666 + }, + { + "epoch": 0.08484925581987024, + "grad_norm": 3.1080548763275146, + "learning_rate": 2.823230182280627e-07, + "loss": 0.4456, + "mean_token_accuracy": 0.8539789319038391, + "num_tokens": 25935430.0, + "step": 667 + }, + { + "epoch": 0.08497646609846075, + "grad_norm": 3.0661556720733643, + "learning_rate": 2.827469266638406e-07, + "loss": 0.398, + "mean_token_accuracy": 0.8665187358856201, + "num_tokens": 25970521.0, + "step": 668 + }, + { + "epoch": 0.08510367637705127, + "grad_norm": 2.710716485977173, + "learning_rate": 2.8317083509961846e-07, + "loss": 0.4094, + "mean_token_accuracy": 0.8648982048034668, + "num_tokens": 26010203.0, + "step": 669 + }, + { + "epoch": 0.08523088665564178, + "grad_norm": 2.6834399700164795, + "learning_rate": 2.8359474353539636e-07, + "loss": 0.4049, + "mean_token_accuracy": 0.8645484447479248, + "num_tokens": 26051965.0, + "step": 670 + }, + { + "epoch": 0.08535809693423228, + "grad_norm": 3.1909754276275635, + "learning_rate": 2.840186519711742e-07, + "loss": 0.4502, + "mean_token_accuracy": 0.8481402397155762, + "num_tokens": 26085642.0, + "step": 671 + }, + { + "epoch": 0.0854853072128228, + "grad_norm": 2.9927375316619873, + "learning_rate": 2.844425604069521e-07, + "loss": 0.4755, + "mean_token_accuracy": 0.8412091732025146, + "num_tokens": 26127338.0, + "step": 672 + }, + { + "epoch": 0.08561251749141331, + "grad_norm": 3.0029261112213135, + "learning_rate": 2.8486646884272995e-07, + "loss": 0.4485, + "mean_token_accuracy": 0.8550447225570679, + "num_tokens": 26163796.0, + "step": 673 + }, + { + "epoch": 0.08573972777000381, + "grad_norm": 2.5222997665405273, + "learning_rate": 2.8529037727850785e-07, + "loss": 0.4045, + "mean_token_accuracy": 0.867527961730957, + "num_tokens": 26208212.0, + "step": 674 + }, + { + "epoch": 0.08586693804859433, + "grad_norm": 3.1145870685577393, + "learning_rate": 2.857142857142857e-07, + "loss": 0.4357, + "mean_token_accuracy": 0.8588769435882568, + "num_tokens": 26243932.0, + "step": 675 + }, + { + "epoch": 0.08599414832718484, + "grad_norm": 2.80926251411438, + "learning_rate": 2.861381941500636e-07, + "loss": 0.389, + "mean_token_accuracy": 0.8695273995399475, + "num_tokens": 26282354.0, + "step": 676 + }, + { + "epoch": 0.08612135860577534, + "grad_norm": 2.9545164108276367, + "learning_rate": 2.8656210258584144e-07, + "loss": 0.4428, + "mean_token_accuracy": 0.8514710664749146, + "num_tokens": 26320576.0, + "step": 677 + }, + { + "epoch": 0.08624856888436586, + "grad_norm": 3.0251965522766113, + "learning_rate": 2.869860110216193e-07, + "loss": 0.4893, + "mean_token_accuracy": 0.8377020359039307, + "num_tokens": 26360108.0, + "step": 678 + }, + { + "epoch": 0.08637577916295637, + "grad_norm": 2.894132375717163, + "learning_rate": 2.874099194573972e-07, + "loss": 0.4421, + "mean_token_accuracy": 0.8550804257392883, + "num_tokens": 26401347.0, + "step": 679 + }, + { + "epoch": 0.08650298944154687, + "grad_norm": 2.8958041667938232, + "learning_rate": 2.878338278931751e-07, + "loss": 0.4246, + "mean_token_accuracy": 0.8619457483291626, + "num_tokens": 26438276.0, + "step": 680 + }, + { + "epoch": 0.08663019972013739, + "grad_norm": 2.6557271480560303, + "learning_rate": 2.8825773632895293e-07, + "loss": 0.4236, + "mean_token_accuracy": 0.858047366142273, + "num_tokens": 26479990.0, + "step": 681 + }, + { + "epoch": 0.0867574099987279, + "grad_norm": 2.6619935035705566, + "learning_rate": 2.886816447647308e-07, + "loss": 0.424, + "mean_token_accuracy": 0.8599425554275513, + "num_tokens": 26521496.0, + "step": 682 + }, + { + "epoch": 0.08688462027731841, + "grad_norm": 3.1652348041534424, + "learning_rate": 2.891055532005087e-07, + "loss": 0.4395, + "mean_token_accuracy": 0.8499193787574768, + "num_tokens": 26556922.0, + "step": 683 + }, + { + "epoch": 0.08701183055590891, + "grad_norm": 2.687361717224121, + "learning_rate": 2.8952946163628657e-07, + "loss": 0.376, + "mean_token_accuracy": 0.8724969029426575, + "num_tokens": 26597324.0, + "step": 684 + }, + { + "epoch": 0.08713904083449943, + "grad_norm": 2.9370505809783936, + "learning_rate": 2.899533700720644e-07, + "loss": 0.4966, + "mean_token_accuracy": 0.836087167263031, + "num_tokens": 26637228.0, + "step": 685 + }, + { + "epoch": 0.08726625111308994, + "grad_norm": 2.8061485290527344, + "learning_rate": 2.9037727850784227e-07, + "loss": 0.5138, + "mean_token_accuracy": 0.8367780447006226, + "num_tokens": 26677910.0, + "step": 686 + }, + { + "epoch": 0.08739346139168044, + "grad_norm": 3.094682455062866, + "learning_rate": 2.9080118694362016e-07, + "loss": 0.4499, + "mean_token_accuracy": 0.8524328470230103, + "num_tokens": 26712826.0, + "step": 687 + }, + { + "epoch": 0.08752067167027096, + "grad_norm": 3.021409511566162, + "learning_rate": 2.9122509537939806e-07, + "loss": 0.4761, + "mean_token_accuracy": 0.8415564298629761, + "num_tokens": 26752900.0, + "step": 688 + }, + { + "epoch": 0.08764788194886147, + "grad_norm": 3.1381685733795166, + "learning_rate": 2.916490038151759e-07, + "loss": 0.5, + "mean_token_accuracy": 0.8399975299835205, + "num_tokens": 26794076.0, + "step": 689 + }, + { + "epoch": 0.08777509222745197, + "grad_norm": 2.890188217163086, + "learning_rate": 2.9207291225095376e-07, + "loss": 0.4928, + "mean_token_accuracy": 0.8396776914596558, + "num_tokens": 26832459.0, + "step": 690 + }, + { + "epoch": 0.08790230250604249, + "grad_norm": 3.3513948917388916, + "learning_rate": 2.9249682068673166e-07, + "loss": 0.4904, + "mean_token_accuracy": 0.8361996412277222, + "num_tokens": 26863876.0, + "step": 691 + }, + { + "epoch": 0.088029512784633, + "grad_norm": 2.9479708671569824, + "learning_rate": 2.9292072912250955e-07, + "loss": 0.4333, + "mean_token_accuracy": 0.854143500328064, + "num_tokens": 26903637.0, + "step": 692 + }, + { + "epoch": 0.0881567230632235, + "grad_norm": 2.6967194080352783, + "learning_rate": 2.933446375582874e-07, + "loss": 0.3838, + "mean_token_accuracy": 0.8697229027748108, + "num_tokens": 26948269.0, + "step": 693 + }, + { + "epoch": 0.08828393334181402, + "grad_norm": 2.9869329929351807, + "learning_rate": 2.9376854599406525e-07, + "loss": 0.4136, + "mean_token_accuracy": 0.8632926940917969, + "num_tokens": 26986749.0, + "step": 694 + }, + { + "epoch": 0.08841114362040453, + "grad_norm": 2.9330122470855713, + "learning_rate": 2.9419245442984315e-07, + "loss": 0.4163, + "mean_token_accuracy": 0.8638838529586792, + "num_tokens": 27021668.0, + "step": 695 + }, + { + "epoch": 0.08853835389899505, + "grad_norm": 2.549880266189575, + "learning_rate": 2.9461636286562104e-07, + "loss": 0.4687, + "mean_token_accuracy": 0.847666323184967, + "num_tokens": 27074350.0, + "step": 696 + }, + { + "epoch": 0.08866556417758555, + "grad_norm": 2.7057619094848633, + "learning_rate": 2.9504027130139884e-07, + "loss": 0.4633, + "mean_token_accuracy": 0.8506920337677002, + "num_tokens": 27119337.0, + "step": 697 + }, + { + "epoch": 0.08879277445617606, + "grad_norm": 3.051544666290283, + "learning_rate": 2.9546417973717674e-07, + "loss": 0.5312, + "mean_token_accuracy": 0.8274299502372742, + "num_tokens": 27159525.0, + "step": 698 + }, + { + "epoch": 0.08891998473476657, + "grad_norm": 2.930191993713379, + "learning_rate": 2.9588808817295464e-07, + "loss": 0.4569, + "mean_token_accuracy": 0.8483388423919678, + "num_tokens": 27197746.0, + "step": 699 + }, + { + "epoch": 0.08904719501335707, + "grad_norm": 2.7221219539642334, + "learning_rate": 2.9631199660873253e-07, + "loss": 0.3973, + "mean_token_accuracy": 0.8686877489089966, + "num_tokens": 27235986.0, + "step": 700 + }, + { + "epoch": 0.08917440529194759, + "grad_norm": 2.819042682647705, + "learning_rate": 2.9673590504451033e-07, + "loss": 0.4494, + "mean_token_accuracy": 0.8500218391418457, + "num_tokens": 27277840.0, + "step": 701 + }, + { + "epoch": 0.0893016155705381, + "grad_norm": 2.8779282569885254, + "learning_rate": 2.9715981348028823e-07, + "loss": 0.4479, + "mean_token_accuracy": 0.8557592630386353, + "num_tokens": 27317999.0, + "step": 702 + }, + { + "epoch": 0.0894288258491286, + "grad_norm": 3.0352816581726074, + "learning_rate": 2.975837219160661e-07, + "loss": 0.4316, + "mean_token_accuracy": 0.8579015731811523, + "num_tokens": 27355163.0, + "step": 703 + }, + { + "epoch": 0.08955603612771912, + "grad_norm": 2.823840618133545, + "learning_rate": 2.98007630351844e-07, + "loss": 0.5069, + "mean_token_accuracy": 0.8335480093955994, + "num_tokens": 27400827.0, + "step": 704 + }, + { + "epoch": 0.08968324640630963, + "grad_norm": 2.689521074295044, + "learning_rate": 2.984315387876218e-07, + "loss": 0.4606, + "mean_token_accuracy": 0.8497809767723083, + "num_tokens": 27447595.0, + "step": 705 + }, + { + "epoch": 0.08981045668490013, + "grad_norm": 3.025529623031616, + "learning_rate": 2.988554472233997e-07, + "loss": 0.4161, + "mean_token_accuracy": 0.8598623871803284, + "num_tokens": 27481267.0, + "step": 706 + }, + { + "epoch": 0.08993766696349065, + "grad_norm": 2.8117802143096924, + "learning_rate": 2.992793556591776e-07, + "loss": 0.4843, + "mean_token_accuracy": 0.8406538963317871, + "num_tokens": 27521331.0, + "step": 707 + }, + { + "epoch": 0.09006487724208116, + "grad_norm": 2.772376298904419, + "learning_rate": 2.997032640949555e-07, + "loss": 0.4621, + "mean_token_accuracy": 0.846778154373169, + "num_tokens": 27565706.0, + "step": 708 + }, + { + "epoch": 0.09019208752067168, + "grad_norm": 3.2305049896240234, + "learning_rate": 3.001271725307333e-07, + "loss": 0.493, + "mean_token_accuracy": 0.8439432382583618, + "num_tokens": 27602005.0, + "step": 709 + }, + { + "epoch": 0.09031929779926218, + "grad_norm": 2.882101058959961, + "learning_rate": 3.005510809665112e-07, + "loss": 0.4432, + "mean_token_accuracy": 0.8522078394889832, + "num_tokens": 27640826.0, + "step": 710 + }, + { + "epoch": 0.09044650807785269, + "grad_norm": 2.822490930557251, + "learning_rate": 3.009749894022891e-07, + "loss": 0.4548, + "mean_token_accuracy": 0.8521125912666321, + "num_tokens": 27681222.0, + "step": 711 + }, + { + "epoch": 0.0905737183564432, + "grad_norm": 2.692427635192871, + "learning_rate": 3.01398897838067e-07, + "loss": 0.4135, + "mean_token_accuracy": 0.859312117099762, + "num_tokens": 27726374.0, + "step": 712 + }, + { + "epoch": 0.0907009286350337, + "grad_norm": 2.7939915657043457, + "learning_rate": 3.018228062738448e-07, + "loss": 0.4025, + "mean_token_accuracy": 0.865740180015564, + "num_tokens": 27761849.0, + "step": 713 + }, + { + "epoch": 0.09082813891362422, + "grad_norm": 2.937980890274048, + "learning_rate": 3.022467147096227e-07, + "loss": 0.4197, + "mean_token_accuracy": 0.8597393035888672, + "num_tokens": 27797414.0, + "step": 714 + }, + { + "epoch": 0.09095534919221474, + "grad_norm": 2.7297167778015137, + "learning_rate": 3.026706231454006e-07, + "loss": 0.4192, + "mean_token_accuracy": 0.8604477643966675, + "num_tokens": 27836572.0, + "step": 715 + }, + { + "epoch": 0.09108255947080524, + "grad_norm": 3.1354901790618896, + "learning_rate": 3.0309453158117844e-07, + "loss": 0.4189, + "mean_token_accuracy": 0.8620175719261169, + "num_tokens": 27872627.0, + "step": 716 + }, + { + "epoch": 0.09120976974939575, + "grad_norm": 2.6532979011535645, + "learning_rate": 3.035184400169563e-07, + "loss": 0.3908, + "mean_token_accuracy": 0.865391731262207, + "num_tokens": 27910889.0, + "step": 717 + }, + { + "epoch": 0.09133698002798626, + "grad_norm": 2.5933046340942383, + "learning_rate": 3.039423484527342e-07, + "loss": 0.3894, + "mean_token_accuracy": 0.8693938255310059, + "num_tokens": 27947862.0, + "step": 718 + }, + { + "epoch": 0.09146419030657676, + "grad_norm": 3.1169004440307617, + "learning_rate": 3.043662568885121e-07, + "loss": 0.4479, + "mean_token_accuracy": 0.8521614074707031, + "num_tokens": 27980761.0, + "step": 719 + }, + { + "epoch": 0.09159140058516728, + "grad_norm": 2.5922422409057617, + "learning_rate": 3.0479016532428993e-07, + "loss": 0.4501, + "mean_token_accuracy": 0.8506655693054199, + "num_tokens": 28025780.0, + "step": 720 + }, + { + "epoch": 0.0917186108637578, + "grad_norm": 2.492323160171509, + "learning_rate": 3.052140737600678e-07, + "loss": 0.4161, + "mean_token_accuracy": 0.8625723123550415, + "num_tokens": 28069043.0, + "step": 721 + }, + { + "epoch": 0.09184582114234831, + "grad_norm": 2.727525472640991, + "learning_rate": 3.056379821958457e-07, + "loss": 0.3957, + "mean_token_accuracy": 0.8664670586585999, + "num_tokens": 28107067.0, + "step": 722 + }, + { + "epoch": 0.09197303142093881, + "grad_norm": 2.7353227138519287, + "learning_rate": 3.060618906316236e-07, + "loss": 0.4608, + "mean_token_accuracy": 0.8488886952400208, + "num_tokens": 28149653.0, + "step": 723 + }, + { + "epoch": 0.09210024169952932, + "grad_norm": 2.9262752532958984, + "learning_rate": 3.064857990674014e-07, + "loss": 0.3982, + "mean_token_accuracy": 0.8651278018951416, + "num_tokens": 28183560.0, + "step": 724 + }, + { + "epoch": 0.09222745197811984, + "grad_norm": 2.8729543685913086, + "learning_rate": 3.0690970750317927e-07, + "loss": 0.4182, + "mean_token_accuracy": 0.8595748543739319, + "num_tokens": 28222180.0, + "step": 725 + }, + { + "epoch": 0.09235466225671034, + "grad_norm": 2.956915855407715, + "learning_rate": 3.0733361593895717e-07, + "loss": 0.4252, + "mean_token_accuracy": 0.8588425517082214, + "num_tokens": 28258682.0, + "step": 726 + }, + { + "epoch": 0.09248187253530085, + "grad_norm": 2.6765217781066895, + "learning_rate": 3.0775752437473507e-07, + "loss": 0.4339, + "mean_token_accuracy": 0.8598107695579529, + "num_tokens": 28300110.0, + "step": 727 + }, + { + "epoch": 0.09260908281389137, + "grad_norm": 2.6612966060638428, + "learning_rate": 3.081814328105129e-07, + "loss": 0.3913, + "mean_token_accuracy": 0.8685066103935242, + "num_tokens": 28338313.0, + "step": 728 + }, + { + "epoch": 0.09273629309248187, + "grad_norm": 3.279305934906006, + "learning_rate": 3.0860534124629076e-07, + "loss": 0.433, + "mean_token_accuracy": 0.8570994138717651, + "num_tokens": 28373356.0, + "step": 729 + }, + { + "epoch": 0.09286350337107238, + "grad_norm": 2.957317352294922, + "learning_rate": 3.0902924968206866e-07, + "loss": 0.449, + "mean_token_accuracy": 0.8493174314498901, + "num_tokens": 28410721.0, + "step": 730 + }, + { + "epoch": 0.0929907136496629, + "grad_norm": 2.87004017829895, + "learning_rate": 3.0945315811784656e-07, + "loss": 0.4114, + "mean_token_accuracy": 0.8601545691490173, + "num_tokens": 28447429.0, + "step": 731 + }, + { + "epoch": 0.0931179239282534, + "grad_norm": 2.9085001945495605, + "learning_rate": 3.098770665536244e-07, + "loss": 0.4615, + "mean_token_accuracy": 0.8481404185295105, + "num_tokens": 28487805.0, + "step": 732 + }, + { + "epoch": 0.09324513420684391, + "grad_norm": 2.960580587387085, + "learning_rate": 3.1030097498940225e-07, + "loss": 0.4257, + "mean_token_accuracy": 0.8611960411071777, + "num_tokens": 28524759.0, + "step": 733 + }, + { + "epoch": 0.09337234448543442, + "grad_norm": 2.6477158069610596, + "learning_rate": 3.1072488342518015e-07, + "loss": 0.3871, + "mean_token_accuracy": 0.8710211515426636, + "num_tokens": 28564110.0, + "step": 734 + }, + { + "epoch": 0.09349955476402494, + "grad_norm": 2.7684872150421143, + "learning_rate": 3.11148791860958e-07, + "loss": 0.4245, + "mean_token_accuracy": 0.8596705198287964, + "num_tokens": 28604768.0, + "step": 735 + }, + { + "epoch": 0.09362676504261544, + "grad_norm": 3.0466251373291016, + "learning_rate": 3.115727002967359e-07, + "loss": 0.4464, + "mean_token_accuracy": 0.8503625392913818, + "num_tokens": 28639712.0, + "step": 736 + }, + { + "epoch": 0.09375397532120595, + "grad_norm": 3.1124908924102783, + "learning_rate": 3.1199660873251374e-07, + "loss": 0.4216, + "mean_token_accuracy": 0.8611330986022949, + "num_tokens": 28672466.0, + "step": 737 + }, + { + "epoch": 0.09388118559979647, + "grad_norm": 3.0189149379730225, + "learning_rate": 3.1242051716829164e-07, + "loss": 0.4359, + "mean_token_accuracy": 0.8578939437866211, + "num_tokens": 28707953.0, + "step": 738 + }, + { + "epoch": 0.09400839587838697, + "grad_norm": 2.6745564937591553, + "learning_rate": 3.128444256040695e-07, + "loss": 0.3943, + "mean_token_accuracy": 0.8658502101898193, + "num_tokens": 28749638.0, + "step": 739 + }, + { + "epoch": 0.09413560615697748, + "grad_norm": 3.021698474884033, + "learning_rate": 3.132683340398474e-07, + "loss": 0.4372, + "mean_token_accuracy": 0.8536290526390076, + "num_tokens": 28786241.0, + "step": 740 + }, + { + "epoch": 0.094262816435568, + "grad_norm": 2.5810275077819824, + "learning_rate": 3.1369224247562523e-07, + "loss": 0.3732, + "mean_token_accuracy": 0.8746166229248047, + "num_tokens": 28824964.0, + "step": 741 + }, + { + "epoch": 0.0943900267141585, + "grad_norm": 2.9007222652435303, + "learning_rate": 3.1411615091140313e-07, + "loss": 0.4256, + "mean_token_accuracy": 0.8608142137527466, + "num_tokens": 28860502.0, + "step": 742 + }, + { + "epoch": 0.09451723699274901, + "grad_norm": 2.545074701309204, + "learning_rate": 3.14540059347181e-07, + "loss": 0.4263, + "mean_token_accuracy": 0.8591489791870117, + "num_tokens": 28905546.0, + "step": 743 + }, + { + "epoch": 0.09464444727133953, + "grad_norm": 3.1143040657043457, + "learning_rate": 3.149639677829589e-07, + "loss": 0.446, + "mean_token_accuracy": 0.8538666367530823, + "num_tokens": 28939330.0, + "step": 744 + }, + { + "epoch": 0.09477165754993004, + "grad_norm": 2.963064670562744, + "learning_rate": 3.153878762187368e-07, + "loss": 0.4145, + "mean_token_accuracy": 0.857477605342865, + "num_tokens": 28977375.0, + "step": 745 + }, + { + "epoch": 0.09489886782852054, + "grad_norm": 2.9687399864196777, + "learning_rate": 3.158117846545146e-07, + "loss": 0.4375, + "mean_token_accuracy": 0.8606466054916382, + "num_tokens": 29016979.0, + "step": 746 + }, + { + "epoch": 0.09502607810711106, + "grad_norm": 2.7894790172576904, + "learning_rate": 3.1623569309029247e-07, + "loss": 0.4738, + "mean_token_accuracy": 0.8392453193664551, + "num_tokens": 29056280.0, + "step": 747 + }, + { + "epoch": 0.09515328838570157, + "grad_norm": 2.9902260303497314, + "learning_rate": 3.1665960152607037e-07, + "loss": 0.4511, + "mean_token_accuracy": 0.8505188226699829, + "num_tokens": 29091946.0, + "step": 748 + }, + { + "epoch": 0.09528049866429207, + "grad_norm": 3.003805160522461, + "learning_rate": 3.1708350996184826e-07, + "loss": 0.4286, + "mean_token_accuracy": 0.8559589385986328, + "num_tokens": 29126927.0, + "step": 749 + }, + { + "epoch": 0.09540770894288259, + "grad_norm": 2.8602638244628906, + "learning_rate": 3.175074183976261e-07, + "loss": 0.4561, + "mean_token_accuracy": 0.85016930103302, + "num_tokens": 29168873.0, + "step": 750 + }, + { + "epoch": 0.0955349192214731, + "grad_norm": 2.660355806350708, + "learning_rate": 3.1793132683340396e-07, + "loss": 0.4078, + "mean_token_accuracy": 0.8622951507568359, + "num_tokens": 29208708.0, + "step": 751 + }, + { + "epoch": 0.0956621295000636, + "grad_norm": 2.9818286895751953, + "learning_rate": 3.1835523526918186e-07, + "loss": 0.4786, + "mean_token_accuracy": 0.8411779403686523, + "num_tokens": 29246680.0, + "step": 752 + }, + { + "epoch": 0.09578933977865411, + "grad_norm": 2.9285457134246826, + "learning_rate": 3.1877914370495975e-07, + "loss": 0.3985, + "mean_token_accuracy": 0.8650662899017334, + "num_tokens": 29280008.0, + "step": 753 + }, + { + "epoch": 0.09591655005724463, + "grad_norm": 2.754424571990967, + "learning_rate": 3.1920305214073755e-07, + "loss": 0.401, + "mean_token_accuracy": 0.8650378584861755, + "num_tokens": 29318547.0, + "step": 754 + }, + { + "epoch": 0.09604376033583513, + "grad_norm": 2.8755764961242676, + "learning_rate": 3.1962696057651545e-07, + "loss": 0.4255, + "mean_token_accuracy": 0.8580299019813538, + "num_tokens": 29356401.0, + "step": 755 + }, + { + "epoch": 0.09617097061442564, + "grad_norm": 2.8099424839019775, + "learning_rate": 3.2005086901229335e-07, + "loss": 0.3674, + "mean_token_accuracy": 0.874613344669342, + "num_tokens": 29389646.0, + "step": 756 + }, + { + "epoch": 0.09629818089301616, + "grad_norm": 2.682870626449585, + "learning_rate": 3.2047477744807125e-07, + "loss": 0.4178, + "mean_token_accuracy": 0.861273467540741, + "num_tokens": 29431591.0, + "step": 757 + }, + { + "epoch": 0.09642539117160667, + "grad_norm": 2.998338460922241, + "learning_rate": 3.2089868588384904e-07, + "loss": 0.4487, + "mean_token_accuracy": 0.8480445146560669, + "num_tokens": 29468010.0, + "step": 758 + }, + { + "epoch": 0.09655260145019717, + "grad_norm": 2.8356223106384277, + "learning_rate": 3.2132259431962694e-07, + "loss": 0.3847, + "mean_token_accuracy": 0.8673666715621948, + "num_tokens": 29502897.0, + "step": 759 + }, + { + "epoch": 0.09667981172878769, + "grad_norm": 2.861487865447998, + "learning_rate": 3.2174650275540484e-07, + "loss": 0.3824, + "mean_token_accuracy": 0.8736017942428589, + "num_tokens": 29539586.0, + "step": 760 + }, + { + "epoch": 0.0968070220073782, + "grad_norm": 2.897087574005127, + "learning_rate": 3.2217041119118274e-07, + "loss": 0.4398, + "mean_token_accuracy": 0.8526968359947205, + "num_tokens": 29577598.0, + "step": 761 + }, + { + "epoch": 0.0969342322859687, + "grad_norm": 2.718775987625122, + "learning_rate": 3.2259431962696053e-07, + "loss": 0.3796, + "mean_token_accuracy": 0.8739378452301025, + "num_tokens": 29614369.0, + "step": 762 + }, + { + "epoch": 0.09706144256455922, + "grad_norm": 2.962606191635132, + "learning_rate": 3.2301822806273843e-07, + "loss": 0.4831, + "mean_token_accuracy": 0.8389495611190796, + "num_tokens": 29655138.0, + "step": 763 + }, + { + "epoch": 0.09718865284314973, + "grad_norm": 2.754972457885742, + "learning_rate": 3.2344213649851633e-07, + "loss": 0.3849, + "mean_token_accuracy": 0.8731626272201538, + "num_tokens": 29700663.0, + "step": 764 + }, + { + "epoch": 0.09731586312174023, + "grad_norm": 2.8522987365722656, + "learning_rate": 3.238660449342942e-07, + "loss": 0.41, + "mean_token_accuracy": 0.860392689704895, + "num_tokens": 29737050.0, + "step": 765 + }, + { + "epoch": 0.09744307340033075, + "grad_norm": 3.077165126800537, + "learning_rate": 3.24289953370072e-07, + "loss": 0.4055, + "mean_token_accuracy": 0.8643521070480347, + "num_tokens": 29770104.0, + "step": 766 + }, + { + "epoch": 0.09757028367892126, + "grad_norm": 2.993706703186035, + "learning_rate": 3.247138618058499e-07, + "loss": 0.5028, + "mean_token_accuracy": 0.8348210453987122, + "num_tokens": 29809158.0, + "step": 767 + }, + { + "epoch": 0.09769749395751176, + "grad_norm": 2.7270755767822266, + "learning_rate": 3.251377702416278e-07, + "loss": 0.3739, + "mean_token_accuracy": 0.8767487406730652, + "num_tokens": 29845666.0, + "step": 768 + }, + { + "epoch": 0.09782470423610228, + "grad_norm": 2.783796787261963, + "learning_rate": 3.255616786774057e-07, + "loss": 0.4377, + "mean_token_accuracy": 0.8533982038497925, + "num_tokens": 29887396.0, + "step": 769 + }, + { + "epoch": 0.09795191451469279, + "grad_norm": 3.079270601272583, + "learning_rate": 3.259855871131835e-07, + "loss": 0.4335, + "mean_token_accuracy": 0.8582950830459595, + "num_tokens": 29920941.0, + "step": 770 + }, + { + "epoch": 0.0980791247932833, + "grad_norm": 2.861194372177124, + "learning_rate": 3.264094955489614e-07, + "loss": 0.4221, + "mean_token_accuracy": 0.8614653944969177, + "num_tokens": 29960475.0, + "step": 771 + }, + { + "epoch": 0.0982063350718738, + "grad_norm": 2.695204973220825, + "learning_rate": 3.268334039847393e-07, + "loss": 0.4447, + "mean_token_accuracy": 0.8481158018112183, + "num_tokens": 30003906.0, + "step": 772 + }, + { + "epoch": 0.09833354535046432, + "grad_norm": 2.8253071308135986, + "learning_rate": 3.2725731242051715e-07, + "loss": 0.4121, + "mean_token_accuracy": 0.8609205484390259, + "num_tokens": 30040012.0, + "step": 773 + }, + { + "epoch": 0.09846075562905483, + "grad_norm": 3.1171557903289795, + "learning_rate": 3.27681220856295e-07, + "loss": 0.4841, + "mean_token_accuracy": 0.841873049736023, + "num_tokens": 30075104.0, + "step": 774 + }, + { + "epoch": 0.09858796590764533, + "grad_norm": 2.942277669906616, + "learning_rate": 3.281051292920729e-07, + "loss": 0.4636, + "mean_token_accuracy": 0.8497517108917236, + "num_tokens": 30113306.0, + "step": 775 + }, + { + "epoch": 0.09871517618623585, + "grad_norm": 2.819589138031006, + "learning_rate": 3.285290377278508e-07, + "loss": 0.3979, + "mean_token_accuracy": 0.8698340654373169, + "num_tokens": 30152050.0, + "step": 776 + }, + { + "epoch": 0.09884238646482636, + "grad_norm": 2.781806230545044, + "learning_rate": 3.2895294616362864e-07, + "loss": 0.4264, + "mean_token_accuracy": 0.8627758622169495, + "num_tokens": 30197207.0, + "step": 777 + }, + { + "epoch": 0.09896959674341686, + "grad_norm": 2.5498690605163574, + "learning_rate": 3.293768545994065e-07, + "loss": 0.4679, + "mean_token_accuracy": 0.8453963994979858, + "num_tokens": 30246798.0, + "step": 778 + }, + { + "epoch": 0.09909680702200738, + "grad_norm": 2.6026175022125244, + "learning_rate": 3.298007630351844e-07, + "loss": 0.373, + "mean_token_accuracy": 0.8747968077659607, + "num_tokens": 30290105.0, + "step": 779 + }, + { + "epoch": 0.09922401730059789, + "grad_norm": 2.6598808765411377, + "learning_rate": 3.302246714709623e-07, + "loss": 0.4135, + "mean_token_accuracy": 0.8634341955184937, + "num_tokens": 30332282.0, + "step": 780 + }, + { + "epoch": 0.09935122757918839, + "grad_norm": 2.648743152618408, + "learning_rate": 3.3064857990674013e-07, + "loss": 0.3857, + "mean_token_accuracy": 0.8708937764167786, + "num_tokens": 30373673.0, + "step": 781 + }, + { + "epoch": 0.0994784378577789, + "grad_norm": 3.1391024589538574, + "learning_rate": 3.31072488342518e-07, + "loss": 0.4284, + "mean_token_accuracy": 0.8560410737991333, + "num_tokens": 30407510.0, + "step": 782 + }, + { + "epoch": 0.09960564813636942, + "grad_norm": 2.796837568283081, + "learning_rate": 3.314963967782959e-07, + "loss": 0.4044, + "mean_token_accuracy": 0.8653059005737305, + "num_tokens": 30445092.0, + "step": 783 + }, + { + "epoch": 0.09973285841495994, + "grad_norm": 2.6902801990509033, + "learning_rate": 3.319203052140738e-07, + "loss": 0.4532, + "mean_token_accuracy": 0.8517742156982422, + "num_tokens": 30488250.0, + "step": 784 + }, + { + "epoch": 0.09986006869355044, + "grad_norm": 3.1773874759674072, + "learning_rate": 3.323442136498516e-07, + "loss": 0.4049, + "mean_token_accuracy": 0.8662338852882385, + "num_tokens": 30521418.0, + "step": 785 + }, + { + "epoch": 0.09998727897214095, + "grad_norm": 2.8506886959075928, + "learning_rate": 3.3276812208562947e-07, + "loss": 0.4575, + "mean_token_accuracy": 0.8413897156715393, + "num_tokens": 30562317.0, + "step": 786 + }, + { + "epoch": 0.10011448925073146, + "grad_norm": 2.9183952808380127, + "learning_rate": 3.3319203052140737e-07, + "loss": 0.3862, + "mean_token_accuracy": 0.8695852756500244, + "num_tokens": 30595158.0, + "step": 787 + }, + { + "epoch": 0.10024169952932196, + "grad_norm": 3.0884299278259277, + "learning_rate": 3.336159389571852e-07, + "loss": 0.4486, + "mean_token_accuracy": 0.8519614934921265, + "num_tokens": 30629607.0, + "step": 788 + }, + { + "epoch": 0.10036890980791248, + "grad_norm": 2.9198005199432373, + "learning_rate": 3.340398473929631e-07, + "loss": 0.4194, + "mean_token_accuracy": 0.8600955605506897, + "num_tokens": 30667662.0, + "step": 789 + }, + { + "epoch": 0.100496120086503, + "grad_norm": 2.878326416015625, + "learning_rate": 3.3446375582874096e-07, + "loss": 0.4256, + "mean_token_accuracy": 0.8565289974212646, + "num_tokens": 30706133.0, + "step": 790 + }, + { + "epoch": 0.1006233303650935, + "grad_norm": 3.111365795135498, + "learning_rate": 3.3488766426451886e-07, + "loss": 0.396, + "mean_token_accuracy": 0.8690518736839294, + "num_tokens": 30739183.0, + "step": 791 + }, + { + "epoch": 0.10075054064368401, + "grad_norm": 2.907367706298828, + "learning_rate": 3.353115727002967e-07, + "loss": 0.4343, + "mean_token_accuracy": 0.8561940789222717, + "num_tokens": 30779724.0, + "step": 792 + }, + { + "epoch": 0.10087775092227452, + "grad_norm": 2.8298161029815674, + "learning_rate": 3.357354811360746e-07, + "loss": 0.4458, + "mean_token_accuracy": 0.8520951271057129, + "num_tokens": 30819431.0, + "step": 793 + }, + { + "epoch": 0.10100496120086502, + "grad_norm": 2.7483716011047363, + "learning_rate": 3.3615938957185245e-07, + "loss": 0.4613, + "mean_token_accuracy": 0.8469327688217163, + "num_tokens": 30863485.0, + "step": 794 + }, + { + "epoch": 0.10113217147945554, + "grad_norm": 2.838848114013672, + "learning_rate": 3.3658329800763035e-07, + "loss": 0.4165, + "mean_token_accuracy": 0.8592993021011353, + "num_tokens": 30900248.0, + "step": 795 + }, + { + "epoch": 0.10125938175804605, + "grad_norm": 2.6737220287323, + "learning_rate": 3.370072064434082e-07, + "loss": 0.4256, + "mean_token_accuracy": 0.861075758934021, + "num_tokens": 30945651.0, + "step": 796 + }, + { + "epoch": 0.10138659203663657, + "grad_norm": 2.9824399948120117, + "learning_rate": 3.374311148791861e-07, + "loss": 0.4384, + "mean_token_accuracy": 0.8565686345100403, + "num_tokens": 30986477.0, + "step": 797 + }, + { + "epoch": 0.10151380231522707, + "grad_norm": 2.8327994346618652, + "learning_rate": 3.3785502331496394e-07, + "loss": 0.4123, + "mean_token_accuracy": 0.8609697818756104, + "num_tokens": 31022642.0, + "step": 798 + }, + { + "epoch": 0.10164101259381758, + "grad_norm": 2.8622334003448486, + "learning_rate": 3.3827893175074184e-07, + "loss": 0.4115, + "mean_token_accuracy": 0.8636876940727234, + "num_tokens": 31060967.0, + "step": 799 + }, + { + "epoch": 0.1017682228724081, + "grad_norm": 2.7016513347625732, + "learning_rate": 3.387028401865197e-07, + "loss": 0.4233, + "mean_token_accuracy": 0.8546864986419678, + "num_tokens": 31103523.0, + "step": 800 + }, + { + "epoch": 0.1018954331509986, + "grad_norm": 2.9703943729400635, + "learning_rate": 3.391267486222976e-07, + "loss": 0.413, + "mean_token_accuracy": 0.8612384796142578, + "num_tokens": 31140688.0, + "step": 801 + }, + { + "epoch": 0.10202264342958911, + "grad_norm": 2.8345141410827637, + "learning_rate": 3.3955065705807543e-07, + "loss": 0.4113, + "mean_token_accuracy": 0.858860194683075, + "num_tokens": 31179448.0, + "step": 802 + }, + { + "epoch": 0.10214985370817962, + "grad_norm": 3.307574987411499, + "learning_rate": 3.3997456549385333e-07, + "loss": 0.465, + "mean_token_accuracy": 0.8460612297058105, + "num_tokens": 31215392.0, + "step": 803 + }, + { + "epoch": 0.10227706398677013, + "grad_norm": 2.811976194381714, + "learning_rate": 3.403984739296312e-07, + "loss": 0.4568, + "mean_token_accuracy": 0.8495973348617554, + "num_tokens": 31256556.0, + "step": 804 + }, + { + "epoch": 0.10240427426536064, + "grad_norm": 2.8650455474853516, + "learning_rate": 3.408223823654091e-07, + "loss": 0.4082, + "mean_token_accuracy": 0.8630320429801941, + "num_tokens": 31292794.0, + "step": 805 + }, + { + "epoch": 0.10253148454395115, + "grad_norm": 3.3868565559387207, + "learning_rate": 3.412462908011869e-07, + "loss": 0.4616, + "mean_token_accuracy": 0.849116861820221, + "num_tokens": 31325626.0, + "step": 806 + }, + { + "epoch": 0.10265869482254165, + "grad_norm": 2.872730016708374, + "learning_rate": 3.4167019923696477e-07, + "loss": 0.3845, + "mean_token_accuracy": 0.8732919096946716, + "num_tokens": 31360011.0, + "step": 807 + }, + { + "epoch": 0.10278590510113217, + "grad_norm": 2.49716854095459, + "learning_rate": 3.4209410767274267e-07, + "loss": 0.3952, + "mean_token_accuracy": 0.8690108060836792, + "num_tokens": 31405191.0, + "step": 808 + }, + { + "epoch": 0.10291311537972268, + "grad_norm": 2.9275946617126465, + "learning_rate": 3.4251801610852057e-07, + "loss": 0.3657, + "mean_token_accuracy": 0.8798779249191284, + "num_tokens": 31441371.0, + "step": 809 + }, + { + "epoch": 0.1030403256583132, + "grad_norm": 2.9380877017974854, + "learning_rate": 3.429419245442984e-07, + "loss": 0.4504, + "mean_token_accuracy": 0.8479812741279602, + "num_tokens": 31480501.0, + "step": 810 + }, + { + "epoch": 0.1031675359369037, + "grad_norm": 3.0262949466705322, + "learning_rate": 3.4336583298007626e-07, + "loss": 0.4486, + "mean_token_accuracy": 0.8535415530204773, + "num_tokens": 31517171.0, + "step": 811 + }, + { + "epoch": 0.10329474621549421, + "grad_norm": 2.525743007659912, + "learning_rate": 3.4378974141585416e-07, + "loss": 0.3959, + "mean_token_accuracy": 0.8656994104385376, + "num_tokens": 31558964.0, + "step": 812 + }, + { + "epoch": 0.10342195649408473, + "grad_norm": 2.8851683139801025, + "learning_rate": 3.4421364985163206e-07, + "loss": 0.4273, + "mean_token_accuracy": 0.8561896085739136, + "num_tokens": 31596620.0, + "step": 813 + }, + { + "epoch": 0.10354916677267523, + "grad_norm": 2.849065065383911, + "learning_rate": 3.446375582874099e-07, + "loss": 0.424, + "mean_token_accuracy": 0.8616191744804382, + "num_tokens": 31637292.0, + "step": 814 + }, + { + "epoch": 0.10367637705126574, + "grad_norm": 3.054518699645996, + "learning_rate": 3.4506146672318775e-07, + "loss": 0.4123, + "mean_token_accuracy": 0.8594712018966675, + "num_tokens": 31668134.0, + "step": 815 + }, + { + "epoch": 0.10380358732985626, + "grad_norm": 2.8842241764068604, + "learning_rate": 3.4548537515896565e-07, + "loss": 0.4385, + "mean_token_accuracy": 0.8534254431724548, + "num_tokens": 31705624.0, + "step": 816 + }, + { + "epoch": 0.10393079760844676, + "grad_norm": 2.706235885620117, + "learning_rate": 3.4590928359474355e-07, + "loss": 0.4159, + "mean_token_accuracy": 0.8620871305465698, + "num_tokens": 31750229.0, + "step": 817 + }, + { + "epoch": 0.10405800788703727, + "grad_norm": 2.9571776390075684, + "learning_rate": 3.463331920305214e-07, + "loss": 0.4243, + "mean_token_accuracy": 0.8587911128997803, + "num_tokens": 31788908.0, + "step": 818 + }, + { + "epoch": 0.10418521816562779, + "grad_norm": 3.0634162425994873, + "learning_rate": 3.4675710046629924e-07, + "loss": 0.4219, + "mean_token_accuracy": 0.862287163734436, + "num_tokens": 31823522.0, + "step": 819 + }, + { + "epoch": 0.1043124284442183, + "grad_norm": 3.0632975101470947, + "learning_rate": 3.4718100890207714e-07, + "loss": 0.4693, + "mean_token_accuracy": 0.8489570617675781, + "num_tokens": 31861506.0, + "step": 820 + }, + { + "epoch": 0.1044396387228088, + "grad_norm": 3.016679525375366, + "learning_rate": 3.4760491733785504e-07, + "loss": 0.4102, + "mean_token_accuracy": 0.8636928200721741, + "num_tokens": 31897575.0, + "step": 821 + }, + { + "epoch": 0.10456684900139931, + "grad_norm": 2.843050241470337, + "learning_rate": 3.480288257736329e-07, + "loss": 0.4677, + "mean_token_accuracy": 0.8450868129730225, + "num_tokens": 31940063.0, + "step": 822 + }, + { + "epoch": 0.10469405927998983, + "grad_norm": 11.955665588378906, + "learning_rate": 3.4845273420941073e-07, + "loss": 0.4548, + "mean_token_accuracy": 0.8504536151885986, + "num_tokens": 31977789.0, + "step": 823 + }, + { + "epoch": 0.10482126955858033, + "grad_norm": 2.858490467071533, + "learning_rate": 3.4887664264518863e-07, + "loss": 0.4265, + "mean_token_accuracy": 0.8539507389068604, + "num_tokens": 32017596.0, + "step": 824 + }, + { + "epoch": 0.10494847983717084, + "grad_norm": 3.0772552490234375, + "learning_rate": 3.4930055108096653e-07, + "loss": 0.4318, + "mean_token_accuracy": 0.8591512441635132, + "num_tokens": 32052838.0, + "step": 825 + }, + { + "epoch": 0.10507569011576136, + "grad_norm": 2.7059268951416016, + "learning_rate": 3.497244595167443e-07, + "loss": 0.391, + "mean_token_accuracy": 0.865963339805603, + "num_tokens": 32095753.0, + "step": 826 + }, + { + "epoch": 0.10520290039435186, + "grad_norm": 2.8940067291259766, + "learning_rate": 3.501483679525222e-07, + "loss": 0.4445, + "mean_token_accuracy": 0.8548101186752319, + "num_tokens": 32133125.0, + "step": 827 + }, + { + "epoch": 0.10533011067294237, + "grad_norm": 2.934887409210205, + "learning_rate": 3.505722763883001e-07, + "loss": 0.4253, + "mean_token_accuracy": 0.8564348220825195, + "num_tokens": 32169112.0, + "step": 828 + }, + { + "epoch": 0.10545732095153289, + "grad_norm": 3.215749979019165, + "learning_rate": 3.50996184824078e-07, + "loss": 0.4843, + "mean_token_accuracy": 0.8429996967315674, + "num_tokens": 32204501.0, + "step": 829 + }, + { + "epoch": 0.10558453123012339, + "grad_norm": 2.8700549602508545, + "learning_rate": 3.514200932598558e-07, + "loss": 0.4181, + "mean_token_accuracy": 0.8560409545898438, + "num_tokens": 32242337.0, + "step": 830 + }, + { + "epoch": 0.1057117415087139, + "grad_norm": 10.24837875366211, + "learning_rate": 3.518440016956337e-07, + "loss": 0.4135, + "mean_token_accuracy": 0.863000214099884, + "num_tokens": 32274069.0, + "step": 831 + }, + { + "epoch": 0.10583895178730442, + "grad_norm": 2.6710541248321533, + "learning_rate": 3.522679101314116e-07, + "loss": 0.44, + "mean_token_accuracy": 0.853961706161499, + "num_tokens": 32318585.0, + "step": 832 + }, + { + "epoch": 0.10596616206589493, + "grad_norm": 3.042990207672119, + "learning_rate": 3.526918185671895e-07, + "loss": 0.4516, + "mean_token_accuracy": 0.8501008152961731, + "num_tokens": 32356789.0, + "step": 833 + }, + { + "epoch": 0.10609337234448543, + "grad_norm": 2.7006750106811523, + "learning_rate": 3.531157270029673e-07, + "loss": 0.4099, + "mean_token_accuracy": 0.8620035648345947, + "num_tokens": 32400176.0, + "step": 834 + }, + { + "epoch": 0.10622058262307595, + "grad_norm": 2.8065123558044434, + "learning_rate": 3.535396354387452e-07, + "loss": 0.3852, + "mean_token_accuracy": 0.8701838850975037, + "num_tokens": 32435720.0, + "step": 835 + }, + { + "epoch": 0.10634779290166646, + "grad_norm": 2.838495969772339, + "learning_rate": 3.539635438745231e-07, + "loss": 0.4074, + "mean_token_accuracy": 0.8635042905807495, + "num_tokens": 32472943.0, + "step": 836 + }, + { + "epoch": 0.10647500318025696, + "grad_norm": 2.719841718673706, + "learning_rate": 3.54387452310301e-07, + "loss": 0.4416, + "mean_token_accuracy": 0.8544507622718811, + "num_tokens": 32515928.0, + "step": 837 + }, + { + "epoch": 0.10660221345884748, + "grad_norm": 2.8890435695648193, + "learning_rate": 3.548113607460788e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8724361658096313, + "num_tokens": 32551170.0, + "step": 838 + }, + { + "epoch": 0.10672942373743799, + "grad_norm": 2.866062641143799, + "learning_rate": 3.552352691818567e-07, + "loss": 0.4473, + "mean_token_accuracy": 0.846739649772644, + "num_tokens": 32591652.0, + "step": 839 + }, + { + "epoch": 0.10685663401602849, + "grad_norm": 2.796517848968506, + "learning_rate": 3.556591776176346e-07, + "loss": 0.4643, + "mean_token_accuracy": 0.845600426197052, + "num_tokens": 32632202.0, + "step": 840 + }, + { + "epoch": 0.106983844294619, + "grad_norm": 2.780662775039673, + "learning_rate": 3.560830860534125e-07, + "loss": 0.4038, + "mean_token_accuracy": 0.866546630859375, + "num_tokens": 32673184.0, + "step": 841 + }, + { + "epoch": 0.10711105457320952, + "grad_norm": 2.7687454223632812, + "learning_rate": 3.565069944891903e-07, + "loss": 0.3907, + "mean_token_accuracy": 0.8687150478363037, + "num_tokens": 32714156.0, + "step": 842 + }, + { + "epoch": 0.10723826485180002, + "grad_norm": 3.023287773132324, + "learning_rate": 3.569309029249682e-07, + "loss": 0.4225, + "mean_token_accuracy": 0.8593257069587708, + "num_tokens": 32749270.0, + "step": 843 + }, + { + "epoch": 0.10736547513039053, + "grad_norm": 3.070145606994629, + "learning_rate": 3.573548113607461e-07, + "loss": 0.4186, + "mean_token_accuracy": 0.8585644960403442, + "num_tokens": 32786234.0, + "step": 844 + }, + { + "epoch": 0.10749268540898105, + "grad_norm": 2.869621515274048, + "learning_rate": 3.577787197965239e-07, + "loss": 0.4073, + "mean_token_accuracy": 0.8644293546676636, + "num_tokens": 32823728.0, + "step": 845 + }, + { + "epoch": 0.10761989568757156, + "grad_norm": 2.6133511066436768, + "learning_rate": 3.5820262823230177e-07, + "loss": 0.4058, + "mean_token_accuracy": 0.8636163473129272, + "num_tokens": 32864703.0, + "step": 846 + }, + { + "epoch": 0.10774710596616206, + "grad_norm": 3.228503465652466, + "learning_rate": 3.5862653666807967e-07, + "loss": 0.4576, + "mean_token_accuracy": 0.8473770022392273, + "num_tokens": 32896167.0, + "step": 847 + }, + { + "epoch": 0.10787431624475258, + "grad_norm": 3.5299758911132812, + "learning_rate": 3.5905044510385757e-07, + "loss": 0.3991, + "mean_token_accuracy": 0.8683265447616577, + "num_tokens": 32932710.0, + "step": 848 + }, + { + "epoch": 0.10800152652334309, + "grad_norm": 3.0840723514556885, + "learning_rate": 3.594743535396354e-07, + "loss": 0.4074, + "mean_token_accuracy": 0.8632700443267822, + "num_tokens": 32966117.0, + "step": 849 + }, + { + "epoch": 0.10812873680193359, + "grad_norm": 2.8371598720550537, + "learning_rate": 3.5989826197541326e-07, + "loss": 0.3592, + "mean_token_accuracy": 0.880911111831665, + "num_tokens": 33002155.0, + "step": 850 + }, + { + "epoch": 0.1082559470805241, + "grad_norm": 2.7750582695007324, + "learning_rate": 3.6032217041119116e-07, + "loss": 0.435, + "mean_token_accuracy": 0.8566617965698242, + "num_tokens": 33043698.0, + "step": 851 + }, + { + "epoch": 0.10838315735911462, + "grad_norm": 2.73366641998291, + "learning_rate": 3.6074607884696906e-07, + "loss": 0.39, + "mean_token_accuracy": 0.8678088188171387, + "num_tokens": 33081454.0, + "step": 852 + }, + { + "epoch": 0.10851036763770512, + "grad_norm": 2.937621593475342, + "learning_rate": 3.611699872827469e-07, + "loss": 0.4338, + "mean_token_accuracy": 0.857989490032196, + "num_tokens": 33118965.0, + "step": 853 + }, + { + "epoch": 0.10863757791629564, + "grad_norm": 2.8612263202667236, + "learning_rate": 3.6159389571852475e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8641276359558105, + "num_tokens": 33155441.0, + "step": 854 + }, + { + "epoch": 0.10876478819488615, + "grad_norm": 2.664900302886963, + "learning_rate": 3.6201780415430265e-07, + "loss": 0.4199, + "mean_token_accuracy": 0.8594881892204285, + "num_tokens": 33201507.0, + "step": 855 + }, + { + "epoch": 0.10889199847347665, + "grad_norm": 2.9109654426574707, + "learning_rate": 3.6244171259008055e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8630119562149048, + "num_tokens": 33236878.0, + "step": 856 + }, + { + "epoch": 0.10901920875206716, + "grad_norm": 2.821553945541382, + "learning_rate": 3.628656210258584e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8611323833465576, + "num_tokens": 33274001.0, + "step": 857 + }, + { + "epoch": 0.10914641903065768, + "grad_norm": 3.1077868938446045, + "learning_rate": 3.6328952946163624e-07, + "loss": 0.467, + "mean_token_accuracy": 0.8444423675537109, + "num_tokens": 33312504.0, + "step": 858 + }, + { + "epoch": 0.1092736293092482, + "grad_norm": 2.916288137435913, + "learning_rate": 3.6371343789741414e-07, + "loss": 0.4083, + "mean_token_accuracy": 0.8638054728507996, + "num_tokens": 33347544.0, + "step": 859 + }, + { + "epoch": 0.1094008395878387, + "grad_norm": 2.792921781539917, + "learning_rate": 3.6413734633319204e-07, + "loss": 0.3931, + "mean_token_accuracy": 0.8702471256256104, + "num_tokens": 33386116.0, + "step": 860 + }, + { + "epoch": 0.10952804986642921, + "grad_norm": 2.7738184928894043, + "learning_rate": 3.645612547689699e-07, + "loss": 0.4204, + "mean_token_accuracy": 0.8586231470108032, + "num_tokens": 33425505.0, + "step": 861 + }, + { + "epoch": 0.10965526014501972, + "grad_norm": 2.740318536758423, + "learning_rate": 3.6498516320474773e-07, + "loss": 0.3999, + "mean_token_accuracy": 0.8685191869735718, + "num_tokens": 33465887.0, + "step": 862 + }, + { + "epoch": 0.10978247042361022, + "grad_norm": 2.516329288482666, + "learning_rate": 3.6540907164052563e-07, + "loss": 0.3868, + "mean_token_accuracy": 0.8687772750854492, + "num_tokens": 33510412.0, + "step": 863 + }, + { + "epoch": 0.10990968070220074, + "grad_norm": 3.087984561920166, + "learning_rate": 3.658329800763035e-07, + "loss": 0.3825, + "mean_token_accuracy": 0.8701563477516174, + "num_tokens": 33542358.0, + "step": 864 + }, + { + "epoch": 0.11003689098079125, + "grad_norm": 2.722395181655884, + "learning_rate": 3.662568885120814e-07, + "loss": 0.4352, + "mean_token_accuracy": 0.8543053865432739, + "num_tokens": 33586295.0, + "step": 865 + }, + { + "epoch": 0.11016410125938175, + "grad_norm": 3.014500856399536, + "learning_rate": 3.666807969478592e-07, + "loss": 0.4825, + "mean_token_accuracy": 0.841713547706604, + "num_tokens": 33627393.0, + "step": 866 + }, + { + "epoch": 0.11029131153797227, + "grad_norm": 3.1008200645446777, + "learning_rate": 3.671047053836371e-07, + "loss": 0.4522, + "mean_token_accuracy": 0.8484143018722534, + "num_tokens": 33662859.0, + "step": 867 + }, + { + "epoch": 0.11041852181656278, + "grad_norm": 2.9590351581573486, + "learning_rate": 3.6752861381941497e-07, + "loss": 0.4488, + "mean_token_accuracy": 0.8486953973770142, + "num_tokens": 33699648.0, + "step": 868 + }, + { + "epoch": 0.11054573209515328, + "grad_norm": 2.906020402908325, + "learning_rate": 3.6795252225519287e-07, + "loss": 0.4668, + "mean_token_accuracy": 0.8402621746063232, + "num_tokens": 33736701.0, + "step": 869 + }, + { + "epoch": 0.1106729423737438, + "grad_norm": 2.860661506652832, + "learning_rate": 3.6837643069097077e-07, + "loss": 0.4475, + "mean_token_accuracy": 0.8513143062591553, + "num_tokens": 33778609.0, + "step": 870 + }, + { + "epoch": 0.11080015265233431, + "grad_norm": 3.1513547897338867, + "learning_rate": 3.688003391267486e-07, + "loss": 0.4767, + "mean_token_accuracy": 0.8431017398834229, + "num_tokens": 33812387.0, + "step": 871 + }, + { + "epoch": 0.11092736293092482, + "grad_norm": 2.909362554550171, + "learning_rate": 3.6922424756252646e-07, + "loss": 0.4269, + "mean_token_accuracy": 0.8579689264297485, + "num_tokens": 33850519.0, + "step": 872 + }, + { + "epoch": 0.11105457320951533, + "grad_norm": 2.817857265472412, + "learning_rate": 3.6964815599830436e-07, + "loss": 0.4186, + "mean_token_accuracy": 0.8610883355140686, + "num_tokens": 33893141.0, + "step": 873 + }, + { + "epoch": 0.11118178348810584, + "grad_norm": 2.942392110824585, + "learning_rate": 3.7007206443408226e-07, + "loss": 0.4, + "mean_token_accuracy": 0.8625423908233643, + "num_tokens": 33929247.0, + "step": 874 + }, + { + "epoch": 0.11130899376669635, + "grad_norm": 2.695087194442749, + "learning_rate": 3.704959728698601e-07, + "loss": 0.4355, + "mean_token_accuracy": 0.8559930324554443, + "num_tokens": 33976463.0, + "step": 875 + }, + { + "epoch": 0.11143620404528685, + "grad_norm": 2.9087507724761963, + "learning_rate": 3.7091988130563795e-07, + "loss": 0.4087, + "mean_token_accuracy": 0.8619875907897949, + "num_tokens": 34014324.0, + "step": 876 + }, + { + "epoch": 0.11156341432387737, + "grad_norm": 2.864060163497925, + "learning_rate": 3.7134378974141585e-07, + "loss": 0.4124, + "mean_token_accuracy": 0.8652048110961914, + "num_tokens": 34055462.0, + "step": 877 + }, + { + "epoch": 0.11169062460246788, + "grad_norm": 3.1816041469573975, + "learning_rate": 3.7176769817719375e-07, + "loss": 0.3784, + "mean_token_accuracy": 0.8735336065292358, + "num_tokens": 34084175.0, + "step": 878 + }, + { + "epoch": 0.11181783488105838, + "grad_norm": 2.831693649291992, + "learning_rate": 3.7219160661297154e-07, + "loss": 0.4387, + "mean_token_accuracy": 0.8552631139755249, + "num_tokens": 34126253.0, + "step": 879 + }, + { + "epoch": 0.1119450451596489, + "grad_norm": 2.964442729949951, + "learning_rate": 3.7261551504874944e-07, + "loss": 0.3824, + "mean_token_accuracy": 0.8741121888160706, + "num_tokens": 34159965.0, + "step": 880 + }, + { + "epoch": 0.11207225543823941, + "grad_norm": 2.759265422821045, + "learning_rate": 3.7303942348452734e-07, + "loss": 0.4138, + "mean_token_accuracy": 0.8656080961227417, + "num_tokens": 34201227.0, + "step": 881 + }, + { + "epoch": 0.11219946571682991, + "grad_norm": 3.4059319496154785, + "learning_rate": 3.7346333192030524e-07, + "loss": 0.4988, + "mean_token_accuracy": 0.8367485404014587, + "num_tokens": 34235076.0, + "step": 882 + }, + { + "epoch": 0.11232667599542043, + "grad_norm": 2.9914345741271973, + "learning_rate": 3.7388724035608303e-07, + "loss": 0.4347, + "mean_token_accuracy": 0.8542037010192871, + "num_tokens": 34270045.0, + "step": 883 + }, + { + "epoch": 0.11245388627401094, + "grad_norm": 2.7287542819976807, + "learning_rate": 3.7431114879186093e-07, + "loss": 0.4221, + "mean_token_accuracy": 0.8593251705169678, + "num_tokens": 34313419.0, + "step": 884 + }, + { + "epoch": 0.11258109655260146, + "grad_norm": 2.8630781173706055, + "learning_rate": 3.7473505722763883e-07, + "loss": 0.4298, + "mean_token_accuracy": 0.854006826877594, + "num_tokens": 34350911.0, + "step": 885 + }, + { + "epoch": 0.11270830683119196, + "grad_norm": 3.118273973464966, + "learning_rate": 3.7515896566341673e-07, + "loss": 0.4087, + "mean_token_accuracy": 0.8631477355957031, + "num_tokens": 34382468.0, + "step": 886 + }, + { + "epoch": 0.11283551710978247, + "grad_norm": 2.781517267227173, + "learning_rate": 3.755828740991945e-07, + "loss": 0.3968, + "mean_token_accuracy": 0.8629004955291748, + "num_tokens": 34419546.0, + "step": 887 + }, + { + "epoch": 0.11296272738837299, + "grad_norm": 2.7671127319335938, + "learning_rate": 3.760067825349724e-07, + "loss": 0.4474, + "mean_token_accuracy": 0.8504451513290405, + "num_tokens": 34462793.0, + "step": 888 + }, + { + "epoch": 0.11308993766696349, + "grad_norm": 2.8714962005615234, + "learning_rate": 3.764306909707503e-07, + "loss": 0.4147, + "mean_token_accuracy": 0.8607999682426453, + "num_tokens": 34499630.0, + "step": 889 + }, + { + "epoch": 0.113217147945554, + "grad_norm": 2.8535420894622803, + "learning_rate": 3.768545994065282e-07, + "loss": 0.4339, + "mean_token_accuracy": 0.8552308678627014, + "num_tokens": 34539177.0, + "step": 890 + }, + { + "epoch": 0.11334435822414451, + "grad_norm": 2.685021162033081, + "learning_rate": 3.77278507842306e-07, + "loss": 0.4195, + "mean_token_accuracy": 0.8608841896057129, + "num_tokens": 34580193.0, + "step": 891 + }, + { + "epoch": 0.11347156850273502, + "grad_norm": 2.831564426422119, + "learning_rate": 3.777024162780839e-07, + "loss": 0.4177, + "mean_token_accuracy": 0.8626022338867188, + "num_tokens": 34619847.0, + "step": 892 + }, + { + "epoch": 0.11359877878132553, + "grad_norm": 3.2036352157592773, + "learning_rate": 3.781263247138618e-07, + "loss": 0.4116, + "mean_token_accuracy": 0.8641805648803711, + "num_tokens": 34654178.0, + "step": 893 + }, + { + "epoch": 0.11372598905991604, + "grad_norm": 2.730647563934326, + "learning_rate": 3.785502331496397e-07, + "loss": 0.4382, + "mean_token_accuracy": 0.8555030226707458, + "num_tokens": 34697527.0, + "step": 894 + }, + { + "epoch": 0.11385319933850654, + "grad_norm": 2.861483335494995, + "learning_rate": 3.789741415854175e-07, + "loss": 0.4588, + "mean_token_accuracy": 0.846812903881073, + "num_tokens": 34736948.0, + "step": 895 + }, + { + "epoch": 0.11398040961709706, + "grad_norm": 2.6261537075042725, + "learning_rate": 3.793980500211954e-07, + "loss": 0.4121, + "mean_token_accuracy": 0.860377848148346, + "num_tokens": 34782047.0, + "step": 896 + }, + { + "epoch": 0.11410761989568757, + "grad_norm": 2.857239007949829, + "learning_rate": 3.798219584569733e-07, + "loss": 0.3748, + "mean_token_accuracy": 0.8691748380661011, + "num_tokens": 34817275.0, + "step": 897 + }, + { + "epoch": 0.11423483017427809, + "grad_norm": 3.034956932067871, + "learning_rate": 3.8024586689275115e-07, + "loss": 0.4126, + "mean_token_accuracy": 0.864865243434906, + "num_tokens": 34851101.0, + "step": 898 + }, + { + "epoch": 0.11436204045286859, + "grad_norm": 2.6609950065612793, + "learning_rate": 3.80669775328529e-07, + "loss": 0.4051, + "mean_token_accuracy": 0.8657782077789307, + "num_tokens": 34891566.0, + "step": 899 + }, + { + "epoch": 0.1144892507314591, + "grad_norm": 2.748481035232544, + "learning_rate": 3.810936837643069e-07, + "loss": 0.4373, + "mean_token_accuracy": 0.8532437086105347, + "num_tokens": 34932394.0, + "step": 900 + }, + { + "epoch": 0.11461646101004962, + "grad_norm": 2.832509756088257, + "learning_rate": 3.815175922000848e-07, + "loss": 0.4466, + "mean_token_accuracy": 0.8542412519454956, + "num_tokens": 34972365.0, + "step": 901 + }, + { + "epoch": 0.11474367128864012, + "grad_norm": 3.0739243030548096, + "learning_rate": 3.8194150063586264e-07, + "loss": 0.4711, + "mean_token_accuracy": 0.8452808856964111, + "num_tokens": 35011878.0, + "step": 902 + }, + { + "epoch": 0.11487088156723063, + "grad_norm": 2.9096148014068604, + "learning_rate": 3.823654090716405e-07, + "loss": 0.4216, + "mean_token_accuracy": 0.8564046025276184, + "num_tokens": 35048659.0, + "step": 903 + }, + { + "epoch": 0.11499809184582115, + "grad_norm": 3.074805736541748, + "learning_rate": 3.827893175074184e-07, + "loss": 0.4614, + "mean_token_accuracy": 0.8521406650543213, + "num_tokens": 35084635.0, + "step": 904 + }, + { + "epoch": 0.11512530212441165, + "grad_norm": 3.1223068237304688, + "learning_rate": 3.832132259431963e-07, + "loss": 0.4115, + "mean_token_accuracy": 0.8594790697097778, + "num_tokens": 35116105.0, + "step": 905 + }, + { + "epoch": 0.11525251240300216, + "grad_norm": 3.103553056716919, + "learning_rate": 3.8363713437897413e-07, + "loss": 0.4484, + "mean_token_accuracy": 0.8463575839996338, + "num_tokens": 35152956.0, + "step": 906 + }, + { + "epoch": 0.11537972268159268, + "grad_norm": 2.617396116256714, + "learning_rate": 3.8406104281475197e-07, + "loss": 0.4121, + "mean_token_accuracy": 0.8603935837745667, + "num_tokens": 35196216.0, + "step": 907 + }, + { + "epoch": 0.11550693296018319, + "grad_norm": 2.7527856826782227, + "learning_rate": 3.8448495125052987e-07, + "loss": 0.4049, + "mean_token_accuracy": 0.8640074729919434, + "num_tokens": 35236538.0, + "step": 908 + }, + { + "epoch": 0.11563414323877369, + "grad_norm": 2.9836480617523193, + "learning_rate": 3.8490885968630777e-07, + "loss": 0.4635, + "mean_token_accuracy": 0.8482269048690796, + "num_tokens": 35274151.0, + "step": 909 + }, + { + "epoch": 0.1157613535173642, + "grad_norm": 3.1469128131866455, + "learning_rate": 3.853327681220856e-07, + "loss": 0.426, + "mean_token_accuracy": 0.8555285334587097, + "num_tokens": 35310077.0, + "step": 910 + }, + { + "epoch": 0.11588856379595472, + "grad_norm": 2.9095399379730225, + "learning_rate": 3.8575667655786346e-07, + "loss": 0.4891, + "mean_token_accuracy": 0.8365516662597656, + "num_tokens": 35351559.0, + "step": 911 + }, + { + "epoch": 0.11601577407454522, + "grad_norm": 2.6329879760742188, + "learning_rate": 3.8618058499364136e-07, + "loss": 0.3962, + "mean_token_accuracy": 0.86997389793396, + "num_tokens": 35395387.0, + "step": 912 + }, + { + "epoch": 0.11614298435313573, + "grad_norm": 2.9616506099700928, + "learning_rate": 3.8660449342941926e-07, + "loss": 0.3889, + "mean_token_accuracy": 0.8679407835006714, + "num_tokens": 35428709.0, + "step": 913 + }, + { + "epoch": 0.11627019463172625, + "grad_norm": 3.0347213745117188, + "learning_rate": 3.870284018651971e-07, + "loss": 0.423, + "mean_token_accuracy": 0.8549226522445679, + "num_tokens": 35465516.0, + "step": 914 + }, + { + "epoch": 0.11639740491031675, + "grad_norm": 2.6888318061828613, + "learning_rate": 3.8745231030097495e-07, + "loss": 0.3838, + "mean_token_accuracy": 0.8692813515663147, + "num_tokens": 35505885.0, + "step": 915 + }, + { + "epoch": 0.11652461518890726, + "grad_norm": 3.0358307361602783, + "learning_rate": 3.8787621873675285e-07, + "loss": 0.4224, + "mean_token_accuracy": 0.8601640462875366, + "num_tokens": 35543919.0, + "step": 916 + }, + { + "epoch": 0.11665182546749778, + "grad_norm": 3.3654420375823975, + "learning_rate": 3.883001271725307e-07, + "loss": 0.4246, + "mean_token_accuracy": 0.8597571849822998, + "num_tokens": 35576050.0, + "step": 917 + }, + { + "epoch": 0.11677903574608828, + "grad_norm": 3.096879005432129, + "learning_rate": 3.887240356083086e-07, + "loss": 0.4167, + "mean_token_accuracy": 0.8551962375640869, + "num_tokens": 35609178.0, + "step": 918 + }, + { + "epoch": 0.11690624602467879, + "grad_norm": 2.6905479431152344, + "learning_rate": 3.8914794404408644e-07, + "loss": 0.4339, + "mean_token_accuracy": 0.8501272201538086, + "num_tokens": 35652899.0, + "step": 919 + }, + { + "epoch": 0.1170334563032693, + "grad_norm": 3.054121732711792, + "learning_rate": 3.8957185247986434e-07, + "loss": 0.4656, + "mean_token_accuracy": 0.8489157557487488, + "num_tokens": 35689544.0, + "step": 920 + }, + { + "epoch": 0.11716066658185982, + "grad_norm": 2.828920602798462, + "learning_rate": 3.899957609156422e-07, + "loss": 0.419, + "mean_token_accuracy": 0.8578213453292847, + "num_tokens": 35728999.0, + "step": 921 + }, + { + "epoch": 0.11728787686045032, + "grad_norm": 3.0648438930511475, + "learning_rate": 3.904196693514201e-07, + "loss": 0.3886, + "mean_token_accuracy": 0.868686318397522, + "num_tokens": 35766758.0, + "step": 922 + }, + { + "epoch": 0.11741508713904084, + "grad_norm": 2.764962673187256, + "learning_rate": 3.9084357778719793e-07, + "loss": 0.4078, + "mean_token_accuracy": 0.8590884208679199, + "num_tokens": 35807188.0, + "step": 923 + }, + { + "epoch": 0.11754229741763135, + "grad_norm": 3.0529794692993164, + "learning_rate": 3.9126748622297583e-07, + "loss": 0.3944, + "mean_token_accuracy": 0.8644976615905762, + "num_tokens": 35842332.0, + "step": 924 + }, + { + "epoch": 0.11766950769622185, + "grad_norm": 2.6799728870391846, + "learning_rate": 3.916913946587537e-07, + "loss": 0.4149, + "mean_token_accuracy": 0.8596159219741821, + "num_tokens": 35886480.0, + "step": 925 + }, + { + "epoch": 0.11779671797481236, + "grad_norm": 2.775979995727539, + "learning_rate": 3.921153030945316e-07, + "loss": 0.4049, + "mean_token_accuracy": 0.8639020919799805, + "num_tokens": 35925604.0, + "step": 926 + }, + { + "epoch": 0.11792392825340288, + "grad_norm": 2.6497883796691895, + "learning_rate": 3.925392115303094e-07, + "loss": 0.3838, + "mean_token_accuracy": 0.8707143068313599, + "num_tokens": 35966328.0, + "step": 927 + }, + { + "epoch": 0.11805113853199338, + "grad_norm": 2.9090640544891357, + "learning_rate": 3.929631199660873e-07, + "loss": 0.4061, + "mean_token_accuracy": 0.8637390732765198, + "num_tokens": 36003901.0, + "step": 928 + }, + { + "epoch": 0.1181783488105839, + "grad_norm": 2.782412052154541, + "learning_rate": 3.9338702840186517e-07, + "loss": 0.4014, + "mean_token_accuracy": 0.8654537796974182, + "num_tokens": 36045800.0, + "step": 929 + }, + { + "epoch": 0.11830555908917441, + "grad_norm": 3.008913993835449, + "learning_rate": 3.9381093683764307e-07, + "loss": 0.4451, + "mean_token_accuracy": 0.8502379059791565, + "num_tokens": 36081828.0, + "step": 930 + }, + { + "epoch": 0.11843276936776491, + "grad_norm": 2.952564001083374, + "learning_rate": 3.942348452734209e-07, + "loss": 0.398, + "mean_token_accuracy": 0.8651202917098999, + "num_tokens": 36116886.0, + "step": 931 + }, + { + "epoch": 0.11855997964635542, + "grad_norm": 3.0109000205993652, + "learning_rate": 3.946587537091988e-07, + "loss": 0.4181, + "mean_token_accuracy": 0.8615788221359253, + "num_tokens": 36153698.0, + "step": 932 + }, + { + "epoch": 0.11868718992494594, + "grad_norm": 2.730973482131958, + "learning_rate": 3.9508266214497666e-07, + "loss": 0.431, + "mean_token_accuracy": 0.8550617098808289, + "num_tokens": 36195135.0, + "step": 933 + }, + { + "epoch": 0.11881440020353645, + "grad_norm": 3.219263792037964, + "learning_rate": 3.9550657058075456e-07, + "loss": 0.3921, + "mean_token_accuracy": 0.8686977624893188, + "num_tokens": 36229509.0, + "step": 934 + }, + { + "epoch": 0.11894161048212695, + "grad_norm": 3.0496695041656494, + "learning_rate": 3.959304790165324e-07, + "loss": 0.4247, + "mean_token_accuracy": 0.8567042350769043, + "num_tokens": 36263325.0, + "step": 935 + }, + { + "epoch": 0.11906882076071747, + "grad_norm": 3.1111598014831543, + "learning_rate": 3.9635438745231025e-07, + "loss": 0.4687, + "mean_token_accuracy": 0.8428953886032104, + "num_tokens": 36298667.0, + "step": 936 + }, + { + "epoch": 0.11919603103930798, + "grad_norm": 2.761941432952881, + "learning_rate": 3.9677829588808815e-07, + "loss": 0.4095, + "mean_token_accuracy": 0.8603213429450989, + "num_tokens": 36339568.0, + "step": 937 + }, + { + "epoch": 0.11932324131789848, + "grad_norm": 2.7837727069854736, + "learning_rate": 3.9720220432386605e-07, + "loss": 0.3961, + "mean_token_accuracy": 0.8651134967803955, + "num_tokens": 36379988.0, + "step": 938 + }, + { + "epoch": 0.119450451596489, + "grad_norm": 3.121720314025879, + "learning_rate": 3.976261127596439e-07, + "loss": 0.4737, + "mean_token_accuracy": 0.8440864682197571, + "num_tokens": 36414937.0, + "step": 939 + }, + { + "epoch": 0.11957766187507951, + "grad_norm": 3.035221815109253, + "learning_rate": 3.9805002119542174e-07, + "loss": 0.4538, + "mean_token_accuracy": 0.8476459383964539, + "num_tokens": 36450893.0, + "step": 940 + }, + { + "epoch": 0.11970487215367001, + "grad_norm": 2.8535850048065186, + "learning_rate": 3.9847392963119964e-07, + "loss": 0.4506, + "mean_token_accuracy": 0.8550583720207214, + "num_tokens": 36489524.0, + "step": 941 + }, + { + "epoch": 0.11983208243226053, + "grad_norm": 3.118825912475586, + "learning_rate": 3.9889783806697754e-07, + "loss": 0.474, + "mean_token_accuracy": 0.8453030586242676, + "num_tokens": 36527716.0, + "step": 942 + }, + { + "epoch": 0.11995929271085104, + "grad_norm": 2.805999279022217, + "learning_rate": 3.993217465027554e-07, + "loss": 0.4299, + "mean_token_accuracy": 0.8575674295425415, + "num_tokens": 36566828.0, + "step": 943 + }, + { + "epoch": 0.12008650298944154, + "grad_norm": 2.7045509815216064, + "learning_rate": 3.9974565493853323e-07, + "loss": 0.4791, + "mean_token_accuracy": 0.8400170207023621, + "num_tokens": 36613105.0, + "step": 944 + }, + { + "epoch": 0.12021371326803205, + "grad_norm": 3.0776867866516113, + "learning_rate": 4.0016956337431113e-07, + "loss": 0.4058, + "mean_token_accuracy": 0.8645336031913757, + "num_tokens": 36648425.0, + "step": 945 + }, + { + "epoch": 0.12034092354662257, + "grad_norm": 2.8314547538757324, + "learning_rate": 4.0059347181008903e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8582897186279297, + "num_tokens": 36685476.0, + "step": 946 + }, + { + "epoch": 0.12046813382521308, + "grad_norm": 2.6980438232421875, + "learning_rate": 4.010173802458669e-07, + "loss": 0.4503, + "mean_token_accuracy": 0.8500725030899048, + "num_tokens": 36730392.0, + "step": 947 + }, + { + "epoch": 0.12059534410380358, + "grad_norm": 2.934630870819092, + "learning_rate": 4.014412886816447e-07, + "loss": 0.435, + "mean_token_accuracy": 0.8553253412246704, + "num_tokens": 36767823.0, + "step": 948 + }, + { + "epoch": 0.1207225543823941, + "grad_norm": 2.671346426010132, + "learning_rate": 4.018651971174226e-07, + "loss": 0.3901, + "mean_token_accuracy": 0.8681510090827942, + "num_tokens": 36806787.0, + "step": 949 + }, + { + "epoch": 0.12084976466098461, + "grad_norm": 2.843898057937622, + "learning_rate": 4.022891055532005e-07, + "loss": 0.4301, + "mean_token_accuracy": 0.8576736450195312, + "num_tokens": 36846552.0, + "step": 950 + }, + { + "epoch": 0.12097697493957511, + "grad_norm": 2.6667070388793945, + "learning_rate": 4.0271301398897837e-07, + "loss": 0.399, + "mean_token_accuracy": 0.8665982484817505, + "num_tokens": 36889198.0, + "step": 951 + }, + { + "epoch": 0.12110418521816563, + "grad_norm": 3.080505132675171, + "learning_rate": 4.031369224247562e-07, + "loss": 0.4395, + "mean_token_accuracy": 0.854210376739502, + "num_tokens": 36922166.0, + "step": 952 + }, + { + "epoch": 0.12123139549675614, + "grad_norm": 2.7030646800994873, + "learning_rate": 4.035608308605341e-07, + "loss": 0.4279, + "mean_token_accuracy": 0.8562039136886597, + "num_tokens": 36966403.0, + "step": 953 + }, + { + "epoch": 0.12135860577534664, + "grad_norm": 2.6012346744537354, + "learning_rate": 4.03984739296312e-07, + "loss": 0.3834, + "mean_token_accuracy": 0.8733225464820862, + "num_tokens": 37009615.0, + "step": 954 + }, + { + "epoch": 0.12148581605393716, + "grad_norm": 2.8571224212646484, + "learning_rate": 4.044086477320898e-07, + "loss": 0.4081, + "mean_token_accuracy": 0.8626607656478882, + "num_tokens": 37044902.0, + "step": 955 + }, + { + "epoch": 0.12161302633252767, + "grad_norm": 2.983691453933716, + "learning_rate": 4.048325561678677e-07, + "loss": 0.4661, + "mean_token_accuracy": 0.842439591884613, + "num_tokens": 37083580.0, + "step": 956 + }, + { + "epoch": 0.12174023661111817, + "grad_norm": 2.813829183578491, + "learning_rate": 4.052564646036456e-07, + "loss": 0.3876, + "mean_token_accuracy": 0.8717062473297119, + "num_tokens": 37119943.0, + "step": 957 + }, + { + "epoch": 0.12186744688970869, + "grad_norm": 2.6236419677734375, + "learning_rate": 4.056803730394235e-07, + "loss": 0.3984, + "mean_token_accuracy": 0.8657732009887695, + "num_tokens": 37161179.0, + "step": 958 + }, + { + "epoch": 0.1219946571682992, + "grad_norm": 2.779665470123291, + "learning_rate": 4.061042814752013e-07, + "loss": 0.4364, + "mean_token_accuracy": 0.8562721014022827, + "num_tokens": 37200765.0, + "step": 959 + }, + { + "epoch": 0.12212186744688971, + "grad_norm": 2.71427845954895, + "learning_rate": 4.065281899109792e-07, + "loss": 0.4054, + "mean_token_accuracy": 0.8619648218154907, + "num_tokens": 37240700.0, + "step": 960 + }, + { + "epoch": 0.12224907772548022, + "grad_norm": 2.9521195888519287, + "learning_rate": 4.069520983467571e-07, + "loss": 0.4631, + "mean_token_accuracy": 0.8440587520599365, + "num_tokens": 37282756.0, + "step": 961 + }, + { + "epoch": 0.12237628800407073, + "grad_norm": 2.5614185333251953, + "learning_rate": 4.07376006782535e-07, + "loss": 0.3973, + "mean_token_accuracy": 0.8673103451728821, + "num_tokens": 37326843.0, + "step": 962 + }, + { + "epoch": 0.12250349828266124, + "grad_norm": 2.6043052673339844, + "learning_rate": 4.077999152183128e-07, + "loss": 0.4291, + "mean_token_accuracy": 0.8540984392166138, + "num_tokens": 37370732.0, + "step": 963 + }, + { + "epoch": 0.12263070856125174, + "grad_norm": 2.7569191455841064, + "learning_rate": 4.082238236540907e-07, + "loss": 0.4107, + "mean_token_accuracy": 0.8652428388595581, + "num_tokens": 37412296.0, + "step": 964 + }, + { + "epoch": 0.12275791883984226, + "grad_norm": 2.8863906860351562, + "learning_rate": 4.086477320898686e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8610423803329468, + "num_tokens": 37446881.0, + "step": 965 + }, + { + "epoch": 0.12288512911843277, + "grad_norm": 2.648444414138794, + "learning_rate": 4.090716405256465e-07, + "loss": 0.3589, + "mean_token_accuracy": 0.8795431852340698, + "num_tokens": 37485692.0, + "step": 966 + }, + { + "epoch": 0.12301233939702327, + "grad_norm": 2.7935478687286377, + "learning_rate": 4.094955489614243e-07, + "loss": 0.4098, + "mean_token_accuracy": 0.8632317781448364, + "num_tokens": 37524582.0, + "step": 967 + }, + { + "epoch": 0.12313954967561379, + "grad_norm": 3.224567174911499, + "learning_rate": 4.099194573972022e-07, + "loss": 0.4494, + "mean_token_accuracy": 0.8509343862533569, + "num_tokens": 37556312.0, + "step": 968 + }, + { + "epoch": 0.1232667599542043, + "grad_norm": 2.9251651763916016, + "learning_rate": 4.1034336583298007e-07, + "loss": 0.4025, + "mean_token_accuracy": 0.8630812764167786, + "num_tokens": 37594730.0, + "step": 969 + }, + { + "epoch": 0.1233939702327948, + "grad_norm": 2.5262837409973145, + "learning_rate": 4.1076727426875797e-07, + "loss": 0.397, + "mean_token_accuracy": 0.8680325746536255, + "num_tokens": 37639354.0, + "step": 970 + }, + { + "epoch": 0.12352118051138532, + "grad_norm": 2.888690233230591, + "learning_rate": 4.1119118270453577e-07, + "loss": 0.4, + "mean_token_accuracy": 0.8667482137680054, + "num_tokens": 37677121.0, + "step": 971 + }, + { + "epoch": 0.12364839078997583, + "grad_norm": 2.9878110885620117, + "learning_rate": 4.1161509114031366e-07, + "loss": 0.3668, + "mean_token_accuracy": 0.8735477328300476, + "num_tokens": 37708558.0, + "step": 972 + }, + { + "epoch": 0.12377560106856635, + "grad_norm": 3.0185821056365967, + "learning_rate": 4.1203899957609156e-07, + "loss": 0.4115, + "mean_token_accuracy": 0.8610020875930786, + "num_tokens": 37741395.0, + "step": 973 + }, + { + "epoch": 0.12390281134715685, + "grad_norm": 2.971496820449829, + "learning_rate": 4.124629080118694e-07, + "loss": 0.376, + "mean_token_accuracy": 0.8699425458908081, + "num_tokens": 37773341.0, + "step": 974 + }, + { + "epoch": 0.12403002162574736, + "grad_norm": 2.5999341011047363, + "learning_rate": 4.1288681644764726e-07, + "loss": 0.4241, + "mean_token_accuracy": 0.8558759689331055, + "num_tokens": 37818227.0, + "step": 975 + }, + { + "epoch": 0.12415723190433788, + "grad_norm": 2.7880361080169678, + "learning_rate": 4.1331072488342515e-07, + "loss": 0.4461, + "mean_token_accuracy": 0.8515418767929077, + "num_tokens": 37857338.0, + "step": 976 + }, + { + "epoch": 0.12428444218292838, + "grad_norm": 2.8820345401763916, + "learning_rate": 4.1373463331920305e-07, + "loss": 0.3867, + "mean_token_accuracy": 0.8683695197105408, + "num_tokens": 37893504.0, + "step": 977 + }, + { + "epoch": 0.12441165246151889, + "grad_norm": 2.8002612590789795, + "learning_rate": 4.141585417549809e-07, + "loss": 0.4363, + "mean_token_accuracy": 0.8501102328300476, + "num_tokens": 37933994.0, + "step": 978 + }, + { + "epoch": 0.1245388627401094, + "grad_norm": 3.185739040374756, + "learning_rate": 4.1458245019075875e-07, + "loss": 0.5025, + "mean_token_accuracy": 0.8346392512321472, + "num_tokens": 37968934.0, + "step": 979 + }, + { + "epoch": 0.1246660730186999, + "grad_norm": 2.800450325012207, + "learning_rate": 4.1500635862653664e-07, + "loss": 0.3661, + "mean_token_accuracy": 0.8764479160308838, + "num_tokens": 38006322.0, + "step": 980 + }, + { + "epoch": 0.12479328329729042, + "grad_norm": 2.83242130279541, + "learning_rate": 4.1543026706231454e-07, + "loss": 0.4161, + "mean_token_accuracy": 0.8588470220565796, + "num_tokens": 38045641.0, + "step": 981 + }, + { + "epoch": 0.12492049357588093, + "grad_norm": 2.935861825942993, + "learning_rate": 4.158541754980924e-07, + "loss": 0.3877, + "mean_token_accuracy": 0.873420238494873, + "num_tokens": 38080796.0, + "step": 982 + }, + { + "epoch": 0.12504770385447145, + "grad_norm": 2.962684154510498, + "learning_rate": 4.1627808393387024e-07, + "loss": 0.4278, + "mean_token_accuracy": 0.85638827085495, + "num_tokens": 38117427.0, + "step": 983 + }, + { + "epoch": 0.12517491413306195, + "grad_norm": 3.044066905975342, + "learning_rate": 4.1670199236964813e-07, + "loss": 0.4254, + "mean_token_accuracy": 0.8570389151573181, + "num_tokens": 38153726.0, + "step": 984 + }, + { + "epoch": 0.12530212441165245, + "grad_norm": 2.6287879943847656, + "learning_rate": 4.1712590080542603e-07, + "loss": 0.4327, + "mean_token_accuracy": 0.8525205850601196, + "num_tokens": 38201747.0, + "step": 985 + }, + { + "epoch": 0.12542933469024298, + "grad_norm": 2.6612589359283447, + "learning_rate": 4.175498092412039e-07, + "loss": 0.3883, + "mean_token_accuracy": 0.8701173067092896, + "num_tokens": 38242282.0, + "step": 986 + }, + { + "epoch": 0.12555654496883348, + "grad_norm": 2.789006233215332, + "learning_rate": 4.179737176769817e-07, + "loss": 0.4075, + "mean_token_accuracy": 0.8647491931915283, + "num_tokens": 38279318.0, + "step": 987 + }, + { + "epoch": 0.12568375524742398, + "grad_norm": 3.121138095855713, + "learning_rate": 4.183976261127596e-07, + "loss": 0.46, + "mean_token_accuracy": 0.8479369878768921, + "num_tokens": 38314055.0, + "step": 988 + }, + { + "epoch": 0.1258109655260145, + "grad_norm": 2.876959800720215, + "learning_rate": 4.1882153454853747e-07, + "loss": 0.4129, + "mean_token_accuracy": 0.8628233671188354, + "num_tokens": 38350769.0, + "step": 989 + }, + { + "epoch": 0.125938175804605, + "grad_norm": 2.7407732009887695, + "learning_rate": 4.1924544298431537e-07, + "loss": 0.4132, + "mean_token_accuracy": 0.8637378215789795, + "num_tokens": 38392263.0, + "step": 990 + }, + { + "epoch": 0.12606538608319554, + "grad_norm": 2.7115702629089355, + "learning_rate": 4.196693514200932e-07, + "loss": 0.3909, + "mean_token_accuracy": 0.8670095801353455, + "num_tokens": 38433753.0, + "step": 991 + }, + { + "epoch": 0.12619259636178604, + "grad_norm": 3.0207390785217285, + "learning_rate": 4.200932598558711e-07, + "loss": 0.4581, + "mean_token_accuracy": 0.8484587669372559, + "num_tokens": 38471818.0, + "step": 992 + }, + { + "epoch": 0.12631980664037654, + "grad_norm": 2.991777181625366, + "learning_rate": 4.2051716829164896e-07, + "loss": 0.4235, + "mean_token_accuracy": 0.8539257049560547, + "num_tokens": 38507957.0, + "step": 993 + }, + { + "epoch": 0.12644701691896706, + "grad_norm": 2.993981122970581, + "learning_rate": 4.2094107672742686e-07, + "loss": 0.4361, + "mean_token_accuracy": 0.8556779623031616, + "num_tokens": 38543683.0, + "step": 994 + }, + { + "epoch": 0.12657422719755757, + "grad_norm": 2.5954084396362305, + "learning_rate": 4.2136498516320476e-07, + "loss": 0.4221, + "mean_token_accuracy": 0.858239471912384, + "num_tokens": 38586793.0, + "step": 995 + }, + { + "epoch": 0.12670143747614807, + "grad_norm": 2.9232661724090576, + "learning_rate": 4.217888935989826e-07, + "loss": 0.4279, + "mean_token_accuracy": 0.8578855991363525, + "num_tokens": 38625243.0, + "step": 996 + }, + { + "epoch": 0.1268286477547386, + "grad_norm": 3.3243253231048584, + "learning_rate": 4.2221280203476045e-07, + "loss": 0.4037, + "mean_token_accuracy": 0.8614130020141602, + "num_tokens": 38653346.0, + "step": 997 + }, + { + "epoch": 0.1269558580333291, + "grad_norm": 3.1226601600646973, + "learning_rate": 4.2263671047053835e-07, + "loss": 0.4106, + "mean_token_accuracy": 0.8607900738716125, + "num_tokens": 38687645.0, + "step": 998 + }, + { + "epoch": 0.1270830683119196, + "grad_norm": 2.9050545692443848, + "learning_rate": 4.2306061890631625e-07, + "loss": 0.423, + "mean_token_accuracy": 0.8570535182952881, + "num_tokens": 38722711.0, + "step": 999 + }, + { + "epoch": 0.12721027859051012, + "grad_norm": 2.8355374336242676, + "learning_rate": 4.234845273420941e-07, + "loss": 0.4116, + "mean_token_accuracy": 0.8646612167358398, + "num_tokens": 38760910.0, + "step": 1000 + }, + { + "epoch": 0.12733748886910062, + "grad_norm": 2.8374338150024414, + "learning_rate": 4.2390843577787194e-07, + "loss": 0.4048, + "mean_token_accuracy": 0.8658711910247803, + "num_tokens": 38801118.0, + "step": 1001 + }, + { + "epoch": 0.12746469914769112, + "grad_norm": 3.004688024520874, + "learning_rate": 4.2433234421364984e-07, + "loss": 0.4527, + "mean_token_accuracy": 0.8463118672370911, + "num_tokens": 38840092.0, + "step": 1002 + }, + { + "epoch": 0.12759190942628165, + "grad_norm": 2.9007344245910645, + "learning_rate": 4.2475625264942774e-07, + "loss": 0.4454, + "mean_token_accuracy": 0.8514674305915833, + "num_tokens": 38883136.0, + "step": 1003 + }, + { + "epoch": 0.12771911970487215, + "grad_norm": 2.5930168628692627, + "learning_rate": 4.251801610852056e-07, + "loss": 0.3724, + "mean_token_accuracy": 0.872494101524353, + "num_tokens": 38928041.0, + "step": 1004 + }, + { + "epoch": 0.12784632998346265, + "grad_norm": 2.9195525646209717, + "learning_rate": 4.2560406952098343e-07, + "loss": 0.4588, + "mean_token_accuracy": 0.8478045463562012, + "num_tokens": 38966231.0, + "step": 1005 + }, + { + "epoch": 0.12797354026205318, + "grad_norm": 2.941213369369507, + "learning_rate": 4.2602797795676133e-07, + "loss": 0.4306, + "mean_token_accuracy": 0.8580199480056763, + "num_tokens": 39006881.0, + "step": 1006 + }, + { + "epoch": 0.12810075054064368, + "grad_norm": 2.749077558517456, + "learning_rate": 4.2645188639253923e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8721065521240234, + "num_tokens": 39050581.0, + "step": 1007 + }, + { + "epoch": 0.12822796081923418, + "grad_norm": 2.8547942638397217, + "learning_rate": 4.26875794828317e-07, + "loss": 0.4544, + "mean_token_accuracy": 0.849971354007721, + "num_tokens": 39096146.0, + "step": 1008 + }, + { + "epoch": 0.1283551710978247, + "grad_norm": 2.9684011936187744, + "learning_rate": 4.272997032640949e-07, + "loss": 0.3847, + "mean_token_accuracy": 0.8676964044570923, + "num_tokens": 39130538.0, + "step": 1009 + }, + { + "epoch": 0.1284823813764152, + "grad_norm": 2.9309003353118896, + "learning_rate": 4.277236116998728e-07, + "loss": 0.4455, + "mean_token_accuracy": 0.850055456161499, + "num_tokens": 39170934.0, + "step": 1010 + }, + { + "epoch": 0.1286095916550057, + "grad_norm": 2.64949107170105, + "learning_rate": 4.281475201356507e-07, + "loss": 0.4109, + "mean_token_accuracy": 0.8627166748046875, + "num_tokens": 39213466.0, + "step": 1011 + }, + { + "epoch": 0.12873680193359624, + "grad_norm": 3.0146965980529785, + "learning_rate": 4.285714285714285e-07, + "loss": 0.4025, + "mean_token_accuracy": 0.8638460636138916, + "num_tokens": 39249898.0, + "step": 1012 + }, + { + "epoch": 0.12886401221218674, + "grad_norm": 2.8163883686065674, + "learning_rate": 4.289953370072064e-07, + "loss": 0.4758, + "mean_token_accuracy": 0.8443858027458191, + "num_tokens": 39293071.0, + "step": 1013 + }, + { + "epoch": 0.12899122249077727, + "grad_norm": 2.836292028427124, + "learning_rate": 4.294192454429843e-07, + "loss": 0.4189, + "mean_token_accuracy": 0.8605660796165466, + "num_tokens": 39329275.0, + "step": 1014 + }, + { + "epoch": 0.12911843276936777, + "grad_norm": 3.0088951587677, + "learning_rate": 4.298431538787622e-07, + "loss": 0.4077, + "mean_token_accuracy": 0.8641385436058044, + "num_tokens": 39365449.0, + "step": 1015 + }, + { + "epoch": 0.12924564304795827, + "grad_norm": 3.0098893642425537, + "learning_rate": 4.3026706231454e-07, + "loss": 0.4224, + "mean_token_accuracy": 0.862053394317627, + "num_tokens": 39405408.0, + "step": 1016 + }, + { + "epoch": 0.1293728533265488, + "grad_norm": 2.9645135402679443, + "learning_rate": 4.306909707503179e-07, + "loss": 0.4057, + "mean_token_accuracy": 0.86491858959198, + "num_tokens": 39442789.0, + "step": 1017 + }, + { + "epoch": 0.1295000636051393, + "grad_norm": 2.841505527496338, + "learning_rate": 4.311148791860958e-07, + "loss": 0.4051, + "mean_token_accuracy": 0.8647568225860596, + "num_tokens": 39481407.0, + "step": 1018 + }, + { + "epoch": 0.1296272738837298, + "grad_norm": 2.788179874420166, + "learning_rate": 4.315387876218737e-07, + "loss": 0.3635, + "mean_token_accuracy": 0.8732691407203674, + "num_tokens": 39517175.0, + "step": 1019 + }, + { + "epoch": 0.12975448416232033, + "grad_norm": 3.1165523529052734, + "learning_rate": 4.319626960576515e-07, + "loss": 0.3976, + "mean_token_accuracy": 0.8616728186607361, + "num_tokens": 39549384.0, + "step": 1020 + }, + { + "epoch": 0.12988169444091083, + "grad_norm": 2.883955478668213, + "learning_rate": 4.323866044934294e-07, + "loss": 0.4168, + "mean_token_accuracy": 0.8574845790863037, + "num_tokens": 39585571.0, + "step": 1021 + }, + { + "epoch": 0.13000890471950133, + "grad_norm": 2.6111786365509033, + "learning_rate": 4.328105129292073e-07, + "loss": 0.4433, + "mean_token_accuracy": 0.8494824171066284, + "num_tokens": 39630871.0, + "step": 1022 + }, + { + "epoch": 0.13013611499809186, + "grad_norm": 2.805617094039917, + "learning_rate": 4.332344213649852e-07, + "loss": 0.4087, + "mean_token_accuracy": 0.863040030002594, + "num_tokens": 39670415.0, + "step": 1023 + }, + { + "epoch": 0.13026332527668236, + "grad_norm": 2.8160288333892822, + "learning_rate": 4.33658329800763e-07, + "loss": 0.4425, + "mean_token_accuracy": 0.8564861416816711, + "num_tokens": 39711364.0, + "step": 1024 + }, + { + "epoch": 0.13039053555527286, + "grad_norm": 2.9637012481689453, + "learning_rate": 4.340822382365409e-07, + "loss": 0.4514, + "mean_token_accuracy": 0.8448246121406555, + "num_tokens": 39748307.0, + "step": 1025 + }, + { + "epoch": 0.13051774583386339, + "grad_norm": 3.2282755374908447, + "learning_rate": 4.345061466723188e-07, + "loss": 0.4569, + "mean_token_accuracy": 0.8479827642440796, + "num_tokens": 39782841.0, + "step": 1026 + }, + { + "epoch": 0.13064495611245389, + "grad_norm": 2.7832138538360596, + "learning_rate": 4.3493005510809663e-07, + "loss": 0.4174, + "mean_token_accuracy": 0.8601952195167542, + "num_tokens": 39823440.0, + "step": 1027 + }, + { + "epoch": 0.1307721663910444, + "grad_norm": 3.011110782623291, + "learning_rate": 4.353539635438745e-07, + "loss": 0.449, + "mean_token_accuracy": 0.8492259979248047, + "num_tokens": 39860348.0, + "step": 1028 + }, + { + "epoch": 0.13089937666963491, + "grad_norm": 2.755582332611084, + "learning_rate": 4.357778719796524e-07, + "loss": 0.4359, + "mean_token_accuracy": 0.8552533388137817, + "num_tokens": 39900071.0, + "step": 1029 + }, + { + "epoch": 0.13102658694822542, + "grad_norm": 2.934344530105591, + "learning_rate": 4.362017804154303e-07, + "loss": 0.3879, + "mean_token_accuracy": 0.8702017068862915, + "num_tokens": 39939537.0, + "step": 1030 + }, + { + "epoch": 0.13115379722681592, + "grad_norm": 2.947167158126831, + "learning_rate": 4.366256888512081e-07, + "loss": 0.4265, + "mean_token_accuracy": 0.8578008413314819, + "num_tokens": 39977987.0, + "step": 1031 + }, + { + "epoch": 0.13128100750540644, + "grad_norm": 3.0113942623138428, + "learning_rate": 4.3704959728698597e-07, + "loss": 0.3768, + "mean_token_accuracy": 0.8669897317886353, + "num_tokens": 40010532.0, + "step": 1032 + }, + { + "epoch": 0.13140821778399694, + "grad_norm": 2.851977586746216, + "learning_rate": 4.3747350572276386e-07, + "loss": 0.4264, + "mean_token_accuracy": 0.8538599610328674, + "num_tokens": 40049868.0, + "step": 1033 + }, + { + "epoch": 0.13153542806258745, + "grad_norm": 2.7237613201141357, + "learning_rate": 4.3789741415854176e-07, + "loss": 0.4054, + "mean_token_accuracy": 0.8609695434570312, + "num_tokens": 40091286.0, + "step": 1034 + }, + { + "epoch": 0.13166263834117797, + "grad_norm": 2.8652780055999756, + "learning_rate": 4.383213225943196e-07, + "loss": 0.407, + "mean_token_accuracy": 0.8633021116256714, + "num_tokens": 40132667.0, + "step": 1035 + }, + { + "epoch": 0.13178984861976847, + "grad_norm": 2.830280303955078, + "learning_rate": 4.3874523103009746e-07, + "loss": 0.4441, + "mean_token_accuracy": 0.8512459397315979, + "num_tokens": 40173661.0, + "step": 1036 + }, + { + "epoch": 0.13191705889835897, + "grad_norm": 2.911421775817871, + "learning_rate": 4.3916913946587536e-07, + "loss": 0.4226, + "mean_token_accuracy": 0.8586868643760681, + "num_tokens": 40211473.0, + "step": 1037 + }, + { + "epoch": 0.1320442691769495, + "grad_norm": 2.5295097827911377, + "learning_rate": 4.3959304790165325e-07, + "loss": 0.4039, + "mean_token_accuracy": 0.8640115261077881, + "num_tokens": 40258564.0, + "step": 1038 + }, + { + "epoch": 0.13217147945554, + "grad_norm": 2.732191801071167, + "learning_rate": 4.400169563374311e-07, + "loss": 0.4103, + "mean_token_accuracy": 0.8644980788230896, + "num_tokens": 40301401.0, + "step": 1039 + }, + { + "epoch": 0.13229868973413053, + "grad_norm": 2.930161714553833, + "learning_rate": 4.4044086477320895e-07, + "loss": 0.4661, + "mean_token_accuracy": 0.8482805490493774, + "num_tokens": 40341053.0, + "step": 1040 + }, + { + "epoch": 0.13242590001272103, + "grad_norm": 2.9840595722198486, + "learning_rate": 4.4086477320898685e-07, + "loss": 0.4093, + "mean_token_accuracy": 0.8588439226150513, + "num_tokens": 40376141.0, + "step": 1041 + }, + { + "epoch": 0.13255311029131153, + "grad_norm": 2.6062498092651367, + "learning_rate": 4.4128868164476474e-07, + "loss": 0.3842, + "mean_token_accuracy": 0.8721513748168945, + "num_tokens": 40417323.0, + "step": 1042 + }, + { + "epoch": 0.13268032056990206, + "grad_norm": 2.871034860610962, + "learning_rate": 4.417125900805426e-07, + "loss": 0.4396, + "mean_token_accuracy": 0.8528345823287964, + "num_tokens": 40458023.0, + "step": 1043 + }, + { + "epoch": 0.13280753084849256, + "grad_norm": 2.8390309810638428, + "learning_rate": 4.4213649851632044e-07, + "loss": 0.4097, + "mean_token_accuracy": 0.8592066764831543, + "num_tokens": 40496847.0, + "step": 1044 + }, + { + "epoch": 0.13293474112708306, + "grad_norm": 2.8775556087493896, + "learning_rate": 4.4256040695209834e-07, + "loss": 0.4061, + "mean_token_accuracy": 0.8599992990493774, + "num_tokens": 40533539.0, + "step": 1045 + }, + { + "epoch": 0.1330619514056736, + "grad_norm": 2.850593328475952, + "learning_rate": 4.429843153878762e-07, + "loss": 0.367, + "mean_token_accuracy": 0.8734802603721619, + "num_tokens": 40569565.0, + "step": 1046 + }, + { + "epoch": 0.1331891616842641, + "grad_norm": 2.866884231567383, + "learning_rate": 4.434082238236541e-07, + "loss": 0.4387, + "mean_token_accuracy": 0.8560354709625244, + "num_tokens": 40608452.0, + "step": 1047 + }, + { + "epoch": 0.1333163719628546, + "grad_norm": 2.9707350730895996, + "learning_rate": 4.4383213225943193e-07, + "loss": 0.4116, + "mean_token_accuracy": 0.8590748906135559, + "num_tokens": 40641899.0, + "step": 1048 + }, + { + "epoch": 0.13344358224144512, + "grad_norm": 2.659420967102051, + "learning_rate": 4.442560406952098e-07, + "loss": 0.4091, + "mean_token_accuracy": 0.8617557287216187, + "num_tokens": 40685864.0, + "step": 1049 + }, + { + "epoch": 0.13357079252003562, + "grad_norm": 2.6404783725738525, + "learning_rate": 4.4467994913098767e-07, + "loss": 0.4059, + "mean_token_accuracy": 0.8659340143203735, + "num_tokens": 40727714.0, + "step": 1050 + }, + { + "epoch": 0.13369800279862612, + "grad_norm": 2.6655802726745605, + "learning_rate": 4.4510385756676557e-07, + "loss": 0.3617, + "mean_token_accuracy": 0.8748683333396912, + "num_tokens": 40767402.0, + "step": 1051 + }, + { + "epoch": 0.13382521307721665, + "grad_norm": 2.6349613666534424, + "learning_rate": 4.455277660025434e-07, + "loss": 0.3679, + "mean_token_accuracy": 0.8695060014724731, + "num_tokens": 40807844.0, + "step": 1052 + }, + { + "epoch": 0.13395242335580715, + "grad_norm": 2.832610845565796, + "learning_rate": 4.459516744383213e-07, + "loss": 0.427, + "mean_token_accuracy": 0.8509171009063721, + "num_tokens": 40846873.0, + "step": 1053 + }, + { + "epoch": 0.13407963363439765, + "grad_norm": 3.4160141944885254, + "learning_rate": 4.4637558287409916e-07, + "loss": 0.4841, + "mean_token_accuracy": 0.8437716960906982, + "num_tokens": 40881572.0, + "step": 1054 + }, + { + "epoch": 0.13420684391298818, + "grad_norm": 2.937290668487549, + "learning_rate": 4.4679949130987706e-07, + "loss": 0.4552, + "mean_token_accuracy": 0.8514159917831421, + "num_tokens": 40921933.0, + "step": 1055 + }, + { + "epoch": 0.13433405419157868, + "grad_norm": 2.7459444999694824, + "learning_rate": 4.472233997456549e-07, + "loss": 0.4618, + "mean_token_accuracy": 0.8451858758926392, + "num_tokens": 40968439.0, + "step": 1056 + }, + { + "epoch": 0.13446126447016918, + "grad_norm": 3.0704526901245117, + "learning_rate": 4.476473081814328e-07, + "loss": 0.4276, + "mean_token_accuracy": 0.85410076379776, + "num_tokens": 41004969.0, + "step": 1057 + }, + { + "epoch": 0.1345884747487597, + "grad_norm": 2.7414748668670654, + "learning_rate": 4.4807121661721065e-07, + "loss": 0.426, + "mean_token_accuracy": 0.8573044538497925, + "num_tokens": 41047241.0, + "step": 1058 + }, + { + "epoch": 0.1347156850273502, + "grad_norm": 2.7307896614074707, + "learning_rate": 4.4849512505298855e-07, + "loss": 0.3929, + "mean_token_accuracy": 0.8644355535507202, + "num_tokens": 41085780.0, + "step": 1059 + }, + { + "epoch": 0.1348428953059407, + "grad_norm": 2.922332286834717, + "learning_rate": 4.489190334887664e-07, + "loss": 0.4628, + "mean_token_accuracy": 0.8423210382461548, + "num_tokens": 41124401.0, + "step": 1060 + }, + { + "epoch": 0.13497010558453124, + "grad_norm": 2.61688494682312, + "learning_rate": 4.493429419245443e-07, + "loss": 0.3412, + "mean_token_accuracy": 0.8837506771087646, + "num_tokens": 41166773.0, + "step": 1061 + }, + { + "epoch": 0.13509731586312174, + "grad_norm": 2.8544795513153076, + "learning_rate": 4.4976685036032214e-07, + "loss": 0.4327, + "mean_token_accuracy": 0.8544676303863525, + "num_tokens": 41205382.0, + "step": 1062 + }, + { + "epoch": 0.13522452614171224, + "grad_norm": 3.085507392883301, + "learning_rate": 4.5019075879610004e-07, + "loss": 0.4146, + "mean_token_accuracy": 0.8594121932983398, + "num_tokens": 41240396.0, + "step": 1063 + }, + { + "epoch": 0.13535173642030277, + "grad_norm": 2.9273650646209717, + "learning_rate": 4.506146672318779e-07, + "loss": 0.3721, + "mean_token_accuracy": 0.8732407093048096, + "num_tokens": 41275773.0, + "step": 1064 + }, + { + "epoch": 0.13547894669889327, + "grad_norm": 2.615001916885376, + "learning_rate": 4.5103857566765573e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8676689863204956, + "num_tokens": 41318632.0, + "step": 1065 + }, + { + "epoch": 0.1356061569774838, + "grad_norm": 2.5744922161102295, + "learning_rate": 4.5146248410343363e-07, + "loss": 0.3918, + "mean_token_accuracy": 0.8672397136688232, + "num_tokens": 41360547.0, + "step": 1066 + }, + { + "epoch": 0.1357333672560743, + "grad_norm": 2.8534395694732666, + "learning_rate": 4.5188639253921153e-07, + "loss": 0.3913, + "mean_token_accuracy": 0.8664030432701111, + "num_tokens": 41396151.0, + "step": 1067 + }, + { + "epoch": 0.1358605775346648, + "grad_norm": 2.9487264156341553, + "learning_rate": 4.523103009749894e-07, + "loss": 0.4404, + "mean_token_accuracy": 0.8522404432296753, + "num_tokens": 41433414.0, + "step": 1068 + }, + { + "epoch": 0.13598778781325532, + "grad_norm": 2.4305453300476074, + "learning_rate": 4.527342094107672e-07, + "loss": 0.3898, + "mean_token_accuracy": 0.867996871471405, + "num_tokens": 41480091.0, + "step": 1069 + }, + { + "epoch": 0.13611499809184582, + "grad_norm": 2.737745761871338, + "learning_rate": 4.531581178465451e-07, + "loss": 0.3737, + "mean_token_accuracy": 0.8720686435699463, + "num_tokens": 41518526.0, + "step": 1070 + }, + { + "epoch": 0.13624220837043632, + "grad_norm": 2.6470959186553955, + "learning_rate": 4.53582026282323e-07, + "loss": 0.4258, + "mean_token_accuracy": 0.8592341542243958, + "num_tokens": 41565767.0, + "step": 1071 + }, + { + "epoch": 0.13636941864902685, + "grad_norm": 2.752967119216919, + "learning_rate": 4.5400593471810087e-07, + "loss": 0.3958, + "mean_token_accuracy": 0.8628331422805786, + "num_tokens": 41604942.0, + "step": 1072 + }, + { + "epoch": 0.13649662892761735, + "grad_norm": 2.9068856239318848, + "learning_rate": 4.544298431538787e-07, + "loss": 0.3876, + "mean_token_accuracy": 0.8692964315414429, + "num_tokens": 41642979.0, + "step": 1073 + }, + { + "epoch": 0.13662383920620785, + "grad_norm": 2.6191298961639404, + "learning_rate": 4.548537515896566e-07, + "loss": 0.3555, + "mean_token_accuracy": 0.87703537940979, + "num_tokens": 41684644.0, + "step": 1074 + }, + { + "epoch": 0.13675104948479838, + "grad_norm": 2.8875794410705566, + "learning_rate": 4.552776600254345e-07, + "loss": 0.422, + "mean_token_accuracy": 0.856176495552063, + "num_tokens": 41723725.0, + "step": 1075 + }, + { + "epoch": 0.13687825976338888, + "grad_norm": 2.791749954223633, + "learning_rate": 4.5570156846121236e-07, + "loss": 0.4109, + "mean_token_accuracy": 0.8587502241134644, + "num_tokens": 41761373.0, + "step": 1076 + }, + { + "epoch": 0.13700547004197938, + "grad_norm": 2.8769686222076416, + "learning_rate": 4.561254768969902e-07, + "loss": 0.3941, + "mean_token_accuracy": 0.8682489991188049, + "num_tokens": 41799892.0, + "step": 1077 + }, + { + "epoch": 0.1371326803205699, + "grad_norm": 2.5124189853668213, + "learning_rate": 4.565493853327681e-07, + "loss": 0.4169, + "mean_token_accuracy": 0.8617348670959473, + "num_tokens": 41846052.0, + "step": 1078 + }, + { + "epoch": 0.1372598905991604, + "grad_norm": 2.9377543926239014, + "learning_rate": 4.56973293768546e-07, + "loss": 0.4209, + "mean_token_accuracy": 0.859390377998352, + "num_tokens": 41884956.0, + "step": 1079 + }, + { + "epoch": 0.1373871008777509, + "grad_norm": 2.9164934158325195, + "learning_rate": 4.573972022043238e-07, + "loss": 0.4213, + "mean_token_accuracy": 0.8622229695320129, + "num_tokens": 41921682.0, + "step": 1080 + }, + { + "epoch": 0.13751431115634144, + "grad_norm": 3.2225265502929688, + "learning_rate": 4.578211106401017e-07, + "loss": 0.4518, + "mean_token_accuracy": 0.853928804397583, + "num_tokens": 41957865.0, + "step": 1081 + }, + { + "epoch": 0.13764152143493194, + "grad_norm": 2.898556709289551, + "learning_rate": 4.582450190758796e-07, + "loss": 0.3801, + "mean_token_accuracy": 0.8707260489463806, + "num_tokens": 41995451.0, + "step": 1082 + }, + { + "epoch": 0.13776873171352244, + "grad_norm": 2.8979222774505615, + "learning_rate": 4.586689275116575e-07, + "loss": 0.4313, + "mean_token_accuracy": 0.8543671369552612, + "num_tokens": 42034574.0, + "step": 1083 + }, + { + "epoch": 0.13789594199211297, + "grad_norm": 3.084042549133301, + "learning_rate": 4.590928359474353e-07, + "loss": 0.4283, + "mean_token_accuracy": 0.8535892963409424, + "num_tokens": 42070427.0, + "step": 1084 + }, + { + "epoch": 0.13802315227070347, + "grad_norm": 2.870990514755249, + "learning_rate": 4.595167443832132e-07, + "loss": 0.4002, + "mean_token_accuracy": 0.8644556999206543, + "num_tokens": 42111578.0, + "step": 1085 + }, + { + "epoch": 0.13815036254929397, + "grad_norm": 2.8468105792999268, + "learning_rate": 4.599406528189911e-07, + "loss": 0.4009, + "mean_token_accuracy": 0.8641381859779358, + "num_tokens": 42148874.0, + "step": 1086 + }, + { + "epoch": 0.1382775728278845, + "grad_norm": 2.7522575855255127, + "learning_rate": 4.60364561254769e-07, + "loss": 0.4883, + "mean_token_accuracy": 0.8390804529190063, + "num_tokens": 42196153.0, + "step": 1087 + }, + { + "epoch": 0.138404783106475, + "grad_norm": 2.670895576477051, + "learning_rate": 4.607884696905468e-07, + "loss": 0.3686, + "mean_token_accuracy": 0.874391496181488, + "num_tokens": 42240779.0, + "step": 1088 + }, + { + "epoch": 0.1385319933850655, + "grad_norm": 2.964017391204834, + "learning_rate": 4.612123781263247e-07, + "loss": 0.3823, + "mean_token_accuracy": 0.869458019733429, + "num_tokens": 42274811.0, + "step": 1089 + }, + { + "epoch": 0.13865920366365603, + "grad_norm": 2.8859922885894775, + "learning_rate": 4.616362865621026e-07, + "loss": 0.417, + "mean_token_accuracy": 0.8598135113716125, + "num_tokens": 42315148.0, + "step": 1090 + }, + { + "epoch": 0.13878641394224653, + "grad_norm": 2.621424913406372, + "learning_rate": 4.620601949978805e-07, + "loss": 0.3923, + "mean_token_accuracy": 0.8669700622558594, + "num_tokens": 42356466.0, + "step": 1091 + }, + { + "epoch": 0.13891362422083706, + "grad_norm": 2.9828829765319824, + "learning_rate": 4.6248410343365827e-07, + "loss": 0.473, + "mean_token_accuracy": 0.8435489535331726, + "num_tokens": 42397266.0, + "step": 1092 + }, + { + "epoch": 0.13904083449942756, + "grad_norm": 3.166926383972168, + "learning_rate": 4.6290801186943617e-07, + "loss": 0.4178, + "mean_token_accuracy": 0.85588538646698, + "num_tokens": 42431047.0, + "step": 1093 + }, + { + "epoch": 0.13916804477801806, + "grad_norm": 2.558521032333374, + "learning_rate": 4.6333192030521407e-07, + "loss": 0.4098, + "mean_token_accuracy": 0.8611462116241455, + "num_tokens": 42477100.0, + "step": 1094 + }, + { + "epoch": 0.13929525505660859, + "grad_norm": 2.7707269191741943, + "learning_rate": 4.6375582874099196e-07, + "loss": 0.3692, + "mean_token_accuracy": 0.8716941475868225, + "num_tokens": 42513845.0, + "step": 1095 + }, + { + "epoch": 0.1394224653351991, + "grad_norm": 3.032184362411499, + "learning_rate": 4.6417973717676976e-07, + "loss": 0.4305, + "mean_token_accuracy": 0.8530531525611877, + "num_tokens": 42548886.0, + "step": 1096 + }, + { + "epoch": 0.1395496756137896, + "grad_norm": 3.3074631690979004, + "learning_rate": 4.6460364561254766e-07, + "loss": 0.4029, + "mean_token_accuracy": 0.8638414144515991, + "num_tokens": 42580011.0, + "step": 1097 + }, + { + "epoch": 0.13967688589238011, + "grad_norm": 2.625669002532959, + "learning_rate": 4.6502755404832556e-07, + "loss": 0.4413, + "mean_token_accuracy": 0.8517278432846069, + "num_tokens": 42625667.0, + "step": 1098 + }, + { + "epoch": 0.13980409617097062, + "grad_norm": 2.9449799060821533, + "learning_rate": 4.654514624841034e-07, + "loss": 0.406, + "mean_token_accuracy": 0.8595618605613708, + "num_tokens": 42660561.0, + "step": 1099 + }, + { + "epoch": 0.13993130644956112, + "grad_norm": 3.0360350608825684, + "learning_rate": 4.6587537091988125e-07, + "loss": 0.4158, + "mean_token_accuracy": 0.8618713617324829, + "num_tokens": 42698275.0, + "step": 1100 + }, + { + "epoch": 0.14005851672815164, + "grad_norm": 3.218735456466675, + "learning_rate": 4.6629927935565915e-07, + "loss": 0.4012, + "mean_token_accuracy": 0.8619041442871094, + "num_tokens": 42729171.0, + "step": 1101 + }, + { + "epoch": 0.14018572700674214, + "grad_norm": 2.716587543487549, + "learning_rate": 4.6672318779143705e-07, + "loss": 0.3997, + "mean_token_accuracy": 0.8631354570388794, + "num_tokens": 42771130.0, + "step": 1102 + }, + { + "epoch": 0.14031293728533265, + "grad_norm": 4.047377109527588, + "learning_rate": 4.671470962272149e-07, + "loss": 0.4445, + "mean_token_accuracy": 0.8540980815887451, + "num_tokens": 42814810.0, + "step": 1103 + }, + { + "epoch": 0.14044014756392317, + "grad_norm": 2.998528003692627, + "learning_rate": 4.6757100466299274e-07, + "loss": 0.3847, + "mean_token_accuracy": 0.867976188659668, + "num_tokens": 42850923.0, + "step": 1104 + }, + { + "epoch": 0.14056735784251367, + "grad_norm": 2.69655704498291, + "learning_rate": 4.6799491309877064e-07, + "loss": 0.402, + "mean_token_accuracy": 0.8645148277282715, + "num_tokens": 42894645.0, + "step": 1105 + }, + { + "epoch": 0.14069456812110417, + "grad_norm": 3.140934467315674, + "learning_rate": 4.6841882153454854e-07, + "loss": 0.4564, + "mean_token_accuracy": 0.8477994203567505, + "num_tokens": 42931971.0, + "step": 1106 + }, + { + "epoch": 0.1408217783996947, + "grad_norm": 2.8304126262664795, + "learning_rate": 4.688427299703264e-07, + "loss": 0.4115, + "mean_token_accuracy": 0.8629132509231567, + "num_tokens": 42971603.0, + "step": 1107 + }, + { + "epoch": 0.1409489886782852, + "grad_norm": 2.978703022003174, + "learning_rate": 4.6926663840610423e-07, + "loss": 0.4727, + "mean_token_accuracy": 0.8455787897109985, + "num_tokens": 43013713.0, + "step": 1108 + }, + { + "epoch": 0.1410761989568757, + "grad_norm": 8.326889038085938, + "learning_rate": 4.6969054684188213e-07, + "loss": 0.4539, + "mean_token_accuracy": 0.8459596633911133, + "num_tokens": 43054059.0, + "step": 1109 + }, + { + "epoch": 0.14120340923546623, + "grad_norm": 3.0774428844451904, + "learning_rate": 4.7011445527766003e-07, + "loss": 0.3746, + "mean_token_accuracy": 0.8711234331130981, + "num_tokens": 43086930.0, + "step": 1110 + }, + { + "epoch": 0.14133061951405673, + "grad_norm": 2.7482011318206787, + "learning_rate": 4.7053836371343787e-07, + "loss": 0.3425, + "mean_token_accuracy": 0.8832632303237915, + "num_tokens": 43123362.0, + "step": 1111 + }, + { + "epoch": 0.14145782979264723, + "grad_norm": 2.989131212234497, + "learning_rate": 4.709622721492157e-07, + "loss": 0.4259, + "mean_token_accuracy": 0.8549695611000061, + "num_tokens": 43157844.0, + "step": 1112 + }, + { + "epoch": 0.14158504007123776, + "grad_norm": 2.7028088569641113, + "learning_rate": 4.713861805849936e-07, + "loss": 0.3904, + "mean_token_accuracy": 0.867603063583374, + "num_tokens": 43198805.0, + "step": 1113 + }, + { + "epoch": 0.14171225034982826, + "grad_norm": 2.7344603538513184, + "learning_rate": 4.718100890207715e-07, + "loss": 0.4122, + "mean_token_accuracy": 0.8614523410797119, + "num_tokens": 43240570.0, + "step": 1114 + }, + { + "epoch": 0.1418394606284188, + "grad_norm": 2.643117904663086, + "learning_rate": 4.7223399745654936e-07, + "loss": 0.3968, + "mean_token_accuracy": 0.8670189380645752, + "num_tokens": 43285336.0, + "step": 1115 + }, + { + "epoch": 0.1419666709070093, + "grad_norm": 2.7038302421569824, + "learning_rate": 4.726579058923272e-07, + "loss": 0.364, + "mean_token_accuracy": 0.8754924535751343, + "num_tokens": 43325173.0, + "step": 1116 + }, + { + "epoch": 0.1420938811855998, + "grad_norm": 2.949918270111084, + "learning_rate": 4.730818143281051e-07, + "loss": 0.4087, + "mean_token_accuracy": 0.8649975061416626, + "num_tokens": 43360550.0, + "step": 1117 + }, + { + "epoch": 0.14222109146419032, + "grad_norm": 3.1416234970092773, + "learning_rate": 4.7350572276388295e-07, + "loss": 0.4233, + "mean_token_accuracy": 0.8546674251556396, + "num_tokens": 43393925.0, + "step": 1118 + }, + { + "epoch": 0.14234830174278082, + "grad_norm": 3.0367274284362793, + "learning_rate": 4.7392963119966085e-07, + "loss": 0.4016, + "mean_token_accuracy": 0.8696810603141785, + "num_tokens": 43428759.0, + "step": 1119 + }, + { + "epoch": 0.14247551202137132, + "grad_norm": 2.8121225833892822, + "learning_rate": 4.7435353963543875e-07, + "loss": 0.3899, + "mean_token_accuracy": 0.8712196350097656, + "num_tokens": 43468052.0, + "step": 1120 + }, + { + "epoch": 0.14260272229996185, + "grad_norm": 2.678389549255371, + "learning_rate": 4.747774480712166e-07, + "loss": 0.3792, + "mean_token_accuracy": 0.8671140074729919, + "num_tokens": 43508464.0, + "step": 1121 + }, + { + "epoch": 0.14272993257855235, + "grad_norm": 2.9477016925811768, + "learning_rate": 4.7520135650699444e-07, + "loss": 0.4333, + "mean_token_accuracy": 0.8532754182815552, + "num_tokens": 43546938.0, + "step": 1122 + }, + { + "epoch": 0.14285714285714285, + "grad_norm": 2.728548049926758, + "learning_rate": 4.7562526494277234e-07, + "loss": 0.3941, + "mean_token_accuracy": 0.8697174787521362, + "num_tokens": 43589567.0, + "step": 1123 + }, + { + "epoch": 0.14298435313573338, + "grad_norm": 3.0644373893737793, + "learning_rate": 4.7604917337855024e-07, + "loss": 0.4122, + "mean_token_accuracy": 0.8594734072685242, + "num_tokens": 43624767.0, + "step": 1124 + }, + { + "epoch": 0.14311156341432388, + "grad_norm": 2.8202710151672363, + "learning_rate": 4.764730818143281e-07, + "loss": 0.4384, + "mean_token_accuracy": 0.8558716177940369, + "num_tokens": 43667518.0, + "step": 1125 + }, + { + "epoch": 0.14323877369291438, + "grad_norm": 3.227099895477295, + "learning_rate": 4.768969902501059e-07, + "loss": 0.4107, + "mean_token_accuracy": 0.861481785774231, + "num_tokens": 43702008.0, + "step": 1126 + }, + { + "epoch": 0.1433659839715049, + "grad_norm": 2.973684787750244, + "learning_rate": 4.773208986858838e-07, + "loss": 0.3612, + "mean_token_accuracy": 0.8751287460327148, + "num_tokens": 43737425.0, + "step": 1127 + }, + { + "epoch": 0.1434931942500954, + "grad_norm": 2.7989747524261475, + "learning_rate": 4.777448071216617e-07, + "loss": 0.4305, + "mean_token_accuracy": 0.8538697957992554, + "num_tokens": 43777870.0, + "step": 1128 + }, + { + "epoch": 0.1436204045286859, + "grad_norm": 3.0485646724700928, + "learning_rate": 4.781687155574396e-07, + "loss": 0.4296, + "mean_token_accuracy": 0.859897792339325, + "num_tokens": 43816790.0, + "step": 1129 + }, + { + "epoch": 0.14374761480727644, + "grad_norm": 2.998767137527466, + "learning_rate": 4.785926239932175e-07, + "loss": 0.4653, + "mean_token_accuracy": 0.8420319557189941, + "num_tokens": 43857256.0, + "step": 1130 + }, + { + "epoch": 0.14387482508586694, + "grad_norm": 2.804793357849121, + "learning_rate": 4.790165324289953e-07, + "loss": 0.4038, + "mean_token_accuracy": 0.8629773855209351, + "num_tokens": 43897312.0, + "step": 1131 + }, + { + "epoch": 0.14400203536445744, + "grad_norm": 3.4123780727386475, + "learning_rate": 4.794404408647732e-07, + "loss": 0.3679, + "mean_token_accuracy": 0.870942234992981, + "num_tokens": 43931498.0, + "step": 1132 + }, + { + "epoch": 0.14412924564304797, + "grad_norm": 2.6825785636901855, + "learning_rate": 4.798643493005511e-07, + "loss": 0.3827, + "mean_token_accuracy": 0.870245635509491, + "num_tokens": 43975553.0, + "step": 1133 + }, + { + "epoch": 0.14425645592163847, + "grad_norm": 2.7181107997894287, + "learning_rate": 4.80288257736329e-07, + "loss": 0.3989, + "mean_token_accuracy": 0.8642114400863647, + "num_tokens": 44018297.0, + "step": 1134 + }, + { + "epoch": 0.14438366620022897, + "grad_norm": 2.575732946395874, + "learning_rate": 4.807121661721068e-07, + "loss": 0.4032, + "mean_token_accuracy": 0.8645197153091431, + "num_tokens": 44063984.0, + "step": 1135 + }, + { + "epoch": 0.1445108764788195, + "grad_norm": 2.91690731048584, + "learning_rate": 4.811360746078847e-07, + "loss": 0.4593, + "mean_token_accuracy": 0.846376895904541, + "num_tokens": 44103105.0, + "step": 1136 + }, + { + "epoch": 0.14463808675741, + "grad_norm": 2.8064751625061035, + "learning_rate": 4.815599830436625e-07, + "loss": 0.3605, + "mean_token_accuracy": 0.8757608532905579, + "num_tokens": 44144814.0, + "step": 1137 + }, + { + "epoch": 0.1447652970360005, + "grad_norm": 2.764458656311035, + "learning_rate": 4.819838914794405e-07, + "loss": 0.3858, + "mean_token_accuracy": 0.8681910037994385, + "num_tokens": 44184816.0, + "step": 1138 + }, + { + "epoch": 0.14489250731459102, + "grad_norm": 2.7735278606414795, + "learning_rate": 4.824077999152183e-07, + "loss": 0.3672, + "mean_token_accuracy": 0.874809741973877, + "num_tokens": 44225008.0, + "step": 1139 + }, + { + "epoch": 0.14501971759318152, + "grad_norm": 3.0357987880706787, + "learning_rate": 4.828317083509962e-07, + "loss": 0.3479, + "mean_token_accuracy": 0.879286527633667, + "num_tokens": 44263989.0, + "step": 1140 + }, + { + "epoch": 0.14514692787177205, + "grad_norm": 2.975449800491333, + "learning_rate": 4.83255616786774e-07, + "loss": 0.3935, + "mean_token_accuracy": 0.868320882320404, + "num_tokens": 44301179.0, + "step": 1141 + }, + { + "epoch": 0.14527413815036255, + "grad_norm": 2.8307478427886963, + "learning_rate": 4.83679525222552e-07, + "loss": 0.4177, + "mean_token_accuracy": 0.8584414720535278, + "num_tokens": 44340910.0, + "step": 1142 + }, + { + "epoch": 0.14540134842895305, + "grad_norm": 2.700052261352539, + "learning_rate": 4.841034336583298e-07, + "loss": 0.4222, + "mean_token_accuracy": 0.8576890230178833, + "num_tokens": 44384373.0, + "step": 1143 + }, + { + "epoch": 0.14552855870754358, + "grad_norm": 2.704045057296753, + "learning_rate": 4.845273420941076e-07, + "loss": 0.3876, + "mean_token_accuracy": 0.8681119680404663, + "num_tokens": 44424649.0, + "step": 1144 + }, + { + "epoch": 0.14565576898613408, + "grad_norm": 2.844470262527466, + "learning_rate": 4.849512505298855e-07, + "loss": 0.4547, + "mean_token_accuracy": 0.8462046980857849, + "num_tokens": 44465564.0, + "step": 1145 + }, + { + "epoch": 0.14578297926472458, + "grad_norm": 2.947012186050415, + "learning_rate": 4.853751589656634e-07, + "loss": 0.4038, + "mean_token_accuracy": 0.8642053604125977, + "num_tokens": 44500552.0, + "step": 1146 + }, + { + "epoch": 0.1459101895433151, + "grad_norm": 2.8390052318573, + "learning_rate": 4.857990674014413e-07, + "loss": 0.4462, + "mean_token_accuracy": 0.8498811721801758, + "num_tokens": 44541211.0, + "step": 1147 + }, + { + "epoch": 0.1460373998219056, + "grad_norm": 2.9471077919006348, + "learning_rate": 4.862229758372191e-07, + "loss": 0.397, + "mean_token_accuracy": 0.8651086091995239, + "num_tokens": 44578346.0, + "step": 1148 + }, + { + "epoch": 0.1461646101004961, + "grad_norm": 2.8708574771881104, + "learning_rate": 4.86646884272997e-07, + "loss": 0.4485, + "mean_token_accuracy": 0.851334273815155, + "num_tokens": 44616161.0, + "step": 1149 + }, + { + "epoch": 0.14629182037908664, + "grad_norm": 2.6584863662719727, + "learning_rate": 4.870707927087749e-07, + "loss": 0.4501, + "mean_token_accuracy": 0.8484465479850769, + "num_tokens": 44662831.0, + "step": 1150 + }, + { + "epoch": 0.14641903065767714, + "grad_norm": 2.717919111251831, + "learning_rate": 4.874947011445528e-07, + "loss": 0.3854, + "mean_token_accuracy": 0.8685464859008789, + "num_tokens": 44702753.0, + "step": 1151 + }, + { + "epoch": 0.14654624093626764, + "grad_norm": 3.04394268989563, + "learning_rate": 4.879186095803306e-07, + "loss": 0.4504, + "mean_token_accuracy": 0.8471408486366272, + "num_tokens": 44738327.0, + "step": 1152 + }, + { + "epoch": 0.14667345121485817, + "grad_norm": 2.8013429641723633, + "learning_rate": 4.883425180161085e-07, + "loss": 0.4826, + "mean_token_accuracy": 0.8436276316642761, + "num_tokens": 44783349.0, + "step": 1153 + }, + { + "epoch": 0.14680066149344867, + "grad_norm": 2.640794277191162, + "learning_rate": 4.887664264518864e-07, + "loss": 0.4323, + "mean_token_accuracy": 0.8557714223861694, + "num_tokens": 44829114.0, + "step": 1154 + }, + { + "epoch": 0.14692787177203917, + "grad_norm": 3.058492660522461, + "learning_rate": 4.891903348876643e-07, + "loss": 0.4827, + "mean_token_accuracy": 0.837668776512146, + "num_tokens": 44867649.0, + "step": 1155 + }, + { + "epoch": 0.1470550820506297, + "grad_norm": 2.587554693222046, + "learning_rate": 4.896142433234421e-07, + "loss": 0.3769, + "mean_token_accuracy": 0.8708475828170776, + "num_tokens": 44909889.0, + "step": 1156 + }, + { + "epoch": 0.1471822923292202, + "grad_norm": 2.674124002456665, + "learning_rate": 4.9003815175922e-07, + "loss": 0.4038, + "mean_token_accuracy": 0.8633288741111755, + "num_tokens": 44952128.0, + "step": 1157 + }, + { + "epoch": 0.1473095026078107, + "grad_norm": 2.9901657104492188, + "learning_rate": 4.904620601949979e-07, + "loss": 0.4094, + "mean_token_accuracy": 0.8612309098243713, + "num_tokens": 44986487.0, + "step": 1158 + }, + { + "epoch": 0.14743671288640123, + "grad_norm": 2.7137222290039062, + "learning_rate": 4.908859686307758e-07, + "loss": 0.4609, + "mean_token_accuracy": 0.846146821975708, + "num_tokens": 45032606.0, + "step": 1159 + }, + { + "epoch": 0.14756392316499173, + "grad_norm": 3.0255942344665527, + "learning_rate": 4.913098770665536e-07, + "loss": 0.3688, + "mean_token_accuracy": 0.8755422830581665, + "num_tokens": 45066187.0, + "step": 1160 + }, + { + "epoch": 0.14769113344358223, + "grad_norm": 2.9364867210388184, + "learning_rate": 4.917337855023314e-07, + "loss": 0.3748, + "mean_token_accuracy": 0.8731276988983154, + "num_tokens": 45106753.0, + "step": 1161 + }, + { + "epoch": 0.14781834372217276, + "grad_norm": 2.960442066192627, + "learning_rate": 4.921576939381094e-07, + "loss": 0.4378, + "mean_token_accuracy": 0.8546316623687744, + "num_tokens": 45147952.0, + "step": 1162 + }, + { + "epoch": 0.14794555400076326, + "grad_norm": 2.4814891815185547, + "learning_rate": 4.925816023738872e-07, + "loss": 0.3777, + "mean_token_accuracy": 0.8739768266677856, + "num_tokens": 45194571.0, + "step": 1163 + }, + { + "epoch": 0.14807276427935376, + "grad_norm": 2.982750177383423, + "learning_rate": 4.930055108096651e-07, + "loss": 0.3973, + "mean_token_accuracy": 0.8639512658119202, + "num_tokens": 45230450.0, + "step": 1164 + }, + { + "epoch": 0.1481999745579443, + "grad_norm": 2.5906288623809814, + "learning_rate": 4.934294192454429e-07, + "loss": 0.3484, + "mean_token_accuracy": 0.8806281089782715, + "num_tokens": 45272854.0, + "step": 1165 + }, + { + "epoch": 0.1483271848365348, + "grad_norm": 2.7323031425476074, + "learning_rate": 4.938533276812209e-07, + "loss": 0.3649, + "mean_token_accuracy": 0.874445915222168, + "num_tokens": 45311909.0, + "step": 1166 + }, + { + "epoch": 0.14845439511512531, + "grad_norm": 2.8611717224121094, + "learning_rate": 4.942772361169987e-07, + "loss": 0.4398, + "mean_token_accuracy": 0.8552452325820923, + "num_tokens": 45354517.0, + "step": 1167 + }, + { + "epoch": 0.14858160539371582, + "grad_norm": 2.87379789352417, + "learning_rate": 4.947011445527766e-07, + "loss": 0.3935, + "mean_token_accuracy": 0.8650425672531128, + "num_tokens": 45392771.0, + "step": 1168 + }, + { + "epoch": 0.14870881567230632, + "grad_norm": 3.1686487197875977, + "learning_rate": 4.951250529885544e-07, + "loss": 0.4751, + "mean_token_accuracy": 0.8414303064346313, + "num_tokens": 45427023.0, + "step": 1169 + }, + { + "epoch": 0.14883602595089684, + "grad_norm": 2.875298023223877, + "learning_rate": 4.955489614243324e-07, + "loss": 0.3974, + "mean_token_accuracy": 0.8645700216293335, + "num_tokens": 45464168.0, + "step": 1170 + }, + { + "epoch": 0.14896323622948734, + "grad_norm": 2.787519931793213, + "learning_rate": 4.959728698601102e-07, + "loss": 0.4178, + "mean_token_accuracy": 0.8583654165267944, + "num_tokens": 45504658.0, + "step": 1171 + }, + { + "epoch": 0.14909044650807785, + "grad_norm": 3.2947144508361816, + "learning_rate": 4.963967782958881e-07, + "loss": 0.44, + "mean_token_accuracy": 0.8526566624641418, + "num_tokens": 45538242.0, + "step": 1172 + }, + { + "epoch": 0.14921765678666837, + "grad_norm": 2.880235195159912, + "learning_rate": 4.968206867316659e-07, + "loss": 0.4097, + "mean_token_accuracy": 0.8606793284416199, + "num_tokens": 45576036.0, + "step": 1173 + }, + { + "epoch": 0.14934486706525887, + "grad_norm": 2.7778725624084473, + "learning_rate": 4.972445951674439e-07, + "loss": 0.3899, + "mean_token_accuracy": 0.8660343885421753, + "num_tokens": 45615555.0, + "step": 1174 + }, + { + "epoch": 0.14947207734384937, + "grad_norm": 2.828479766845703, + "learning_rate": 4.976685036032216e-07, + "loss": 0.3875, + "mean_token_accuracy": 0.8678924441337585, + "num_tokens": 45653730.0, + "step": 1175 + }, + { + "epoch": 0.1495992876224399, + "grad_norm": 2.868820905685425, + "learning_rate": 4.980924120389996e-07, + "loss": 0.446, + "mean_token_accuracy": 0.8499760031700134, + "num_tokens": 45698802.0, + "step": 1176 + }, + { + "epoch": 0.1497264979010304, + "grad_norm": 2.7366373538970947, + "learning_rate": 4.985163204747774e-07, + "loss": 0.3829, + "mean_token_accuracy": 0.8662921786308289, + "num_tokens": 45737078.0, + "step": 1177 + }, + { + "epoch": 0.1498537081796209, + "grad_norm": 2.763869285583496, + "learning_rate": 4.989402289105554e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8597198724746704, + "num_tokens": 45781527.0, + "step": 1178 + }, + { + "epoch": 0.14998091845821143, + "grad_norm": 2.938192844390869, + "learning_rate": 4.993641373463331e-07, + "loss": 0.4409, + "mean_token_accuracy": 0.851905107498169, + "num_tokens": 45818809.0, + "step": 1179 + }, + { + "epoch": 0.15010812873680193, + "grad_norm": 2.750814199447632, + "learning_rate": 4.997880457821111e-07, + "loss": 0.3863, + "mean_token_accuracy": 0.8694754242897034, + "num_tokens": 45860787.0, + "step": 1180 + }, + { + "epoch": 0.15023533901539243, + "grad_norm": 2.8704166412353516, + "learning_rate": 5.002119542178889e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8619622588157654, + "num_tokens": 45900783.0, + "step": 1181 + }, + { + "epoch": 0.15036254929398296, + "grad_norm": 2.951784610748291, + "learning_rate": 5.006358626536667e-07, + "loss": 0.4569, + "mean_token_accuracy": 0.8490057587623596, + "num_tokens": 45940253.0, + "step": 1182 + }, + { + "epoch": 0.15048975957257346, + "grad_norm": 2.6935582160949707, + "learning_rate": 5.010597710894446e-07, + "loss": 0.4116, + "mean_token_accuracy": 0.8578410744667053, + "num_tokens": 45984680.0, + "step": 1183 + }, + { + "epoch": 0.15061696985116396, + "grad_norm": 2.9481992721557617, + "learning_rate": 5.014836795252225e-07, + "loss": 0.4505, + "mean_token_accuracy": 0.8504010438919067, + "num_tokens": 46024023.0, + "step": 1184 + }, + { + "epoch": 0.1507441801297545, + "grad_norm": 2.973905324935913, + "learning_rate": 5.019075879610004e-07, + "loss": 0.3916, + "mean_token_accuracy": 0.8658559322357178, + "num_tokens": 46061211.0, + "step": 1185 + }, + { + "epoch": 0.150871390408345, + "grad_norm": 3.0037519931793213, + "learning_rate": 5.023314963967783e-07, + "loss": 0.4159, + "mean_token_accuracy": 0.8617513179779053, + "num_tokens": 46097907.0, + "step": 1186 + }, + { + "epoch": 0.1509986006869355, + "grad_norm": 2.9382615089416504, + "learning_rate": 5.027554048325562e-07, + "loss": 0.4204, + "mean_token_accuracy": 0.8635598421096802, + "num_tokens": 46135172.0, + "step": 1187 + }, + { + "epoch": 0.15112581096552602, + "grad_norm": 2.7497847080230713, + "learning_rate": 5.03179313268334e-07, + "loss": 0.4576, + "mean_token_accuracy": 0.8487474918365479, + "num_tokens": 46181004.0, + "step": 1188 + }, + { + "epoch": 0.15125302124411652, + "grad_norm": 2.974182605743408, + "learning_rate": 5.036032217041119e-07, + "loss": 0.4375, + "mean_token_accuracy": 0.8531477451324463, + "num_tokens": 46218873.0, + "step": 1189 + }, + { + "epoch": 0.15138023152270705, + "grad_norm": 2.897552013397217, + "learning_rate": 5.040271301398897e-07, + "loss": 0.3955, + "mean_token_accuracy": 0.8607338666915894, + "num_tokens": 46257061.0, + "step": 1190 + }, + { + "epoch": 0.15150744180129755, + "grad_norm": 2.9554505348205566, + "learning_rate": 5.044510385756676e-07, + "loss": 0.3892, + "mean_token_accuracy": 0.8672342896461487, + "num_tokens": 46293162.0, + "step": 1191 + }, + { + "epoch": 0.15163465207988805, + "grad_norm": 2.819784164428711, + "learning_rate": 5.048749470114455e-07, + "loss": 0.4347, + "mean_token_accuracy": 0.8567837476730347, + "num_tokens": 46333145.0, + "step": 1192 + }, + { + "epoch": 0.15176186235847858, + "grad_norm": 2.6452765464782715, + "learning_rate": 5.052988554472234e-07, + "loss": 0.3571, + "mean_token_accuracy": 0.8797637224197388, + "num_tokens": 46372232.0, + "step": 1193 + }, + { + "epoch": 0.15188907263706908, + "grad_norm": 2.8538310527801514, + "learning_rate": 5.057227638830013e-07, + "loss": 0.4121, + "mean_token_accuracy": 0.8604187965393066, + "num_tokens": 46408924.0, + "step": 1194 + }, + { + "epoch": 0.15201628291565958, + "grad_norm": 2.5293242931365967, + "learning_rate": 5.061466723187792e-07, + "loss": 0.3538, + "mean_token_accuracy": 0.8784399032592773, + "num_tokens": 46451489.0, + "step": 1195 + }, + { + "epoch": 0.1521434931942501, + "grad_norm": 3.0980193614959717, + "learning_rate": 5.065705807545569e-07, + "loss": 0.4659, + "mean_token_accuracy": 0.8416677117347717, + "num_tokens": 46490787.0, + "step": 1196 + }, + { + "epoch": 0.1522707034728406, + "grad_norm": 2.806338310241699, + "learning_rate": 5.069944891903349e-07, + "loss": 0.423, + "mean_token_accuracy": 0.8584179282188416, + "num_tokens": 46532566.0, + "step": 1197 + }, + { + "epoch": 0.1523979137514311, + "grad_norm": 2.766057252883911, + "learning_rate": 5.074183976261127e-07, + "loss": 0.4399, + "mean_token_accuracy": 0.853913426399231, + "num_tokens": 46576342.0, + "step": 1198 + }, + { + "epoch": 0.15252512403002164, + "grad_norm": 2.8807694911956787, + "learning_rate": 5.078423060618906e-07, + "loss": 0.4582, + "mean_token_accuracy": 0.849717378616333, + "num_tokens": 46617979.0, + "step": 1199 + }, + { + "epoch": 0.15265233430861214, + "grad_norm": 2.714082956314087, + "learning_rate": 5.082662144976685e-07, + "loss": 0.4252, + "mean_token_accuracy": 0.8569777011871338, + "num_tokens": 46659907.0, + "step": 1200 + }, + { + "epoch": 0.15277954458720264, + "grad_norm": 3.0219357013702393, + "learning_rate": 5.086901229334464e-07, + "loss": 0.413, + "mean_token_accuracy": 0.856528103351593, + "num_tokens": 46697400.0, + "step": 1201 + }, + { + "epoch": 0.15290675486579317, + "grad_norm": 2.894122838973999, + "learning_rate": 5.091140313692243e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8606155514717102, + "num_tokens": 46736008.0, + "step": 1202 + }, + { + "epoch": 0.15303396514438367, + "grad_norm": 2.881399154663086, + "learning_rate": 5.095379398050022e-07, + "loss": 0.372, + "mean_token_accuracy": 0.8731207251548767, + "num_tokens": 46769862.0, + "step": 1203 + }, + { + "epoch": 0.15316117542297417, + "grad_norm": 2.774775743484497, + "learning_rate": 5.099618482407799e-07, + "loss": 0.4183, + "mean_token_accuracy": 0.859472930431366, + "num_tokens": 46809998.0, + "step": 1204 + }, + { + "epoch": 0.1532883857015647, + "grad_norm": 2.9548487663269043, + "learning_rate": 5.103857566765578e-07, + "loss": 0.3801, + "mean_token_accuracy": 0.8687626719474792, + "num_tokens": 46843663.0, + "step": 1205 + }, + { + "epoch": 0.1534155959801552, + "grad_norm": 2.7467446327209473, + "learning_rate": 5.108096651123357e-07, + "loss": 0.367, + "mean_token_accuracy": 0.8774465322494507, + "num_tokens": 46883146.0, + "step": 1206 + }, + { + "epoch": 0.1535428062587457, + "grad_norm": 3.137802839279175, + "learning_rate": 5.112335735481135e-07, + "loss": 0.4382, + "mean_token_accuracy": 0.8573218584060669, + "num_tokens": 46917949.0, + "step": 1207 + }, + { + "epoch": 0.15367001653733622, + "grad_norm": 3.2841646671295166, + "learning_rate": 5.116574819838915e-07, + "loss": 0.4255, + "mean_token_accuracy": 0.8581168055534363, + "num_tokens": 46958971.0, + "step": 1208 + }, + { + "epoch": 0.15379722681592672, + "grad_norm": 2.6674485206604004, + "learning_rate": 5.120813904196693e-07, + "loss": 0.3825, + "mean_token_accuracy": 0.8688312768936157, + "num_tokens": 47005124.0, + "step": 1209 + }, + { + "epoch": 0.15392443709451722, + "grad_norm": 2.903398036956787, + "learning_rate": 5.125052988554473e-07, + "loss": 0.4147, + "mean_token_accuracy": 0.860895574092865, + "num_tokens": 47045817.0, + "step": 1210 + }, + { + "epoch": 0.15405164737310775, + "grad_norm": 2.868513584136963, + "learning_rate": 5.12929207291225e-07, + "loss": 0.4306, + "mean_token_accuracy": 0.8539302349090576, + "num_tokens": 47086665.0, + "step": 1211 + }, + { + "epoch": 0.15417885765169825, + "grad_norm": 2.821481227874756, + "learning_rate": 5.133531157270029e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8629249334335327, + "num_tokens": 47123019.0, + "step": 1212 + }, + { + "epoch": 0.15430606793028875, + "grad_norm": 2.9763967990875244, + "learning_rate": 5.137770241627808e-07, + "loss": 0.4353, + "mean_token_accuracy": 0.8530818223953247, + "num_tokens": 47162211.0, + "step": 1213 + }, + { + "epoch": 0.15443327820887928, + "grad_norm": 2.909301996231079, + "learning_rate": 5.142009325985587e-07, + "loss": 0.4719, + "mean_token_accuracy": 0.8434101343154907, + "num_tokens": 47203902.0, + "step": 1214 + }, + { + "epoch": 0.15456048848746978, + "grad_norm": 2.6606686115264893, + "learning_rate": 5.146248410343365e-07, + "loss": 0.4184, + "mean_token_accuracy": 0.8614519834518433, + "num_tokens": 47249313.0, + "step": 1215 + }, + { + "epoch": 0.1546876987660603, + "grad_norm": 2.7785840034484863, + "learning_rate": 5.150487494701145e-07, + "loss": 0.3925, + "mean_token_accuracy": 0.8662320375442505, + "num_tokens": 47287953.0, + "step": 1216 + }, + { + "epoch": 0.1548149090446508, + "grad_norm": 2.5656771659851074, + "learning_rate": 5.154726579058923e-07, + "loss": 0.3572, + "mean_token_accuracy": 0.8796482086181641, + "num_tokens": 47327797.0, + "step": 1217 + }, + { + "epoch": 0.1549421193232413, + "grad_norm": 2.751492738723755, + "learning_rate": 5.158965663416703e-07, + "loss": 0.3297, + "mean_token_accuracy": 0.8844126462936401, + "num_tokens": 47362200.0, + "step": 1218 + }, + { + "epoch": 0.15506932960183184, + "grad_norm": 3.083059549331665, + "learning_rate": 5.16320474777448e-07, + "loss": 0.4427, + "mean_token_accuracy": 0.8508242964744568, + "num_tokens": 47397228.0, + "step": 1219 + }, + { + "epoch": 0.15519653988042234, + "grad_norm": 2.9984450340270996, + "learning_rate": 5.167443832132259e-07, + "loss": 0.4679, + "mean_token_accuracy": 0.844996452331543, + "num_tokens": 47434249.0, + "step": 1220 + }, + { + "epoch": 0.15532375015901284, + "grad_norm": 2.900364637374878, + "learning_rate": 5.171682916490038e-07, + "loss": 0.4589, + "mean_token_accuracy": 0.8464218378067017, + "num_tokens": 47475973.0, + "step": 1221 + }, + { + "epoch": 0.15545096043760337, + "grad_norm": 2.8222362995147705, + "learning_rate": 5.175922000847816e-07, + "loss": 0.4179, + "mean_token_accuracy": 0.8565013408660889, + "num_tokens": 47516197.0, + "step": 1222 + }, + { + "epoch": 0.15557817071619387, + "grad_norm": 2.753033399581909, + "learning_rate": 5.180161085205595e-07, + "loss": 0.3918, + "mean_token_accuracy": 0.8683056831359863, + "num_tokens": 47558766.0, + "step": 1223 + }, + { + "epoch": 0.15570538099478437, + "grad_norm": 2.7943975925445557, + "learning_rate": 5.184400169563374e-07, + "loss": 0.3817, + "mean_token_accuracy": 0.8721527457237244, + "num_tokens": 47596905.0, + "step": 1224 + }, + { + "epoch": 0.1558325912733749, + "grad_norm": 2.7365729808807373, + "learning_rate": 5.188639253921153e-07, + "loss": 0.3852, + "mean_token_accuracy": 0.868621826171875, + "num_tokens": 47635777.0, + "step": 1225 + }, + { + "epoch": 0.1559598015519654, + "grad_norm": 3.032707929611206, + "learning_rate": 5.192878338278932e-07, + "loss": 0.4023, + "mean_token_accuracy": 0.8636321425437927, + "num_tokens": 47669703.0, + "step": 1226 + }, + { + "epoch": 0.1560870118305559, + "grad_norm": 2.8021645545959473, + "learning_rate": 5.19711742263671e-07, + "loss": 0.4216, + "mean_token_accuracy": 0.8595715761184692, + "num_tokens": 47712353.0, + "step": 1227 + }, + { + "epoch": 0.15621422210914643, + "grad_norm": 3.080228328704834, + "learning_rate": 5.201356506994488e-07, + "loss": 0.4405, + "mean_token_accuracy": 0.8487101197242737, + "num_tokens": 47751856.0, + "step": 1228 + }, + { + "epoch": 0.15634143238773693, + "grad_norm": 2.8149521350860596, + "learning_rate": 5.205595591352268e-07, + "loss": 0.4121, + "mean_token_accuracy": 0.8597578406333923, + "num_tokens": 47790985.0, + "step": 1229 + }, + { + "epoch": 0.15646864266632743, + "grad_norm": 2.958489418029785, + "learning_rate": 5.209834675710046e-07, + "loss": 0.4185, + "mean_token_accuracy": 0.8595181703567505, + "num_tokens": 47829213.0, + "step": 1230 + }, + { + "epoch": 0.15659585294491796, + "grad_norm": 2.9028635025024414, + "learning_rate": 5.214073760067825e-07, + "loss": 0.4004, + "mean_token_accuracy": 0.8631324768066406, + "num_tokens": 47863576.0, + "step": 1231 + }, + { + "epoch": 0.15672306322350846, + "grad_norm": 3.157177686691284, + "learning_rate": 5.218312844425604e-07, + "loss": 0.4242, + "mean_token_accuracy": 0.8564239740371704, + "num_tokens": 47899594.0, + "step": 1232 + }, + { + "epoch": 0.15685027350209896, + "grad_norm": 3.0013692378997803, + "learning_rate": 5.222551928783383e-07, + "loss": 0.3888, + "mean_token_accuracy": 0.8684278726577759, + "num_tokens": 47934427.0, + "step": 1233 + }, + { + "epoch": 0.1569774837806895, + "grad_norm": 2.7576680183410645, + "learning_rate": 5.226791013141161e-07, + "loss": 0.3767, + "mean_token_accuracy": 0.8752251863479614, + "num_tokens": 47971533.0, + "step": 1234 + }, + { + "epoch": 0.15710469405928, + "grad_norm": 2.5718891620635986, + "learning_rate": 5.23103009749894e-07, + "loss": 0.3865, + "mean_token_accuracy": 0.8697040677070618, + "num_tokens": 48015743.0, + "step": 1235 + }, + { + "epoch": 0.1572319043378705, + "grad_norm": 2.598513603210449, + "learning_rate": 5.235269181856718e-07, + "loss": 0.3963, + "mean_token_accuracy": 0.864775538444519, + "num_tokens": 48060799.0, + "step": 1236 + }, + { + "epoch": 0.15735911461646102, + "grad_norm": 2.710545301437378, + "learning_rate": 5.239508266214498e-07, + "loss": 0.382, + "mean_token_accuracy": 0.8726362586021423, + "num_tokens": 48102792.0, + "step": 1237 + }, + { + "epoch": 0.15748632489505152, + "grad_norm": 2.905125617980957, + "learning_rate": 5.243747350572276e-07, + "loss": 0.3793, + "mean_token_accuracy": 0.8699899911880493, + "num_tokens": 48136484.0, + "step": 1238 + }, + { + "epoch": 0.15761353517364202, + "grad_norm": 2.8628156185150146, + "learning_rate": 5.247986434930056e-07, + "loss": 0.4248, + "mean_token_accuracy": 0.8555216789245605, + "num_tokens": 48177301.0, + "step": 1239 + }, + { + "epoch": 0.15774074545223254, + "grad_norm": 2.771235942840576, + "learning_rate": 5.252225519287834e-07, + "loss": 0.4153, + "mean_token_accuracy": 0.8605021834373474, + "num_tokens": 48218367.0, + "step": 1240 + }, + { + "epoch": 0.15786795573082305, + "grad_norm": 3.081942319869995, + "learning_rate": 5.256464603645613e-07, + "loss": 0.436, + "mean_token_accuracy": 0.8528093099594116, + "num_tokens": 48254235.0, + "step": 1241 + }, + { + "epoch": 0.15799516600941357, + "grad_norm": 2.7672786712646484, + "learning_rate": 5.260703688003391e-07, + "loss": 0.4301, + "mean_token_accuracy": 0.8516690731048584, + "num_tokens": 48296646.0, + "step": 1242 + }, + { + "epoch": 0.15812237628800407, + "grad_norm": 2.9822585582733154, + "learning_rate": 5.26494277236117e-07, + "loss": 0.375, + "mean_token_accuracy": 0.8717691898345947, + "num_tokens": 48330746.0, + "step": 1243 + }, + { + "epoch": 0.15824958656659457, + "grad_norm": 2.768132448196411, + "learning_rate": 5.269181856718948e-07, + "loss": 0.4381, + "mean_token_accuracy": 0.8500379920005798, + "num_tokens": 48372323.0, + "step": 1244 + }, + { + "epoch": 0.1583767968451851, + "grad_norm": 2.9195199012756348, + "learning_rate": 5.273420941076727e-07, + "loss": 0.4419, + "mean_token_accuracy": 0.8535325527191162, + "num_tokens": 48409986.0, + "step": 1245 + }, + { + "epoch": 0.1585040071237756, + "grad_norm": 2.795499324798584, + "learning_rate": 5.277660025434506e-07, + "loss": 0.4609, + "mean_token_accuracy": 0.8510756492614746, + "num_tokens": 48453754.0, + "step": 1246 + }, + { + "epoch": 0.1586312174023661, + "grad_norm": 2.7026684284210205, + "learning_rate": 5.281899109792285e-07, + "loss": 0.4018, + "mean_token_accuracy": 0.8631753921508789, + "num_tokens": 48495786.0, + "step": 1247 + }, + { + "epoch": 0.15875842768095663, + "grad_norm": 2.7809338569641113, + "learning_rate": 5.286138194150064e-07, + "loss": 0.3941, + "mean_token_accuracy": 0.8648325204849243, + "num_tokens": 48534452.0, + "step": 1248 + }, + { + "epoch": 0.15888563795954713, + "grad_norm": 2.9843034744262695, + "learning_rate": 5.290377278507841e-07, + "loss": 0.4327, + "mean_token_accuracy": 0.8549185991287231, + "num_tokens": 48570349.0, + "step": 1249 + }, + { + "epoch": 0.15901284823813763, + "grad_norm": 2.7383434772491455, + "learning_rate": 5.294616362865621e-07, + "loss": 0.3971, + "mean_token_accuracy": 0.8671259880065918, + "num_tokens": 48611456.0, + "step": 1250 + }, + { + "epoch": 0.15914005851672816, + "grad_norm": 2.7372050285339355, + "learning_rate": 5.298855447223399e-07, + "loss": 0.42, + "mean_token_accuracy": 0.8592525720596313, + "num_tokens": 48651984.0, + "step": 1251 + }, + { + "epoch": 0.15926726879531866, + "grad_norm": 2.8548946380615234, + "learning_rate": 5.303094531581178e-07, + "loss": 0.3552, + "mean_token_accuracy": 0.8765920400619507, + "num_tokens": 48686910.0, + "step": 1252 + }, + { + "epoch": 0.15939447907390916, + "grad_norm": 2.9222843647003174, + "learning_rate": 5.307333615938957e-07, + "loss": 0.4032, + "mean_token_accuracy": 0.8642715811729431, + "num_tokens": 48727519.0, + "step": 1253 + }, + { + "epoch": 0.1595216893524997, + "grad_norm": 2.7891547679901123, + "learning_rate": 5.311572700296736e-07, + "loss": 0.3777, + "mean_token_accuracy": 0.8709690570831299, + "num_tokens": 48765920.0, + "step": 1254 + }, + { + "epoch": 0.1596488996310902, + "grad_norm": 2.6713509559631348, + "learning_rate": 5.315811784654515e-07, + "loss": 0.4418, + "mean_token_accuracy": 0.8503625392913818, + "num_tokens": 48812616.0, + "step": 1255 + }, + { + "epoch": 0.1597761099096807, + "grad_norm": 2.7476842403411865, + "learning_rate": 5.320050869012294e-07, + "loss": 0.409, + "mean_token_accuracy": 0.8594691753387451, + "num_tokens": 48854687.0, + "step": 1256 + }, + { + "epoch": 0.15990332018827122, + "grad_norm": 2.4225571155548096, + "learning_rate": 5.324289953370071e-07, + "loss": 0.353, + "mean_token_accuracy": 0.8814741373062134, + "num_tokens": 48903148.0, + "step": 1257 + }, + { + "epoch": 0.16003053046686172, + "grad_norm": 2.6932380199432373, + "learning_rate": 5.328529037727851e-07, + "loss": 0.4432, + "mean_token_accuracy": 0.8517329096794128, + "num_tokens": 48948303.0, + "step": 1258 + }, + { + "epoch": 0.16015774074545222, + "grad_norm": 2.9959726333618164, + "learning_rate": 5.332768122085629e-07, + "loss": 0.4346, + "mean_token_accuracy": 0.8533912301063538, + "num_tokens": 48987323.0, + "step": 1259 + }, + { + "epoch": 0.16028495102404275, + "grad_norm": 2.8645424842834473, + "learning_rate": 5.337007206443408e-07, + "loss": 0.4436, + "mean_token_accuracy": 0.8529840111732483, + "num_tokens": 49026301.0, + "step": 1260 + }, + { + "epoch": 0.16041216130263325, + "grad_norm": 2.8658130168914795, + "learning_rate": 5.341246290801187e-07, + "loss": 0.3813, + "mean_token_accuracy": 0.8703030943870544, + "num_tokens": 49067178.0, + "step": 1261 + }, + { + "epoch": 0.16053937158122375, + "grad_norm": 2.8948540687561035, + "learning_rate": 5.345485375158966e-07, + "loss": 0.4343, + "mean_token_accuracy": 0.8524728417396545, + "num_tokens": 49106772.0, + "step": 1262 + }, + { + "epoch": 0.16066658185981428, + "grad_norm": 2.9409236907958984, + "learning_rate": 5.349724459516745e-07, + "loss": 0.4581, + "mean_token_accuracy": 0.8487712144851685, + "num_tokens": 49145434.0, + "step": 1263 + }, + { + "epoch": 0.16079379213840478, + "grad_norm": 3.0283946990966797, + "learning_rate": 5.353963543874522e-07, + "loss": 0.4469, + "mean_token_accuracy": 0.8527085185050964, + "num_tokens": 49187364.0, + "step": 1264 + }, + { + "epoch": 0.1609210024169953, + "grad_norm": 2.8735013008117676, + "learning_rate": 5.358202628232301e-07, + "loss": 0.3534, + "mean_token_accuracy": 0.8752714991569519, + "num_tokens": 49223081.0, + "step": 1265 + }, + { + "epoch": 0.1610482126955858, + "grad_norm": 2.6736536026000977, + "learning_rate": 5.36244171259008e-07, + "loss": 0.367, + "mean_token_accuracy": 0.8708652853965759, + "num_tokens": 49265654.0, + "step": 1266 + }, + { + "epoch": 0.1611754229741763, + "grad_norm": 3.007664203643799, + "learning_rate": 5.366680796947859e-07, + "loss": 0.4176, + "mean_token_accuracy": 0.8593379259109497, + "num_tokens": 49301731.0, + "step": 1267 + }, + { + "epoch": 0.16130263325276684, + "grad_norm": 3.366206645965576, + "learning_rate": 5.370919881305637e-07, + "loss": 0.4243, + "mean_token_accuracy": 0.8572887182235718, + "num_tokens": 49338266.0, + "step": 1268 + }, + { + "epoch": 0.16142984353135734, + "grad_norm": 2.736684560775757, + "learning_rate": 5.375158965663417e-07, + "loss": 0.4173, + "mean_token_accuracy": 0.8541845083236694, + "num_tokens": 49383149.0, + "step": 1269 + }, + { + "epoch": 0.16155705380994784, + "grad_norm": 2.668668746948242, + "learning_rate": 5.379398050021195e-07, + "loss": 0.445, + "mean_token_accuracy": 0.8477013111114502, + "num_tokens": 49428970.0, + "step": 1270 + }, + { + "epoch": 0.16168426408853837, + "grad_norm": 2.933011531829834, + "learning_rate": 5.383637134378975e-07, + "loss": 0.376, + "mean_token_accuracy": 0.873649537563324, + "num_tokens": 49464690.0, + "step": 1271 + }, + { + "epoch": 0.16181147436712887, + "grad_norm": 2.8148131370544434, + "learning_rate": 5.387876218736752e-07, + "loss": 0.3696, + "mean_token_accuracy": 0.8748018741607666, + "num_tokens": 49500693.0, + "step": 1272 + }, + { + "epoch": 0.16193868464571937, + "grad_norm": 2.9589791297912598, + "learning_rate": 5.392115303094531e-07, + "loss": 0.3694, + "mean_token_accuracy": 0.8747802376747131, + "num_tokens": 49534412.0, + "step": 1273 + }, + { + "epoch": 0.1620658949243099, + "grad_norm": 3.0575592517852783, + "learning_rate": 5.39635438745231e-07, + "loss": 0.438, + "mean_token_accuracy": 0.8491935729980469, + "num_tokens": 49571065.0, + "step": 1274 + }, + { + "epoch": 0.1621931052029004, + "grad_norm": 2.81976056098938, + "learning_rate": 5.400593471810089e-07, + "loss": 0.3396, + "mean_token_accuracy": 0.8800583481788635, + "num_tokens": 49604432.0, + "step": 1275 + }, + { + "epoch": 0.1623203154814909, + "grad_norm": 2.764692544937134, + "learning_rate": 5.404832556167867e-07, + "loss": 0.388, + "mean_token_accuracy": 0.8695069551467896, + "num_tokens": 49646506.0, + "step": 1276 + }, + { + "epoch": 0.16244752576008142, + "grad_norm": 2.684760332107544, + "learning_rate": 5.409071640525647e-07, + "loss": 0.364, + "mean_token_accuracy": 0.8776482343673706, + "num_tokens": 49685839.0, + "step": 1277 + }, + { + "epoch": 0.16257473603867192, + "grad_norm": 2.5953104496002197, + "learning_rate": 5.413310724883425e-07, + "loss": 0.3649, + "mean_token_accuracy": 0.8742786645889282, + "num_tokens": 49728488.0, + "step": 1278 + }, + { + "epoch": 0.16270194631726242, + "grad_norm": 2.664926767349243, + "learning_rate": 5.417549809241205e-07, + "loss": 0.3905, + "mean_token_accuracy": 0.8673833608627319, + "num_tokens": 49768918.0, + "step": 1279 + }, + { + "epoch": 0.16282915659585295, + "grad_norm": 2.9183878898620605, + "learning_rate": 5.421788893598982e-07, + "loss": 0.4162, + "mean_token_accuracy": 0.8645962476730347, + "num_tokens": 49809448.0, + "step": 1280 + }, + { + "epoch": 0.16295636687444345, + "grad_norm": 2.770261526107788, + "learning_rate": 5.42602797795676e-07, + "loss": 0.3809, + "mean_token_accuracy": 0.8677325248718262, + "num_tokens": 49846639.0, + "step": 1281 + }, + { + "epoch": 0.16308357715303395, + "grad_norm": 2.956136703491211, + "learning_rate": 5.43026706231454e-07, + "loss": 0.3894, + "mean_token_accuracy": 0.864285945892334, + "num_tokens": 49881645.0, + "step": 1282 + }, + { + "epoch": 0.16321078743162448, + "grad_norm": 2.756096363067627, + "learning_rate": 5.434506146672319e-07, + "loss": 0.4013, + "mean_token_accuracy": 0.8653491735458374, + "num_tokens": 49921192.0, + "step": 1283 + }, + { + "epoch": 0.16333799771021498, + "grad_norm": 2.810070753097534, + "learning_rate": 5.438745231030097e-07, + "loss": 0.3865, + "mean_token_accuracy": 0.8666462898254395, + "num_tokens": 49959204.0, + "step": 1284 + }, + { + "epoch": 0.16346520798880548, + "grad_norm": 2.7165181636810303, + "learning_rate": 5.442984315387876e-07, + "loss": 0.4112, + "mean_token_accuracy": 0.8593453168869019, + "num_tokens": 50002622.0, + "step": 1285 + }, + { + "epoch": 0.163592418267396, + "grad_norm": 2.8887135982513428, + "learning_rate": 5.447223399745655e-07, + "loss": 0.4344, + "mean_token_accuracy": 0.8541749715805054, + "num_tokens": 50042461.0, + "step": 1286 + }, + { + "epoch": 0.1637196285459865, + "grad_norm": 2.771894931793213, + "learning_rate": 5.451462484103433e-07, + "loss": 0.4096, + "mean_token_accuracy": 0.8637093305587769, + "num_tokens": 50085610.0, + "step": 1287 + }, + { + "epoch": 0.163846838824577, + "grad_norm": 2.7873475551605225, + "learning_rate": 5.455701568461212e-07, + "loss": 0.4297, + "mean_token_accuracy": 0.8568693399429321, + "num_tokens": 50130040.0, + "step": 1288 + }, + { + "epoch": 0.16397404910316754, + "grad_norm": 2.9462101459503174, + "learning_rate": 5.45994065281899e-07, + "loss": 0.4129, + "mean_token_accuracy": 0.8617526292800903, + "num_tokens": 50167425.0, + "step": 1289 + }, + { + "epoch": 0.16410125938175804, + "grad_norm": 3.0294699668884277, + "learning_rate": 5.46417973717677e-07, + "loss": 0.3611, + "mean_token_accuracy": 0.8739824891090393, + "num_tokens": 50201765.0, + "step": 1290 + }, + { + "epoch": 0.16422846966034857, + "grad_norm": 2.867218017578125, + "learning_rate": 5.468418821534548e-07, + "loss": 0.3749, + "mean_token_accuracy": 0.8736826181411743, + "num_tokens": 50239316.0, + "step": 1291 + }, + { + "epoch": 0.16435567993893907, + "grad_norm": 2.8143417835235596, + "learning_rate": 5.472657905892327e-07, + "loss": 0.4098, + "mean_token_accuracy": 0.8611119985580444, + "num_tokens": 50280308.0, + "step": 1292 + }, + { + "epoch": 0.16448289021752957, + "grad_norm": 2.919891595840454, + "learning_rate": 5.476896990250106e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8696929216384888, + "num_tokens": 50316901.0, + "step": 1293 + }, + { + "epoch": 0.1646101004961201, + "grad_norm": 3.2456130981445312, + "learning_rate": 5.481136074607885e-07, + "loss": 0.4083, + "mean_token_accuracy": 0.8601573705673218, + "num_tokens": 50348176.0, + "step": 1294 + }, + { + "epoch": 0.1647373107747106, + "grad_norm": 2.8563671112060547, + "learning_rate": 5.485375158965663e-07, + "loss": 0.3369, + "mean_token_accuracy": 0.8811579346656799, + "num_tokens": 50383863.0, + "step": 1295 + }, + { + "epoch": 0.1648645210533011, + "grad_norm": 2.7076029777526855, + "learning_rate": 5.489614243323442e-07, + "loss": 0.3607, + "mean_token_accuracy": 0.8783336877822876, + "num_tokens": 50421382.0, + "step": 1296 + }, + { + "epoch": 0.16499173133189163, + "grad_norm": 3.0547471046447754, + "learning_rate": 5.49385332768122e-07, + "loss": 0.4579, + "mean_token_accuracy": 0.848646879196167, + "num_tokens": 50458683.0, + "step": 1297 + }, + { + "epoch": 0.16511894161048213, + "grad_norm": 3.1512978076934814, + "learning_rate": 5.498092412039e-07, + "loss": 0.4129, + "mean_token_accuracy": 0.8597244024276733, + "num_tokens": 50489429.0, + "step": 1298 + }, + { + "epoch": 0.16524615188907263, + "grad_norm": 3.311241388320923, + "learning_rate": 5.502331496396778e-07, + "loss": 0.4643, + "mean_token_accuracy": 0.8476831912994385, + "num_tokens": 50526392.0, + "step": 1299 + }, + { + "epoch": 0.16537336216766316, + "grad_norm": 2.9245223999023438, + "learning_rate": 5.506570580754557e-07, + "loss": 0.369, + "mean_token_accuracy": 0.8714891076087952, + "num_tokens": 50558952.0, + "step": 1300 + }, + { + "epoch": 0.16550057244625366, + "grad_norm": 3.075896739959717, + "learning_rate": 5.510809665112336e-07, + "loss": 0.3779, + "mean_token_accuracy": 0.8689780235290527, + "num_tokens": 50593498.0, + "step": 1301 + }, + { + "epoch": 0.16562778272484416, + "grad_norm": 2.849174737930298, + "learning_rate": 5.515048749470113e-07, + "loss": 0.408, + "mean_token_accuracy": 0.8587365746498108, + "num_tokens": 50632015.0, + "step": 1302 + }, + { + "epoch": 0.1657549930034347, + "grad_norm": 3.1833012104034424, + "learning_rate": 5.519287833827893e-07, + "loss": 0.451, + "mean_token_accuracy": 0.8500630855560303, + "num_tokens": 50669302.0, + "step": 1303 + }, + { + "epoch": 0.1658822032820252, + "grad_norm": 2.8398361206054688, + "learning_rate": 5.523526918185671e-07, + "loss": 0.3964, + "mean_token_accuracy": 0.8663714528083801, + "num_tokens": 50710387.0, + "step": 1304 + }, + { + "epoch": 0.1660094135606157, + "grad_norm": 2.609785318374634, + "learning_rate": 5.52776600254345e-07, + "loss": 0.3318, + "mean_token_accuracy": 0.8869130611419678, + "num_tokens": 50750341.0, + "step": 1305 + }, + { + "epoch": 0.16613662383920622, + "grad_norm": 2.7125532627105713, + "learning_rate": 5.532005086901229e-07, + "loss": 0.3515, + "mean_token_accuracy": 0.8753534555435181, + "num_tokens": 50786790.0, + "step": 1306 + }, + { + "epoch": 0.16626383411779672, + "grad_norm": 3.1261818408966064, + "learning_rate": 5.536244171259008e-07, + "loss": 0.4606, + "mean_token_accuracy": 0.841661810874939, + "num_tokens": 50824427.0, + "step": 1307 + }, + { + "epoch": 0.16639104439638722, + "grad_norm": 2.831486940383911, + "learning_rate": 5.540483255616786e-07, + "loss": 0.4179, + "mean_token_accuracy": 0.8636233806610107, + "num_tokens": 50868393.0, + "step": 1308 + }, + { + "epoch": 0.16651825467497774, + "grad_norm": 2.672116756439209, + "learning_rate": 5.544722339974566e-07, + "loss": 0.3761, + "mean_token_accuracy": 0.8733004331588745, + "num_tokens": 50910403.0, + "step": 1309 + }, + { + "epoch": 0.16664546495356825, + "grad_norm": 2.595560312271118, + "learning_rate": 5.548961424332343e-07, + "loss": 0.3752, + "mean_token_accuracy": 0.8727869987487793, + "num_tokens": 50950975.0, + "step": 1310 + }, + { + "epoch": 0.16677267523215875, + "grad_norm": 3.008758544921875, + "learning_rate": 5.553200508690123e-07, + "loss": 0.356, + "mean_token_accuracy": 0.8774924874305725, + "num_tokens": 50983740.0, + "step": 1311 + }, + { + "epoch": 0.16689988551074927, + "grad_norm": 3.0645432472229004, + "learning_rate": 5.557439593047901e-07, + "loss": 0.4192, + "mean_token_accuracy": 0.8579719066619873, + "num_tokens": 51024098.0, + "step": 1312 + }, + { + "epoch": 0.16702709578933977, + "grad_norm": 3.413275718688965, + "learning_rate": 5.56167867740568e-07, + "loss": 0.4532, + "mean_token_accuracy": 0.8492231965065002, + "num_tokens": 51057222.0, + "step": 1313 + }, + { + "epoch": 0.16715430606793028, + "grad_norm": 2.9352362155914307, + "learning_rate": 5.565917761763459e-07, + "loss": 0.3678, + "mean_token_accuracy": 0.8735597729682922, + "num_tokens": 51097131.0, + "step": 1314 + }, + { + "epoch": 0.1672815163465208, + "grad_norm": 2.6402952671051025, + "learning_rate": 5.570156846121238e-07, + "loss": 0.3737, + "mean_token_accuracy": 0.8713619709014893, + "num_tokens": 51139171.0, + "step": 1315 + }, + { + "epoch": 0.1674087266251113, + "grad_norm": 2.7065281867980957, + "learning_rate": 5.574395930479016e-07, + "loss": 0.3654, + "mean_token_accuracy": 0.8794978260993958, + "num_tokens": 51179283.0, + "step": 1316 + }, + { + "epoch": 0.16753593690370183, + "grad_norm": 3.0009539127349854, + "learning_rate": 5.578635014836796e-07, + "loss": 0.3736, + "mean_token_accuracy": 0.8718616962432861, + "num_tokens": 51213567.0, + "step": 1317 + }, + { + "epoch": 0.16766314718229233, + "grad_norm": 3.1072988510131836, + "learning_rate": 5.582874099194573e-07, + "loss": 0.442, + "mean_token_accuracy": 0.8518145084381104, + "num_tokens": 51250015.0, + "step": 1318 + }, + { + "epoch": 0.16779035746088283, + "grad_norm": 2.596480131149292, + "learning_rate": 5.587113183552353e-07, + "loss": 0.3505, + "mean_token_accuracy": 0.8795177936553955, + "num_tokens": 51292436.0, + "step": 1319 + }, + { + "epoch": 0.16791756773947336, + "grad_norm": 2.9689555168151855, + "learning_rate": 5.591352267910131e-07, + "loss": 0.4565, + "mean_token_accuracy": 0.8445855379104614, + "num_tokens": 51332077.0, + "step": 1320 + }, + { + "epoch": 0.16804477801806386, + "grad_norm": 3.0292835235595703, + "learning_rate": 5.59559135226791e-07, + "loss": 0.4147, + "mean_token_accuracy": 0.858607828617096, + "num_tokens": 51368474.0, + "step": 1321 + }, + { + "epoch": 0.16817198829665436, + "grad_norm": 2.9359562397003174, + "learning_rate": 5.599830436625689e-07, + "loss": 0.4183, + "mean_token_accuracy": 0.8587758541107178, + "num_tokens": 51405948.0, + "step": 1322 + }, + { + "epoch": 0.1682991985752449, + "grad_norm": 3.201153516769409, + "learning_rate": 5.604069520983468e-07, + "loss": 0.4387, + "mean_token_accuracy": 0.8431021571159363, + "num_tokens": 51441181.0, + "step": 1323 + }, + { + "epoch": 0.1684264088538354, + "grad_norm": 3.080669403076172, + "learning_rate": 5.608308605341246e-07, + "loss": 0.4197, + "mean_token_accuracy": 0.856888473033905, + "num_tokens": 51479894.0, + "step": 1324 + }, + { + "epoch": 0.1685536191324259, + "grad_norm": 2.9159939289093018, + "learning_rate": 5.612547689699024e-07, + "loss": 0.4243, + "mean_token_accuracy": 0.8539081811904907, + "num_tokens": 51517739.0, + "step": 1325 + }, + { + "epoch": 0.16868082941101642, + "grad_norm": 2.948115110397339, + "learning_rate": 5.616786774056803e-07, + "loss": 0.3953, + "mean_token_accuracy": 0.8675163984298706, + "num_tokens": 51552210.0, + "step": 1326 + }, + { + "epoch": 0.16880803968960692, + "grad_norm": 2.686558961868286, + "learning_rate": 5.621025858414582e-07, + "loss": 0.383, + "mean_token_accuracy": 0.8679966926574707, + "num_tokens": 51594590.0, + "step": 1327 + }, + { + "epoch": 0.16893524996819742, + "grad_norm": 2.872856855392456, + "learning_rate": 5.625264942772361e-07, + "loss": 0.4225, + "mean_token_accuracy": 0.8602911233901978, + "num_tokens": 51634549.0, + "step": 1328 + }, + { + "epoch": 0.16906246024678795, + "grad_norm": 3.187852382659912, + "learning_rate": 5.629504027130139e-07, + "loss": 0.4326, + "mean_token_accuracy": 0.8517052531242371, + "num_tokens": 51671931.0, + "step": 1329 + }, + { + "epoch": 0.16918967052537845, + "grad_norm": 2.80060076713562, + "learning_rate": 5.633743111487919e-07, + "loss": 0.4037, + "mean_token_accuracy": 0.8613971471786499, + "num_tokens": 51714512.0, + "step": 1330 + }, + { + "epoch": 0.16931688080396895, + "grad_norm": 3.054147720336914, + "learning_rate": 5.637982195845697e-07, + "loss": 0.3869, + "mean_token_accuracy": 0.8667600154876709, + "num_tokens": 51749880.0, + "step": 1331 + }, + { + "epoch": 0.16944409108255948, + "grad_norm": 2.8585848808288574, + "learning_rate": 5.642221280203476e-07, + "loss": 0.4388, + "mean_token_accuracy": 0.8540126085281372, + "num_tokens": 51789556.0, + "step": 1332 + }, + { + "epoch": 0.16957130136114998, + "grad_norm": 3.021646022796631, + "learning_rate": 5.646460364561254e-07, + "loss": 0.4128, + "mean_token_accuracy": 0.8606078624725342, + "num_tokens": 51824152.0, + "step": 1333 + }, + { + "epoch": 0.16969851163974048, + "grad_norm": 3.129744529724121, + "learning_rate": 5.650699448919033e-07, + "loss": 0.3446, + "mean_token_accuracy": 0.883372962474823, + "num_tokens": 51861221.0, + "step": 1334 + }, + { + "epoch": 0.169825721918331, + "grad_norm": 2.7829771041870117, + "learning_rate": 5.654938533276812e-07, + "loss": 0.4008, + "mean_token_accuracy": 0.8664668798446655, + "num_tokens": 51901651.0, + "step": 1335 + }, + { + "epoch": 0.1699529321969215, + "grad_norm": 2.8646926879882812, + "learning_rate": 5.659177617634591e-07, + "loss": 0.3957, + "mean_token_accuracy": 0.8657640814781189, + "num_tokens": 51945296.0, + "step": 1336 + }, + { + "epoch": 0.170080142475512, + "grad_norm": 2.91204833984375, + "learning_rate": 5.663416701992369e-07, + "loss": 0.4317, + "mean_token_accuracy": 0.8517526388168335, + "num_tokens": 51985993.0, + "step": 1337 + }, + { + "epoch": 0.17020735275410254, + "grad_norm": 2.7809736728668213, + "learning_rate": 5.667655786350149e-07, + "loss": 0.3799, + "mean_token_accuracy": 0.870520293712616, + "num_tokens": 52024299.0, + "step": 1338 + }, + { + "epoch": 0.17033456303269304, + "grad_norm": 2.8134467601776123, + "learning_rate": 5.671894870707927e-07, + "loss": 0.3732, + "mean_token_accuracy": 0.8732261657714844, + "num_tokens": 52064023.0, + "step": 1339 + }, + { + "epoch": 0.17046177331128357, + "grad_norm": 2.74287486076355, + "learning_rate": 5.676133955065705e-07, + "loss": 0.4046, + "mean_token_accuracy": 0.8629628419876099, + "num_tokens": 52111100.0, + "step": 1340 + }, + { + "epoch": 0.17058898358987407, + "grad_norm": 2.7637147903442383, + "learning_rate": 5.680373039423484e-07, + "loss": 0.3704, + "mean_token_accuracy": 0.8762487173080444, + "num_tokens": 52149951.0, + "step": 1341 + }, + { + "epoch": 0.17071619386846457, + "grad_norm": 2.570100784301758, + "learning_rate": 5.684612123781263e-07, + "loss": 0.3517, + "mean_token_accuracy": 0.8788602352142334, + "num_tokens": 52189756.0, + "step": 1342 + }, + { + "epoch": 0.1708434041470551, + "grad_norm": 3.1589231491088867, + "learning_rate": 5.688851208139042e-07, + "loss": 0.4003, + "mean_token_accuracy": 0.862512469291687, + "num_tokens": 52229353.0, + "step": 1343 + }, + { + "epoch": 0.1709706144256456, + "grad_norm": 3.1415719985961914, + "learning_rate": 5.69309029249682e-07, + "loss": 0.4398, + "mean_token_accuracy": 0.8508164286613464, + "num_tokens": 52263913.0, + "step": 1344 + }, + { + "epoch": 0.1710978247042361, + "grad_norm": 2.560457468032837, + "learning_rate": 5.697329376854599e-07, + "loss": 0.4083, + "mean_token_accuracy": 0.8602135181427002, + "num_tokens": 52310260.0, + "step": 1345 + }, + { + "epoch": 0.17122503498282662, + "grad_norm": 2.8450686931610107, + "learning_rate": 5.701568461212378e-07, + "loss": 0.3844, + "mean_token_accuracy": 0.8704861998558044, + "num_tokens": 52354223.0, + "step": 1346 + }, + { + "epoch": 0.17135224526141712, + "grad_norm": 2.983407497406006, + "learning_rate": 5.705807545570157e-07, + "loss": 0.4224, + "mean_token_accuracy": 0.8585700988769531, + "num_tokens": 52389629.0, + "step": 1347 + }, + { + "epoch": 0.17147945554000762, + "grad_norm": 2.8769772052764893, + "learning_rate": 5.710046629927934e-07, + "loss": 0.3897, + "mean_token_accuracy": 0.863615095615387, + "num_tokens": 52426884.0, + "step": 1348 + }, + { + "epoch": 0.17160666581859815, + "grad_norm": 2.7605199813842773, + "learning_rate": 5.714285714285714e-07, + "loss": 0.4356, + "mean_token_accuracy": 0.8536889553070068, + "num_tokens": 52473180.0, + "step": 1349 + }, + { + "epoch": 0.17173387609718865, + "grad_norm": 2.956683397293091, + "learning_rate": 5.718524798643492e-07, + "loss": 0.4152, + "mean_token_accuracy": 0.8647564649581909, + "num_tokens": 52511009.0, + "step": 1350 + }, + { + "epoch": 0.17186108637577915, + "grad_norm": 3.082913398742676, + "learning_rate": 5.722763883001272e-07, + "loss": 0.3848, + "mean_token_accuracy": 0.8727186918258667, + "num_tokens": 52545148.0, + "step": 1351 + }, + { + "epoch": 0.17198829665436968, + "grad_norm": 3.277024984359741, + "learning_rate": 5.72700296735905e-07, + "loss": 0.3875, + "mean_token_accuracy": 0.8684790134429932, + "num_tokens": 52574364.0, + "step": 1352 + }, + { + "epoch": 0.17211550693296018, + "grad_norm": 2.905527114868164, + "learning_rate": 5.731242051716829e-07, + "loss": 0.4413, + "mean_token_accuracy": 0.8520529270172119, + "num_tokens": 52613464.0, + "step": 1353 + }, + { + "epoch": 0.17224271721155068, + "grad_norm": 2.8507778644561768, + "learning_rate": 5.735481136074608e-07, + "loss": 0.4086, + "mean_token_accuracy": 0.8592876195907593, + "num_tokens": 52654669.0, + "step": 1354 + }, + { + "epoch": 0.1723699274901412, + "grad_norm": 2.759650945663452, + "learning_rate": 5.739720220432386e-07, + "loss": 0.3898, + "mean_token_accuracy": 0.8656152486801147, + "num_tokens": 52693297.0, + "step": 1355 + }, + { + "epoch": 0.1724971377687317, + "grad_norm": 2.877312660217285, + "learning_rate": 5.743959304790164e-07, + "loss": 0.3943, + "mean_token_accuracy": 0.8662331700325012, + "num_tokens": 52733497.0, + "step": 1356 + }, + { + "epoch": 0.1726243480473222, + "grad_norm": 2.814249038696289, + "learning_rate": 5.748198389147944e-07, + "loss": 0.4149, + "mean_token_accuracy": 0.8597040176391602, + "num_tokens": 52773660.0, + "step": 1357 + }, + { + "epoch": 0.17275155832591274, + "grad_norm": 2.8756022453308105, + "learning_rate": 5.752437473505722e-07, + "loss": 0.3857, + "mean_token_accuracy": 0.8673120141029358, + "num_tokens": 52811018.0, + "step": 1358 + }, + { + "epoch": 0.17287876860450324, + "grad_norm": 2.7554373741149902, + "learning_rate": 5.756676557863502e-07, + "loss": 0.4209, + "mean_token_accuracy": 0.8587183952331543, + "num_tokens": 52852148.0, + "step": 1359 + }, + { + "epoch": 0.17300597888309374, + "grad_norm": 2.819667100906372, + "learning_rate": 5.76091564222128e-07, + "loss": 0.3804, + "mean_token_accuracy": 0.8715484142303467, + "num_tokens": 52886873.0, + "step": 1360 + }, + { + "epoch": 0.17313318916168427, + "grad_norm": 2.4375393390655518, + "learning_rate": 5.765154726579059e-07, + "loss": 0.3504, + "mean_token_accuracy": 0.879844069480896, + "num_tokens": 52931866.0, + "step": 1361 + }, + { + "epoch": 0.17326039944027477, + "grad_norm": 2.8010764122009277, + "learning_rate": 5.769393810936838e-07, + "loss": 0.3639, + "mean_token_accuracy": 0.8769832253456116, + "num_tokens": 52967972.0, + "step": 1362 + }, + { + "epoch": 0.17338760971886527, + "grad_norm": 2.860783576965332, + "learning_rate": 5.773632895294616e-07, + "loss": 0.4061, + "mean_token_accuracy": 0.8571947813034058, + "num_tokens": 53004701.0, + "step": 1363 + }, + { + "epoch": 0.1735148199974558, + "grad_norm": 3.045208692550659, + "learning_rate": 5.777871979652394e-07, + "loss": 0.4179, + "mean_token_accuracy": 0.8585977554321289, + "num_tokens": 53040570.0, + "step": 1364 + }, + { + "epoch": 0.1736420302760463, + "grad_norm": 3.07670259475708, + "learning_rate": 5.782111064010173e-07, + "loss": 0.3992, + "mean_token_accuracy": 0.8647662401199341, + "num_tokens": 53072554.0, + "step": 1365 + }, + { + "epoch": 0.17376924055463683, + "grad_norm": 2.6615726947784424, + "learning_rate": 5.786350148367952e-07, + "loss": 0.4022, + "mean_token_accuracy": 0.8644769191741943, + "num_tokens": 53119368.0, + "step": 1366 + }, + { + "epoch": 0.17389645083322733, + "grad_norm": 3.059166193008423, + "learning_rate": 5.790589232725731e-07, + "loss": 0.3753, + "mean_token_accuracy": 0.8712415099143982, + "num_tokens": 53154686.0, + "step": 1367 + }, + { + "epoch": 0.17402366111181783, + "grad_norm": 2.9604334831237793, + "learning_rate": 5.79482831708351e-07, + "loss": 0.4297, + "mean_token_accuracy": 0.8544814586639404, + "num_tokens": 53193409.0, + "step": 1368 + }, + { + "epoch": 0.17415087139040836, + "grad_norm": 2.973973274230957, + "learning_rate": 5.799067401441288e-07, + "loss": 0.4171, + "mean_token_accuracy": 0.859604001045227, + "num_tokens": 53228555.0, + "step": 1369 + }, + { + "epoch": 0.17427808166899886, + "grad_norm": 2.9460232257843018, + "learning_rate": 5.803306485799068e-07, + "loss": 0.4363, + "mean_token_accuracy": 0.8539038896560669, + "num_tokens": 53267246.0, + "step": 1370 + }, + { + "epoch": 0.17440529194758936, + "grad_norm": 2.7423343658447266, + "learning_rate": 5.807545570156845e-07, + "loss": 0.3816, + "mean_token_accuracy": 0.867711067199707, + "num_tokens": 53307742.0, + "step": 1371 + }, + { + "epoch": 0.1745325022261799, + "grad_norm": 3.1293888092041016, + "learning_rate": 5.811784654514624e-07, + "loss": 0.4111, + "mean_token_accuracy": 0.8584738969802856, + "num_tokens": 53342295.0, + "step": 1372 + }, + { + "epoch": 0.1746597125047704, + "grad_norm": 2.8238301277160645, + "learning_rate": 5.816023738872403e-07, + "loss": 0.3743, + "mean_token_accuracy": 0.8698130249977112, + "num_tokens": 53381141.0, + "step": 1373 + }, + { + "epoch": 0.1747869227833609, + "grad_norm": 3.1249125003814697, + "learning_rate": 5.820262823230182e-07, + "loss": 0.4111, + "mean_token_accuracy": 0.8627212047576904, + "num_tokens": 53414753.0, + "step": 1374 + }, + { + "epoch": 0.17491413306195142, + "grad_norm": 2.795491933822632, + "learning_rate": 5.824501907587961e-07, + "loss": 0.4162, + "mean_token_accuracy": 0.861578106880188, + "num_tokens": 53455985.0, + "step": 1375 + }, + { + "epoch": 0.17504134334054192, + "grad_norm": 3.0577328205108643, + "learning_rate": 5.82874099194574e-07, + "loss": 0.4439, + "mean_token_accuracy": 0.8518282175064087, + "num_tokens": 53493567.0, + "step": 1376 + }, + { + "epoch": 0.17516855361913242, + "grad_norm": 2.622715711593628, + "learning_rate": 5.832980076303518e-07, + "loss": 0.3603, + "mean_token_accuracy": 0.8760230541229248, + "num_tokens": 53531968.0, + "step": 1377 + }, + { + "epoch": 0.17529576389772294, + "grad_norm": 2.592963933944702, + "learning_rate": 5.837219160661297e-07, + "loss": 0.4074, + "mean_token_accuracy": 0.8624589443206787, + "num_tokens": 53581252.0, + "step": 1378 + }, + { + "epoch": 0.17542297417631345, + "grad_norm": 3.2607197761535645, + "learning_rate": 5.841458245019075e-07, + "loss": 0.4652, + "mean_token_accuracy": 0.8440889120101929, + "num_tokens": 53616988.0, + "step": 1379 + }, + { + "epoch": 0.17555018445490395, + "grad_norm": 2.732557535171509, + "learning_rate": 5.845697329376855e-07, + "loss": 0.407, + "mean_token_accuracy": 0.861454427242279, + "num_tokens": 53657763.0, + "step": 1380 + }, + { + "epoch": 0.17567739473349447, + "grad_norm": 2.64965558052063, + "learning_rate": 5.849936413734633e-07, + "loss": 0.3642, + "mean_token_accuracy": 0.8745291233062744, + "num_tokens": 53699107.0, + "step": 1381 + }, + { + "epoch": 0.17580460501208497, + "grad_norm": 2.803971767425537, + "learning_rate": 5.854175498092412e-07, + "loss": 0.365, + "mean_token_accuracy": 0.8744229078292847, + "num_tokens": 53735363.0, + "step": 1382 + }, + { + "epoch": 0.17593181529067548, + "grad_norm": 2.837045431137085, + "learning_rate": 5.858414582450191e-07, + "loss": 0.3702, + "mean_token_accuracy": 0.8725489974021912, + "num_tokens": 53771350.0, + "step": 1383 + }, + { + "epoch": 0.176059025569266, + "grad_norm": 2.8922340869903564, + "learning_rate": 5.86265366680797e-07, + "loss": 0.4057, + "mean_token_accuracy": 0.8610793352127075, + "num_tokens": 53810689.0, + "step": 1384 + }, + { + "epoch": 0.1761862358478565, + "grad_norm": 3.133445978164673, + "learning_rate": 5.866892751165748e-07, + "loss": 0.5006, + "mean_token_accuracy": 0.8354040384292603, + "num_tokens": 53850616.0, + "step": 1385 + }, + { + "epoch": 0.176313446126447, + "grad_norm": 2.9072563648223877, + "learning_rate": 5.871131835523526e-07, + "loss": 0.4274, + "mean_token_accuracy": 0.8569103479385376, + "num_tokens": 53890799.0, + "step": 1386 + }, + { + "epoch": 0.17644065640503753, + "grad_norm": 2.745582342147827, + "learning_rate": 5.875370919881305e-07, + "loss": 0.3642, + "mean_token_accuracy": 0.8729698657989502, + "num_tokens": 53929201.0, + "step": 1387 + }, + { + "epoch": 0.17656786668362803, + "grad_norm": 2.7764997482299805, + "learning_rate": 5.879610004239084e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8580284714698792, + "num_tokens": 53971351.0, + "step": 1388 + }, + { + "epoch": 0.17669507696221853, + "grad_norm": 3.10215163230896, + "learning_rate": 5.883849088596863e-07, + "loss": 0.4316, + "mean_token_accuracy": 0.8523843288421631, + "num_tokens": 54009173.0, + "step": 1389 + }, + { + "epoch": 0.17682228724080906, + "grad_norm": 2.9873666763305664, + "learning_rate": 5.888088172954641e-07, + "loss": 0.3646, + "mean_token_accuracy": 0.8761872053146362, + "num_tokens": 54044548.0, + "step": 1390 + }, + { + "epoch": 0.17694949751939956, + "grad_norm": 2.9686460494995117, + "learning_rate": 5.892327257312421e-07, + "loss": 0.4056, + "mean_token_accuracy": 0.8649898767471313, + "num_tokens": 54081325.0, + "step": 1391 + }, + { + "epoch": 0.1770767077979901, + "grad_norm": 3.3495805263519287, + "learning_rate": 5.896566341670199e-07, + "loss": 0.3882, + "mean_token_accuracy": 0.8678677678108215, + "num_tokens": 54121941.0, + "step": 1392 + }, + { + "epoch": 0.1772039180765806, + "grad_norm": 2.779975175857544, + "learning_rate": 5.900805426027977e-07, + "loss": 0.4033, + "mean_token_accuracy": 0.866204023361206, + "num_tokens": 54162959.0, + "step": 1393 + }, + { + "epoch": 0.1773311283551711, + "grad_norm": 2.940636396408081, + "learning_rate": 5.905044510385756e-07, + "loss": 0.4235, + "mean_token_accuracy": 0.8552618026733398, + "num_tokens": 54200703.0, + "step": 1394 + }, + { + "epoch": 0.17745833863376162, + "grad_norm": 3.1486992835998535, + "learning_rate": 5.909283594743535e-07, + "loss": 0.4151, + "mean_token_accuracy": 0.8600064516067505, + "num_tokens": 54235687.0, + "step": 1395 + }, + { + "epoch": 0.17758554891235212, + "grad_norm": 2.5055038928985596, + "learning_rate": 5.913522679101314e-07, + "loss": 0.3552, + "mean_token_accuracy": 0.877987802028656, + "num_tokens": 54279479.0, + "step": 1396 + }, + { + "epoch": 0.17771275919094262, + "grad_norm": 3.02378249168396, + "learning_rate": 5.917761763459093e-07, + "loss": 0.3675, + "mean_token_accuracy": 0.8733661770820618, + "num_tokens": 54312075.0, + "step": 1397 + }, + { + "epoch": 0.17783996946953315, + "grad_norm": 3.9557042121887207, + "learning_rate": 5.922000847816871e-07, + "loss": 0.3621, + "mean_token_accuracy": 0.8763946890830994, + "num_tokens": 54352988.0, + "step": 1398 + }, + { + "epoch": 0.17796717974812365, + "grad_norm": 2.8933184146881104, + "learning_rate": 5.926239932174651e-07, + "loss": 0.4249, + "mean_token_accuracy": 0.8570479154586792, + "num_tokens": 54395542.0, + "step": 1399 + }, + { + "epoch": 0.17809439002671415, + "grad_norm": 2.603037118911743, + "learning_rate": 5.930479016532429e-07, + "loss": 0.388, + "mean_token_accuracy": 0.8682650327682495, + "num_tokens": 54440219.0, + "step": 1400 + }, + { + "epoch": 0.17822160030530468, + "grad_norm": 3.0032119750976562, + "learning_rate": 5.934718100890207e-07, + "loss": 0.4324, + "mean_token_accuracy": 0.8553858995437622, + "num_tokens": 54476722.0, + "step": 1401 + }, + { + "epoch": 0.17834881058389518, + "grad_norm": 2.62703013420105, + "learning_rate": 5.938957185247986e-07, + "loss": 0.3628, + "mean_token_accuracy": 0.877007246017456, + "num_tokens": 54517320.0, + "step": 1402 + }, + { + "epoch": 0.17847602086248568, + "grad_norm": 2.9860260486602783, + "learning_rate": 5.943196269605765e-07, + "loss": 0.4475, + "mean_token_accuracy": 0.8512412905693054, + "num_tokens": 54556015.0, + "step": 1403 + }, + { + "epoch": 0.1786032311410762, + "grad_norm": 2.4496028423309326, + "learning_rate": 5.947435353963544e-07, + "loss": 0.3387, + "mean_token_accuracy": 0.8821178674697876, + "num_tokens": 54601216.0, + "step": 1404 + }, + { + "epoch": 0.1787304414196667, + "grad_norm": 2.8833377361297607, + "learning_rate": 5.951674438321323e-07, + "loss": 0.3757, + "mean_token_accuracy": 0.8721208572387695, + "num_tokens": 54635243.0, + "step": 1405 + }, + { + "epoch": 0.1788576516982572, + "grad_norm": 2.7708940505981445, + "learning_rate": 5.955913522679101e-07, + "loss": 0.392, + "mean_token_accuracy": 0.8643758296966553, + "num_tokens": 54673147.0, + "step": 1406 + }, + { + "epoch": 0.17898486197684774, + "grad_norm": 2.71010422706604, + "learning_rate": 5.96015260703688e-07, + "loss": 0.3796, + "mean_token_accuracy": 0.8694393634796143, + "num_tokens": 54712267.0, + "step": 1407 + }, + { + "epoch": 0.17911207225543824, + "grad_norm": 2.456280469894409, + "learning_rate": 5.964391691394659e-07, + "loss": 0.3621, + "mean_token_accuracy": 0.8747223019599915, + "num_tokens": 54755465.0, + "step": 1408 + }, + { + "epoch": 0.17923928253402874, + "grad_norm": 2.7064273357391357, + "learning_rate": 5.968630775752436e-07, + "loss": 0.3969, + "mean_token_accuracy": 0.8616200089454651, + "num_tokens": 54798401.0, + "step": 1409 + }, + { + "epoch": 0.17936649281261927, + "grad_norm": 2.7162985801696777, + "learning_rate": 5.972869860110216e-07, + "loss": 0.3764, + "mean_token_accuracy": 0.8702664375305176, + "num_tokens": 54839077.0, + "step": 1410 + }, + { + "epoch": 0.17949370309120977, + "grad_norm": 2.830904722213745, + "learning_rate": 5.977108944467994e-07, + "loss": 0.4099, + "mean_token_accuracy": 0.8581820130348206, + "num_tokens": 54879147.0, + "step": 1411 + }, + { + "epoch": 0.17962091336980027, + "grad_norm": 2.7926297187805176, + "learning_rate": 5.981348028825774e-07, + "loss": 0.3765, + "mean_token_accuracy": 0.8704689741134644, + "num_tokens": 54918707.0, + "step": 1412 + }, + { + "epoch": 0.1797481236483908, + "grad_norm": 2.646573543548584, + "learning_rate": 5.985587113183552e-07, + "loss": 0.3576, + "mean_token_accuracy": 0.8794748783111572, + "num_tokens": 54957069.0, + "step": 1413 + }, + { + "epoch": 0.1798753339269813, + "grad_norm": 2.8575944900512695, + "learning_rate": 5.989826197541331e-07, + "loss": 0.3761, + "mean_token_accuracy": 0.8717994093894958, + "num_tokens": 54994134.0, + "step": 1414 + }, + { + "epoch": 0.18000254420557182, + "grad_norm": 3.1311113834381104, + "learning_rate": 5.99406528189911e-07, + "loss": 0.3812, + "mean_token_accuracy": 0.8710450530052185, + "num_tokens": 55024989.0, + "step": 1415 + }, + { + "epoch": 0.18012975448416232, + "grad_norm": 3.0360782146453857, + "learning_rate": 5.998304366256888e-07, + "loss": 0.4259, + "mean_token_accuracy": 0.8517593741416931, + "num_tokens": 55062324.0, + "step": 1416 + }, + { + "epoch": 0.18025696476275282, + "grad_norm": 2.8878955841064453, + "learning_rate": 6.002543450614666e-07, + "loss": 0.3558, + "mean_token_accuracy": 0.8764629364013672, + "num_tokens": 55096850.0, + "step": 1417 + }, + { + "epoch": 0.18038417504134335, + "grad_norm": 2.9072036743164062, + "learning_rate": 6.006782534972446e-07, + "loss": 0.399, + "mean_token_accuracy": 0.864393413066864, + "num_tokens": 55135997.0, + "step": 1418 + }, + { + "epoch": 0.18051138531993385, + "grad_norm": 3.0240814685821533, + "learning_rate": 6.011021619330224e-07, + "loss": 0.4367, + "mean_token_accuracy": 0.852275013923645, + "num_tokens": 55173183.0, + "step": 1419 + }, + { + "epoch": 0.18063859559852435, + "grad_norm": 2.842078685760498, + "learning_rate": 6.015260703688004e-07, + "loss": 0.4076, + "mean_token_accuracy": 0.8624132871627808, + "num_tokens": 55215673.0, + "step": 1420 + }, + { + "epoch": 0.18076580587711488, + "grad_norm": 2.852297067642212, + "learning_rate": 6.019499788045782e-07, + "loss": 0.3881, + "mean_token_accuracy": 0.8663378953933716, + "num_tokens": 55252911.0, + "step": 1421 + }, + { + "epoch": 0.18089301615570538, + "grad_norm": 2.739847421646118, + "learning_rate": 6.023738872403561e-07, + "loss": 0.3862, + "mean_token_accuracy": 0.8689911961555481, + "num_tokens": 55294032.0, + "step": 1422 + }, + { + "epoch": 0.18102022643429588, + "grad_norm": 3.017237663269043, + "learning_rate": 6.02797795676134e-07, + "loss": 0.4145, + "mean_token_accuracy": 0.8572977781295776, + "num_tokens": 55330162.0, + "step": 1423 + }, + { + "epoch": 0.1811474367128864, + "grad_norm": 2.7210681438446045, + "learning_rate": 6.032217041119118e-07, + "loss": 0.3788, + "mean_token_accuracy": 0.8712959289550781, + "num_tokens": 55370738.0, + "step": 1424 + }, + { + "epoch": 0.1812746469914769, + "grad_norm": 2.9935357570648193, + "learning_rate": 6.036456125476896e-07, + "loss": 0.433, + "mean_token_accuracy": 0.8592502474784851, + "num_tokens": 55407204.0, + "step": 1425 + }, + { + "epoch": 0.1814018572700674, + "grad_norm": 2.768775463104248, + "learning_rate": 6.040695209834675e-07, + "loss": 0.4342, + "mean_token_accuracy": 0.8531688451766968, + "num_tokens": 55448483.0, + "step": 1426 + }, + { + "epoch": 0.18152906754865794, + "grad_norm": 2.9138712882995605, + "learning_rate": 6.044934294192454e-07, + "loss": 0.4052, + "mean_token_accuracy": 0.8584822416305542, + "num_tokens": 55485350.0, + "step": 1427 + }, + { + "epoch": 0.18165627782724844, + "grad_norm": 2.665792942047119, + "learning_rate": 6.049173378550233e-07, + "loss": 0.3923, + "mean_token_accuracy": 0.8687962293624878, + "num_tokens": 55526714.0, + "step": 1428 + }, + { + "epoch": 0.18178348810583894, + "grad_norm": 2.7604098320007324, + "learning_rate": 6.053412462908012e-07, + "loss": 0.3723, + "mean_token_accuracy": 0.8714674711227417, + "num_tokens": 55564371.0, + "step": 1429 + }, + { + "epoch": 0.18191069838442947, + "grad_norm": 2.5510149002075195, + "learning_rate": 6.05765154726579e-07, + "loss": 0.3607, + "mean_token_accuracy": 0.8768192529678345, + "num_tokens": 55610321.0, + "step": 1430 + }, + { + "epoch": 0.18203790866301997, + "grad_norm": 2.916029453277588, + "learning_rate": 6.061890631623569e-07, + "loss": 0.3561, + "mean_token_accuracy": 0.8790686130523682, + "num_tokens": 55647535.0, + "step": 1431 + }, + { + "epoch": 0.18216511894161047, + "grad_norm": 2.9646716117858887, + "learning_rate": 6.066129715981347e-07, + "loss": 0.4306, + "mean_token_accuracy": 0.8555878400802612, + "num_tokens": 55687044.0, + "step": 1432 + }, + { + "epoch": 0.182292329220201, + "grad_norm": 2.897167444229126, + "learning_rate": 6.070368800339126e-07, + "loss": 0.4429, + "mean_token_accuracy": 0.8517294526100159, + "num_tokens": 55726858.0, + "step": 1433 + }, + { + "epoch": 0.1824195394987915, + "grad_norm": 2.6506083011627197, + "learning_rate": 6.074607884696905e-07, + "loss": 0.3701, + "mean_token_accuracy": 0.8723688721656799, + "num_tokens": 55766883.0, + "step": 1434 + }, + { + "epoch": 0.182546749777382, + "grad_norm": 2.7252037525177, + "learning_rate": 6.078846969054684e-07, + "loss": 0.3769, + "mean_token_accuracy": 0.8699120283126831, + "num_tokens": 55807166.0, + "step": 1435 + }, + { + "epoch": 0.18267396005597253, + "grad_norm": 2.7160990238189697, + "learning_rate": 6.083086053412463e-07, + "loss": 0.4307, + "mean_token_accuracy": 0.8597416877746582, + "num_tokens": 55850913.0, + "step": 1436 + }, + { + "epoch": 0.18280117033456303, + "grad_norm": 2.9861655235290527, + "learning_rate": 6.087325137770242e-07, + "loss": 0.4445, + "mean_token_accuracy": 0.8564268350601196, + "num_tokens": 55887951.0, + "step": 1437 + }, + { + "epoch": 0.18292838061315353, + "grad_norm": 3.122870445251465, + "learning_rate": 6.09156422212802e-07, + "loss": 0.4063, + "mean_token_accuracy": 0.8615982532501221, + "num_tokens": 55919749.0, + "step": 1438 + }, + { + "epoch": 0.18305559089174406, + "grad_norm": 2.784308433532715, + "learning_rate": 6.095803306485799e-07, + "loss": 0.4667, + "mean_token_accuracy": 0.8437666893005371, + "num_tokens": 55964241.0, + "step": 1439 + }, + { + "epoch": 0.18318280117033456, + "grad_norm": 2.8354361057281494, + "learning_rate": 6.100042390843577e-07, + "loss": 0.3963, + "mean_token_accuracy": 0.8648825287818909, + "num_tokens": 56002173.0, + "step": 1440 + }, + { + "epoch": 0.1833100114489251, + "grad_norm": 2.5496881008148193, + "learning_rate": 6.104281475201356e-07, + "loss": 0.3556, + "mean_token_accuracy": 0.8785814046859741, + "num_tokens": 56044833.0, + "step": 1441 + }, + { + "epoch": 0.1834372217275156, + "grad_norm": 2.6441051959991455, + "learning_rate": 6.108520559559135e-07, + "loss": 0.3606, + "mean_token_accuracy": 0.8791510462760925, + "num_tokens": 56084733.0, + "step": 1442 + }, + { + "epoch": 0.1835644320061061, + "grad_norm": 2.660372257232666, + "learning_rate": 6.112759643916914e-07, + "loss": 0.3825, + "mean_token_accuracy": 0.8682706356048584, + "num_tokens": 56125846.0, + "step": 1443 + }, + { + "epoch": 0.18369164228469662, + "grad_norm": 2.6283395290374756, + "learning_rate": 6.116998728274693e-07, + "loss": 0.3811, + "mean_token_accuracy": 0.869216799736023, + "num_tokens": 56170344.0, + "step": 1444 + }, + { + "epoch": 0.18381885256328712, + "grad_norm": 3.4332213401794434, + "learning_rate": 6.121237812632472e-07, + "loss": 0.4315, + "mean_token_accuracy": 0.8496399521827698, + "num_tokens": 56201499.0, + "step": 1445 + }, + { + "epoch": 0.18394606284187762, + "grad_norm": 2.988725423812866, + "learning_rate": 6.125476896990249e-07, + "loss": 0.4531, + "mean_token_accuracy": 0.8510436415672302, + "num_tokens": 56240096.0, + "step": 1446 + }, + { + "epoch": 0.18407327312046814, + "grad_norm": 2.9959444999694824, + "learning_rate": 6.129715981348028e-07, + "loss": 0.3865, + "mean_token_accuracy": 0.8711618185043335, + "num_tokens": 56273139.0, + "step": 1447 + }, + { + "epoch": 0.18420048339905865, + "grad_norm": 2.9284279346466064, + "learning_rate": 6.133955065705807e-07, + "loss": 0.4014, + "mean_token_accuracy": 0.8618186116218567, + "num_tokens": 56309199.0, + "step": 1448 + }, + { + "epoch": 0.18432769367764915, + "grad_norm": 2.911255121231079, + "learning_rate": 6.138194150063585e-07, + "loss": 0.3735, + "mean_token_accuracy": 0.8717868328094482, + "num_tokens": 56346771.0, + "step": 1449 + }, + { + "epoch": 0.18445490395623967, + "grad_norm": 3.0535988807678223, + "learning_rate": 6.142433234421365e-07, + "loss": 0.4171, + "mean_token_accuracy": 0.8600670099258423, + "num_tokens": 56383433.0, + "step": 1450 + }, + { + "epoch": 0.18458211423483017, + "grad_norm": 2.9573144912719727, + "learning_rate": 6.146672318779143e-07, + "loss": 0.3736, + "mean_token_accuracy": 0.8719826936721802, + "num_tokens": 56420097.0, + "step": 1451 + }, + { + "epoch": 0.18470932451342068, + "grad_norm": 2.7304415702819824, + "learning_rate": 6.150911403136923e-07, + "loss": 0.3651, + "mean_token_accuracy": 0.8745151162147522, + "num_tokens": 56461456.0, + "step": 1452 + }, + { + "epoch": 0.1848365347920112, + "grad_norm": 2.5349559783935547, + "learning_rate": 6.155150487494701e-07, + "loss": 0.3663, + "mean_token_accuracy": 0.8773422241210938, + "num_tokens": 56510598.0, + "step": 1453 + }, + { + "epoch": 0.1849637450706017, + "grad_norm": 2.9961600303649902, + "learning_rate": 6.159389571852479e-07, + "loss": 0.4105, + "mean_token_accuracy": 0.8584133386611938, + "num_tokens": 56550631.0, + "step": 1454 + }, + { + "epoch": 0.1850909553491922, + "grad_norm": 2.8180317878723145, + "learning_rate": 6.163628656210258e-07, + "loss": 0.3637, + "mean_token_accuracy": 0.8781571388244629, + "num_tokens": 56587253.0, + "step": 1455 + }, + { + "epoch": 0.18521816562778273, + "grad_norm": 2.99180269241333, + "learning_rate": 6.167867740568037e-07, + "loss": 0.4059, + "mean_token_accuracy": 0.8606858253479004, + "num_tokens": 56623670.0, + "step": 1456 + }, + { + "epoch": 0.18534537590637323, + "grad_norm": 2.720529079437256, + "learning_rate": 6.172106824925815e-07, + "loss": 0.3899, + "mean_token_accuracy": 0.8674617409706116, + "num_tokens": 56666671.0, + "step": 1457 + }, + { + "epoch": 0.18547258618496373, + "grad_norm": 2.868020534515381, + "learning_rate": 6.176345909283595e-07, + "loss": 0.398, + "mean_token_accuracy": 0.8660261631011963, + "num_tokens": 56703663.0, + "step": 1458 + }, + { + "epoch": 0.18559979646355426, + "grad_norm": 2.8383731842041016, + "learning_rate": 6.180584993641373e-07, + "loss": 0.4203, + "mean_token_accuracy": 0.8586804270744324, + "num_tokens": 56746478.0, + "step": 1459 + }, + { + "epoch": 0.18572700674214476, + "grad_norm": 2.8572001457214355, + "learning_rate": 6.184824077999153e-07, + "loss": 0.4048, + "mean_token_accuracy": 0.8612281084060669, + "num_tokens": 56787458.0, + "step": 1460 + }, + { + "epoch": 0.18585421702073526, + "grad_norm": 3.249990701675415, + "learning_rate": 6.189063162356931e-07, + "loss": 0.4641, + "mean_token_accuracy": 0.8464479446411133, + "num_tokens": 56821833.0, + "step": 1461 + }, + { + "epoch": 0.1859814272993258, + "grad_norm": 2.750826358795166, + "learning_rate": 6.193302246714709e-07, + "loss": 0.367, + "mean_token_accuracy": 0.8731811046600342, + "num_tokens": 56860212.0, + "step": 1462 + }, + { + "epoch": 0.1861086375779163, + "grad_norm": 3.246164321899414, + "learning_rate": 6.197541331072488e-07, + "loss": 0.382, + "mean_token_accuracy": 0.8695294260978699, + "num_tokens": 56890232.0, + "step": 1463 + }, + { + "epoch": 0.1862358478565068, + "grad_norm": 2.9206552505493164, + "learning_rate": 6.201780415430267e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8615216016769409, + "num_tokens": 56929686.0, + "step": 1464 + }, + { + "epoch": 0.18636305813509732, + "grad_norm": 2.9190754890441895, + "learning_rate": 6.206019499788045e-07, + "loss": 0.4474, + "mean_token_accuracy": 0.8479346036911011, + "num_tokens": 56973192.0, + "step": 1465 + }, + { + "epoch": 0.18649026841368782, + "grad_norm": 2.6251492500305176, + "learning_rate": 6.210258584145825e-07, + "loss": 0.4157, + "mean_token_accuracy": 0.8563649654388428, + "num_tokens": 57017147.0, + "step": 1466 + }, + { + "epoch": 0.18661747869227835, + "grad_norm": 2.73860764503479, + "learning_rate": 6.214497668503603e-07, + "loss": 0.4253, + "mean_token_accuracy": 0.8555728197097778, + "num_tokens": 57060589.0, + "step": 1467 + }, + { + "epoch": 0.18674468897086885, + "grad_norm": 2.8726720809936523, + "learning_rate": 6.218736752861383e-07, + "loss": 0.3589, + "mean_token_accuracy": 0.8760077953338623, + "num_tokens": 57096367.0, + "step": 1468 + }, + { + "epoch": 0.18687189924945935, + "grad_norm": 2.8826446533203125, + "learning_rate": 6.22297583721916e-07, + "loss": 0.4144, + "mean_token_accuracy": 0.8615442514419556, + "num_tokens": 57133579.0, + "step": 1469 + }, + { + "epoch": 0.18699910952804988, + "grad_norm": 2.9549622535705566, + "learning_rate": 6.227214921576938e-07, + "loss": 0.3981, + "mean_token_accuracy": 0.8662415742874146, + "num_tokens": 57170799.0, + "step": 1470 + }, + { + "epoch": 0.18712631980664038, + "grad_norm": 2.792299270629883, + "learning_rate": 6.231454005934718e-07, + "loss": 0.3997, + "mean_token_accuracy": 0.8665176033973694, + "num_tokens": 57213180.0, + "step": 1471 + }, + { + "epoch": 0.18725353008523088, + "grad_norm": 2.873351812362671, + "learning_rate": 6.235693090292496e-07, + "loss": 0.3771, + "mean_token_accuracy": 0.8712728023529053, + "num_tokens": 57248444.0, + "step": 1472 + }, + { + "epoch": 0.1873807403638214, + "grad_norm": 2.6757893562316895, + "learning_rate": 6.239932174650275e-07, + "loss": 0.4028, + "mean_token_accuracy": 0.8617892265319824, + "num_tokens": 57291108.0, + "step": 1473 + }, + { + "epoch": 0.1875079506424119, + "grad_norm": 2.897028684616089, + "learning_rate": 6.244171259008054e-07, + "loss": 0.3862, + "mean_token_accuracy": 0.8693913221359253, + "num_tokens": 57332094.0, + "step": 1474 + }, + { + "epoch": 0.1876351609210024, + "grad_norm": 2.8751978874206543, + "learning_rate": 6.248410343365833e-07, + "loss": 0.4073, + "mean_token_accuracy": 0.862298846244812, + "num_tokens": 57375388.0, + "step": 1475 + }, + { + "epoch": 0.18776237119959294, + "grad_norm": 2.7746267318725586, + "learning_rate": 6.252649427723612e-07, + "loss": 0.4114, + "mean_token_accuracy": 0.8584215641021729, + "num_tokens": 57419447.0, + "step": 1476 + }, + { + "epoch": 0.18788958147818344, + "grad_norm": 3.0144259929656982, + "learning_rate": 6.25688851208139e-07, + "loss": 0.4302, + "mean_token_accuracy": 0.8540439605712891, + "num_tokens": 57457222.0, + "step": 1477 + }, + { + "epoch": 0.18801679175677394, + "grad_norm": 2.623243808746338, + "learning_rate": 6.261127596439168e-07, + "loss": 0.4059, + "mean_token_accuracy": 0.8610223531723022, + "num_tokens": 57501850.0, + "step": 1478 + }, + { + "epoch": 0.18814400203536447, + "grad_norm": 2.907775640487671, + "learning_rate": 6.265366680796948e-07, + "loss": 0.3678, + "mean_token_accuracy": 0.8753387928009033, + "num_tokens": 57536726.0, + "step": 1479 + }, + { + "epoch": 0.18827121231395497, + "grad_norm": 2.656738758087158, + "learning_rate": 6.269605765154726e-07, + "loss": 0.3572, + "mean_token_accuracy": 0.879916787147522, + "num_tokens": 57583813.0, + "step": 1480 + }, + { + "epoch": 0.18839842259254547, + "grad_norm": 2.844971179962158, + "learning_rate": 6.273844849512505e-07, + "loss": 0.3314, + "mean_token_accuracy": 0.8865468502044678, + "num_tokens": 57619696.0, + "step": 1481 + }, + { + "epoch": 0.188525632871136, + "grad_norm": 2.811171770095825, + "learning_rate": 6.278083933870284e-07, + "loss": 0.4034, + "mean_token_accuracy": 0.8620505332946777, + "num_tokens": 57656925.0, + "step": 1482 + }, + { + "epoch": 0.1886528431497265, + "grad_norm": 2.913149833679199, + "learning_rate": 6.282323018228063e-07, + "loss": 0.4341, + "mean_token_accuracy": 0.8550176024436951, + "num_tokens": 57693086.0, + "step": 1483 + }, + { + "epoch": 0.188780053428317, + "grad_norm": 2.996119260787964, + "learning_rate": 6.286562102585841e-07, + "loss": 0.4296, + "mean_token_accuracy": 0.8534726500511169, + "num_tokens": 57730767.0, + "step": 1484 + }, + { + "epoch": 0.18890726370690752, + "grad_norm": 2.754937171936035, + "learning_rate": 6.29080118694362e-07, + "loss": 0.3616, + "mean_token_accuracy": 0.8784911036491394, + "num_tokens": 57769202.0, + "step": 1485 + }, + { + "epoch": 0.18903447398549802, + "grad_norm": 2.901839017868042, + "learning_rate": 6.295040271301398e-07, + "loss": 0.3915, + "mean_token_accuracy": 0.8691878318786621, + "num_tokens": 57805894.0, + "step": 1486 + }, + { + "epoch": 0.18916168426408853, + "grad_norm": 2.977273941040039, + "learning_rate": 6.299279355659178e-07, + "loss": 0.3851, + "mean_token_accuracy": 0.8699048757553101, + "num_tokens": 57840059.0, + "step": 1487 + }, + { + "epoch": 0.18928889454267905, + "grad_norm": 3.081467628479004, + "learning_rate": 6.303518440016956e-07, + "loss": 0.4045, + "mean_token_accuracy": 0.8581491708755493, + "num_tokens": 57871776.0, + "step": 1488 + }, + { + "epoch": 0.18941610482126955, + "grad_norm": 2.706559419631958, + "learning_rate": 6.307757524374735e-07, + "loss": 0.3859, + "mean_token_accuracy": 0.8682788014411926, + "num_tokens": 57914647.0, + "step": 1489 + }, + { + "epoch": 0.18954331509986008, + "grad_norm": 2.9020726680755615, + "learning_rate": 6.311996608732514e-07, + "loss": 0.3589, + "mean_token_accuracy": 0.8778454065322876, + "num_tokens": 57947357.0, + "step": 1490 + }, + { + "epoch": 0.18967052537845058, + "grad_norm": 2.822150230407715, + "learning_rate": 6.316235693090292e-07, + "loss": 0.4293, + "mean_token_accuracy": 0.8551177978515625, + "num_tokens": 57990160.0, + "step": 1491 + }, + { + "epoch": 0.18979773565704108, + "grad_norm": 2.6965415477752686, + "learning_rate": 6.320474777448071e-07, + "loss": 0.4019, + "mean_token_accuracy": 0.8605217337608337, + "num_tokens": 58031778.0, + "step": 1492 + }, + { + "epoch": 0.1899249459356316, + "grad_norm": 2.7707676887512207, + "learning_rate": 6.324713861805849e-07, + "loss": 0.3952, + "mean_token_accuracy": 0.8639481663703918, + "num_tokens": 58072839.0, + "step": 1493 + }, + { + "epoch": 0.1900521562142221, + "grad_norm": 3.125307083129883, + "learning_rate": 6.328952946163628e-07, + "loss": 0.3698, + "mean_token_accuracy": 0.8716013431549072, + "num_tokens": 58105313.0, + "step": 1494 + }, + { + "epoch": 0.1901793664928126, + "grad_norm": 2.838467597961426, + "learning_rate": 6.333192030521407e-07, + "loss": 0.369, + "mean_token_accuracy": 0.8738813996315002, + "num_tokens": 58144999.0, + "step": 1495 + }, + { + "epoch": 0.19030657677140314, + "grad_norm": 2.864887237548828, + "learning_rate": 6.337431114879186e-07, + "loss": 0.3734, + "mean_token_accuracy": 0.8719543218612671, + "num_tokens": 58183971.0, + "step": 1496 + }, + { + "epoch": 0.19043378704999364, + "grad_norm": 2.711362600326538, + "learning_rate": 6.341670199236965e-07, + "loss": 0.3465, + "mean_token_accuracy": 0.8805351257324219, + "num_tokens": 58220305.0, + "step": 1497 + }, + { + "epoch": 0.19056099732858414, + "grad_norm": 2.907447576522827, + "learning_rate": 6.345909283594744e-07, + "loss": 0.4067, + "mean_token_accuracy": 0.8596311807632446, + "num_tokens": 58257703.0, + "step": 1498 + }, + { + "epoch": 0.19068820760717467, + "grad_norm": 2.8883652687072754, + "learning_rate": 6.350148367952522e-07, + "loss": 0.347, + "mean_token_accuracy": 0.8769954442977905, + "num_tokens": 58292504.0, + "step": 1499 + }, + { + "epoch": 0.19081541788576517, + "grad_norm": 2.5475523471832275, + "learning_rate": 6.354387452310301e-07, + "loss": 0.3873, + "mean_token_accuracy": 0.8712427616119385, + "num_tokens": 58335494.0, + "step": 1500 + }, + { + "epoch": 0.19094262816435567, + "grad_norm": 2.9018688201904297, + "learning_rate": 6.358626536668079e-07, + "loss": 0.3949, + "mean_token_accuracy": 0.8660162091255188, + "num_tokens": 58373358.0, + "step": 1501 + }, + { + "epoch": 0.1910698384429462, + "grad_norm": 2.968313694000244, + "learning_rate": 6.362865621025858e-07, + "loss": 0.3813, + "mean_token_accuracy": 0.8689387440681458, + "num_tokens": 58410890.0, + "step": 1502 + }, + { + "epoch": 0.1911970487215367, + "grad_norm": 3.0153651237487793, + "learning_rate": 6.367104705383637e-07, + "loss": 0.425, + "mean_token_accuracy": 0.8548657298088074, + "num_tokens": 58446018.0, + "step": 1503 + }, + { + "epoch": 0.1913242590001272, + "grad_norm": 2.6494762897491455, + "learning_rate": 6.371343789741416e-07, + "loss": 0.4024, + "mean_token_accuracy": 0.8653263449668884, + "num_tokens": 58488359.0, + "step": 1504 + }, + { + "epoch": 0.19145146927871773, + "grad_norm": 2.9547908306121826, + "learning_rate": 6.375582874099195e-07, + "loss": 0.4109, + "mean_token_accuracy": 0.8650082349777222, + "num_tokens": 58526423.0, + "step": 1505 + }, + { + "epoch": 0.19157867955730823, + "grad_norm": 2.786746025085449, + "learning_rate": 6.379821958456974e-07, + "loss": 0.3511, + "mean_token_accuracy": 0.8803816437721252, + "num_tokens": 58562742.0, + "step": 1506 + }, + { + "epoch": 0.19170588983589873, + "grad_norm": 2.597341537475586, + "learning_rate": 6.384061042814751e-07, + "loss": 0.435, + "mean_token_accuracy": 0.8531666994094849, + "num_tokens": 58608761.0, + "step": 1507 + }, + { + "epoch": 0.19183310011448926, + "grad_norm": 2.78898286819458, + "learning_rate": 6.38830012717253e-07, + "loss": 0.4519, + "mean_token_accuracy": 0.8505146503448486, + "num_tokens": 58650594.0, + "step": 1508 + }, + { + "epoch": 0.19196031039307976, + "grad_norm": 2.9359941482543945, + "learning_rate": 6.392539211530309e-07, + "loss": 0.4528, + "mean_token_accuracy": 0.8500818014144897, + "num_tokens": 58691062.0, + "step": 1509 + }, + { + "epoch": 0.19208752067167026, + "grad_norm": 2.8306989669799805, + "learning_rate": 6.396778295888087e-07, + "loss": 0.3967, + "mean_token_accuracy": 0.8638664484024048, + "num_tokens": 58728275.0, + "step": 1510 + }, + { + "epoch": 0.1922147309502608, + "grad_norm": 2.680668354034424, + "learning_rate": 6.401017380245867e-07, + "loss": 0.345, + "mean_token_accuracy": 0.8831197619438171, + "num_tokens": 58771329.0, + "step": 1511 + }, + { + "epoch": 0.1923419412288513, + "grad_norm": 2.577347993850708, + "learning_rate": 6.405256464603645e-07, + "loss": 0.3698, + "mean_token_accuracy": 0.8735523223876953, + "num_tokens": 58812615.0, + "step": 1512 + }, + { + "epoch": 0.1924691515074418, + "grad_norm": 2.7836310863494873, + "learning_rate": 6.409495548961425e-07, + "loss": 0.3918, + "mean_token_accuracy": 0.8656728863716125, + "num_tokens": 58849954.0, + "step": 1513 + }, + { + "epoch": 0.19259636178603232, + "grad_norm": 2.8673994541168213, + "learning_rate": 6.413734633319203e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8700594902038574, + "num_tokens": 58883381.0, + "step": 1514 + }, + { + "epoch": 0.19272357206462282, + "grad_norm": 2.9583969116210938, + "learning_rate": 6.417973717676981e-07, + "loss": 0.4029, + "mean_token_accuracy": 0.863552451133728, + "num_tokens": 58917576.0, + "step": 1515 + }, + { + "epoch": 0.19285078234321335, + "grad_norm": 2.765718698501587, + "learning_rate": 6.42221280203476e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8604615926742554, + "num_tokens": 58958134.0, + "step": 1516 + }, + { + "epoch": 0.19297799262180385, + "grad_norm": 2.5899338722229004, + "learning_rate": 6.426451886392539e-07, + "loss": 0.3882, + "mean_token_accuracy": 0.8674415349960327, + "num_tokens": 59002071.0, + "step": 1517 + }, + { + "epoch": 0.19310520290039435, + "grad_norm": 2.825432300567627, + "learning_rate": 6.430690970750317e-07, + "loss": 0.4167, + "mean_token_accuracy": 0.8591373562812805, + "num_tokens": 59041601.0, + "step": 1518 + }, + { + "epoch": 0.19323241317898487, + "grad_norm": 2.9768130779266357, + "learning_rate": 6.434930055108097e-07, + "loss": 0.4019, + "mean_token_accuracy": 0.8648159503936768, + "num_tokens": 59075366.0, + "step": 1519 + }, + { + "epoch": 0.19335962345757537, + "grad_norm": 2.6310040950775146, + "learning_rate": 6.439169139465875e-07, + "loss": 0.3966, + "mean_token_accuracy": 0.8652262687683105, + "num_tokens": 59118295.0, + "step": 1520 + }, + { + "epoch": 0.19348683373616588, + "grad_norm": 2.644819974899292, + "learning_rate": 6.443408223823655e-07, + "loss": 0.3291, + "mean_token_accuracy": 0.8878926038742065, + "num_tokens": 59156762.0, + "step": 1521 + }, + { + "epoch": 0.1936140440147564, + "grad_norm": 2.9416513442993164, + "learning_rate": 6.447647308181432e-07, + "loss": 0.4061, + "mean_token_accuracy": 0.8631383180618286, + "num_tokens": 59194319.0, + "step": 1522 + }, + { + "epoch": 0.1937412542933469, + "grad_norm": 3.068974256515503, + "learning_rate": 6.451886392539211e-07, + "loss": 0.4138, + "mean_token_accuracy": 0.8579684495925903, + "num_tokens": 59228930.0, + "step": 1523 + }, + { + "epoch": 0.1938684645719374, + "grad_norm": 2.829869270324707, + "learning_rate": 6.45612547689699e-07, + "loss": 0.3941, + "mean_token_accuracy": 0.8634862303733826, + "num_tokens": 59268320.0, + "step": 1524 + }, + { + "epoch": 0.19399567485052793, + "grad_norm": 3.078416109085083, + "learning_rate": 6.460364561254769e-07, + "loss": 0.4264, + "mean_token_accuracy": 0.8510182499885559, + "num_tokens": 59302298.0, + "step": 1525 + }, + { + "epoch": 0.19412288512911843, + "grad_norm": 3.012134075164795, + "learning_rate": 6.464603645612547e-07, + "loss": 0.3887, + "mean_token_accuracy": 0.8679016828536987, + "num_tokens": 59340422.0, + "step": 1526 + }, + { + "epoch": 0.19425009540770893, + "grad_norm": 3.1371030807495117, + "learning_rate": 6.468842729970327e-07, + "loss": 0.4014, + "mean_token_accuracy": 0.862538754940033, + "num_tokens": 59371777.0, + "step": 1527 + }, + { + "epoch": 0.19437730568629946, + "grad_norm": 2.793851852416992, + "learning_rate": 6.473081814328105e-07, + "loss": 0.4203, + "mean_token_accuracy": 0.8583493232727051, + "num_tokens": 59412236.0, + "step": 1528 + }, + { + "epoch": 0.19450451596488996, + "grad_norm": 2.6267244815826416, + "learning_rate": 6.477320898685885e-07, + "loss": 0.3634, + "mean_token_accuracy": 0.8754957914352417, + "num_tokens": 59451386.0, + "step": 1529 + }, + { + "epoch": 0.19463172624348046, + "grad_norm": 2.592703342437744, + "learning_rate": 6.481559983043662e-07, + "loss": 0.3806, + "mean_token_accuracy": 0.8701908588409424, + "num_tokens": 59492846.0, + "step": 1530 + }, + { + "epoch": 0.194758936522071, + "grad_norm": 2.8589842319488525, + "learning_rate": 6.48579906740144e-07, + "loss": 0.3919, + "mean_token_accuracy": 0.8673285245895386, + "num_tokens": 59530916.0, + "step": 1531 + }, + { + "epoch": 0.1948861468006615, + "grad_norm": 2.8488223552703857, + "learning_rate": 6.49003815175922e-07, + "loss": 0.426, + "mean_token_accuracy": 0.853743851184845, + "num_tokens": 59573198.0, + "step": 1532 + }, + { + "epoch": 0.195013357079252, + "grad_norm": 2.8137240409851074, + "learning_rate": 6.494277236116998e-07, + "loss": 0.3907, + "mean_token_accuracy": 0.8691429495811462, + "num_tokens": 59612874.0, + "step": 1533 + }, + { + "epoch": 0.19514056735784252, + "grad_norm": 2.7523701190948486, + "learning_rate": 6.498516320474777e-07, + "loss": 0.4133, + "mean_token_accuracy": 0.8585767149925232, + "num_tokens": 59652623.0, + "step": 1534 + }, + { + "epoch": 0.19526777763643302, + "grad_norm": 2.7125802040100098, + "learning_rate": 6.502755404832556e-07, + "loss": 0.3469, + "mean_token_accuracy": 0.8804317712783813, + "num_tokens": 59688880.0, + "step": 1535 + }, + { + "epoch": 0.19539498791502352, + "grad_norm": 2.850708246231079, + "learning_rate": 6.506994489190335e-07, + "loss": 0.3891, + "mean_token_accuracy": 0.8694292902946472, + "num_tokens": 59726252.0, + "step": 1536 + }, + { + "epoch": 0.19552219819361405, + "grad_norm": 3.118332624435425, + "learning_rate": 6.511233573548114e-07, + "loss": 0.3932, + "mean_token_accuracy": 0.8649881482124329, + "num_tokens": 59760270.0, + "step": 1537 + }, + { + "epoch": 0.19564940847220455, + "grad_norm": 2.7101755142211914, + "learning_rate": 6.515472657905892e-07, + "loss": 0.3921, + "mean_token_accuracy": 0.8656683564186096, + "num_tokens": 59803413.0, + "step": 1538 + }, + { + "epoch": 0.19577661875079505, + "grad_norm": 2.7382655143737793, + "learning_rate": 6.51971174226367e-07, + "loss": 0.3962, + "mean_token_accuracy": 0.8610202074050903, + "num_tokens": 59846862.0, + "step": 1539 + }, + { + "epoch": 0.19590382902938558, + "grad_norm": 2.79150652885437, + "learning_rate": 6.52395082662145e-07, + "loss": 0.3997, + "mean_token_accuracy": 0.8621411919593811, + "num_tokens": 59891204.0, + "step": 1540 + }, + { + "epoch": 0.19603103930797608, + "grad_norm": 2.813706398010254, + "learning_rate": 6.528189910979228e-07, + "loss": 0.3634, + "mean_token_accuracy": 0.8732243776321411, + "num_tokens": 59929825.0, + "step": 1541 + }, + { + "epoch": 0.1961582495865666, + "grad_norm": 2.9540789127349854, + "learning_rate": 6.532428995337007e-07, + "loss": 0.3677, + "mean_token_accuracy": 0.8688076734542847, + "num_tokens": 59963165.0, + "step": 1542 + }, + { + "epoch": 0.1962854598651571, + "grad_norm": 2.7503349781036377, + "learning_rate": 6.536668079694786e-07, + "loss": 0.4033, + "mean_token_accuracy": 0.8616656064987183, + "num_tokens": 60003677.0, + "step": 1543 + }, + { + "epoch": 0.1964126701437476, + "grad_norm": 2.768425703048706, + "learning_rate": 6.540907164052565e-07, + "loss": 0.4021, + "mean_token_accuracy": 0.8604458570480347, + "num_tokens": 60044669.0, + "step": 1544 + }, + { + "epoch": 0.19653988042233814, + "grad_norm": 2.903238296508789, + "learning_rate": 6.545146248410343e-07, + "loss": 0.3605, + "mean_token_accuracy": 0.8734539151191711, + "num_tokens": 60079908.0, + "step": 1545 + }, + { + "epoch": 0.19666709070092864, + "grad_norm": 3.2222652435302734, + "learning_rate": 6.549385332768122e-07, + "loss": 0.4171, + "mean_token_accuracy": 0.8614382743835449, + "num_tokens": 60112851.0, + "step": 1546 + }, + { + "epoch": 0.19679430097951914, + "grad_norm": 2.6898441314697266, + "learning_rate": 6.5536244171259e-07, + "loss": 0.3716, + "mean_token_accuracy": 0.8716934323310852, + "num_tokens": 60154532.0, + "step": 1547 + }, + { + "epoch": 0.19692151125810967, + "grad_norm": 2.791905164718628, + "learning_rate": 6.55786350148368e-07, + "loss": 0.3508, + "mean_token_accuracy": 0.8773916959762573, + "num_tokens": 60193145.0, + "step": 1548 + }, + { + "epoch": 0.19704872153670017, + "grad_norm": 2.5645575523376465, + "learning_rate": 6.562102585841458e-07, + "loss": 0.4152, + "mean_token_accuracy": 0.8602897524833679, + "num_tokens": 60239685.0, + "step": 1549 + }, + { + "epoch": 0.19717593181529067, + "grad_norm": 2.7115426063537598, + "learning_rate": 6.566341670199236e-07, + "loss": 0.357, + "mean_token_accuracy": 0.8770829439163208, + "num_tokens": 60280226.0, + "step": 1550 + }, + { + "epoch": 0.1973031420938812, + "grad_norm": 2.6843154430389404, + "learning_rate": 6.570580754557016e-07, + "loss": 0.3775, + "mean_token_accuracy": 0.8701561689376831, + "num_tokens": 60323565.0, + "step": 1551 + }, + { + "epoch": 0.1974303523724717, + "grad_norm": 2.9498775005340576, + "learning_rate": 6.574819838914794e-07, + "loss": 0.3631, + "mean_token_accuracy": 0.8769283294677734, + "num_tokens": 60357232.0, + "step": 1552 + }, + { + "epoch": 0.1975575626510622, + "grad_norm": 2.8056743144989014, + "learning_rate": 6.579058923272573e-07, + "loss": 0.4017, + "mean_token_accuracy": 0.8624070882797241, + "num_tokens": 60397116.0, + "step": 1553 + }, + { + "epoch": 0.19768477292965272, + "grad_norm": 3.073133945465088, + "learning_rate": 6.583298007630351e-07, + "loss": 0.4441, + "mean_token_accuracy": 0.8501772880554199, + "num_tokens": 60432941.0, + "step": 1554 + }, + { + "epoch": 0.19781198320824323, + "grad_norm": 2.924515724182129, + "learning_rate": 6.58753709198813e-07, + "loss": 0.3886, + "mean_token_accuracy": 0.8691152334213257, + "num_tokens": 60471337.0, + "step": 1555 + }, + { + "epoch": 0.19793919348683373, + "grad_norm": 2.8521370887756348, + "learning_rate": 6.591776176345909e-07, + "loss": 0.4789, + "mean_token_accuracy": 0.8411080837249756, + "num_tokens": 60514049.0, + "step": 1556 + }, + { + "epoch": 0.19806640376542425, + "grad_norm": 2.6430881023406982, + "learning_rate": 6.596015260703688e-07, + "loss": 0.3795, + "mean_token_accuracy": 0.8754500150680542, + "num_tokens": 60556965.0, + "step": 1557 + }, + { + "epoch": 0.19819361404401475, + "grad_norm": 3.1613879203796387, + "learning_rate": 6.600254345061466e-07, + "loss": 0.3933, + "mean_token_accuracy": 0.8661198616027832, + "num_tokens": 60589330.0, + "step": 1558 + }, + { + "epoch": 0.19832082432260525, + "grad_norm": 2.6378235816955566, + "learning_rate": 6.604493429419246e-07, + "loss": 0.3685, + "mean_token_accuracy": 0.8718171119689941, + "num_tokens": 60629768.0, + "step": 1559 + }, + { + "epoch": 0.19844803460119578, + "grad_norm": 3.0094902515411377, + "learning_rate": 6.608732513777023e-07, + "loss": 0.3586, + "mean_token_accuracy": 0.8763245344161987, + "num_tokens": 60661087.0, + "step": 1560 + }, + { + "epoch": 0.19857524487978628, + "grad_norm": 2.9283671379089355, + "learning_rate": 6.612971598134803e-07, + "loss": 0.4137, + "mean_token_accuracy": 0.8624452948570251, + "num_tokens": 60701238.0, + "step": 1561 + }, + { + "epoch": 0.19870245515837678, + "grad_norm": 2.966076612472534, + "learning_rate": 6.617210682492581e-07, + "loss": 0.4227, + "mean_token_accuracy": 0.8554172515869141, + "num_tokens": 60736479.0, + "step": 1562 + }, + { + "epoch": 0.1988296654369673, + "grad_norm": 2.7303478717803955, + "learning_rate": 6.62144976685036e-07, + "loss": 0.4219, + "mean_token_accuracy": 0.8591126203536987, + "num_tokens": 60779975.0, + "step": 1563 + }, + { + "epoch": 0.1989568757155578, + "grad_norm": 2.838460683822632, + "learning_rate": 6.625688851208139e-07, + "loss": 0.3753, + "mean_token_accuracy": 0.8727172017097473, + "num_tokens": 60815245.0, + "step": 1564 + }, + { + "epoch": 0.19908408599414834, + "grad_norm": 2.7766263484954834, + "learning_rate": 6.629927935565918e-07, + "loss": 0.4017, + "mean_token_accuracy": 0.8643547296524048, + "num_tokens": 60856219.0, + "step": 1565 + }, + { + "epoch": 0.19921129627273884, + "grad_norm": 2.931760311126709, + "learning_rate": 6.634167019923696e-07, + "loss": 0.3813, + "mean_token_accuracy": 0.8744102716445923, + "num_tokens": 60893185.0, + "step": 1566 + }, + { + "epoch": 0.19933850655132934, + "grad_norm": 2.707003116607666, + "learning_rate": 6.638406104281476e-07, + "loss": 0.3862, + "mean_token_accuracy": 0.8668105602264404, + "num_tokens": 60932517.0, + "step": 1567 + }, + { + "epoch": 0.19946571682991987, + "grad_norm": 2.9528884887695312, + "learning_rate": 6.642645188639253e-07, + "loss": 0.382, + "mean_token_accuracy": 0.866070032119751, + "num_tokens": 60966933.0, + "step": 1568 + }, + { + "epoch": 0.19959292710851037, + "grad_norm": 2.6691160202026367, + "learning_rate": 6.646884272997032e-07, + "loss": 0.4096, + "mean_token_accuracy": 0.8623931407928467, + "num_tokens": 61008060.0, + "step": 1569 + }, + { + "epoch": 0.19972013738710087, + "grad_norm": 2.705336332321167, + "learning_rate": 6.651123357354811e-07, + "loss": 0.3599, + "mean_token_accuracy": 0.8786802291870117, + "num_tokens": 61050141.0, + "step": 1570 + }, + { + "epoch": 0.1998473476656914, + "grad_norm": 2.813683271408081, + "learning_rate": 6.655362441712589e-07, + "loss": 0.3974, + "mean_token_accuracy": 0.8647062182426453, + "num_tokens": 61092724.0, + "step": 1571 + }, + { + "epoch": 0.1999745579442819, + "grad_norm": 2.490612745285034, + "learning_rate": 6.659601526070369e-07, + "loss": 0.3813, + "mean_token_accuracy": 0.8701642155647278, + "num_tokens": 61138567.0, + "step": 1572 + }, + { + "epoch": 0.2001017682228724, + "grad_norm": 2.895029306411743, + "learning_rate": 6.663840610428147e-07, + "loss": 0.4216, + "mean_token_accuracy": 0.8601850867271423, + "num_tokens": 61176762.0, + "step": 1573 + }, + { + "epoch": 0.20022897850146293, + "grad_norm": 2.830169439315796, + "learning_rate": 6.668079694785926e-07, + "loss": 0.4203, + "mean_token_accuracy": 0.8574901223182678, + "num_tokens": 61216116.0, + "step": 1574 + }, + { + "epoch": 0.20035618878005343, + "grad_norm": 2.7764036655426025, + "learning_rate": 6.672318779143704e-07, + "loss": 0.3612, + "mean_token_accuracy": 0.8748736381530762, + "num_tokens": 61254424.0, + "step": 1575 + }, + { + "epoch": 0.20048339905864393, + "grad_norm": 3.162808656692505, + "learning_rate": 6.676557863501483e-07, + "loss": 0.3985, + "mean_token_accuracy": 0.8646412491798401, + "num_tokens": 61287619.0, + "step": 1576 + }, + { + "epoch": 0.20061060933723446, + "grad_norm": 3.749584674835205, + "learning_rate": 6.680796947859262e-07, + "loss": 0.3698, + "mean_token_accuracy": 0.8754191994667053, + "num_tokens": 61322552.0, + "step": 1577 + }, + { + "epoch": 0.20073781961582496, + "grad_norm": 2.791919231414795, + "learning_rate": 6.685036032217041e-07, + "loss": 0.3984, + "mean_token_accuracy": 0.8579055666923523, + "num_tokens": 61358457.0, + "step": 1578 + }, + { + "epoch": 0.20086502989441546, + "grad_norm": 2.7549808025360107, + "learning_rate": 6.689275116574819e-07, + "loss": 0.3884, + "mean_token_accuracy": 0.8652044534683228, + "num_tokens": 61396720.0, + "step": 1579 + }, + { + "epoch": 0.200992240173006, + "grad_norm": 2.928952217102051, + "learning_rate": 6.693514200932599e-07, + "loss": 0.402, + "mean_token_accuracy": 0.8594474792480469, + "num_tokens": 61433919.0, + "step": 1580 + }, + { + "epoch": 0.2011194504515965, + "grad_norm": 2.6724114418029785, + "learning_rate": 6.697753285290377e-07, + "loss": 0.3817, + "mean_token_accuracy": 0.8710885047912598, + "num_tokens": 61472959.0, + "step": 1581 + }, + { + "epoch": 0.201246660730187, + "grad_norm": 2.7741005420684814, + "learning_rate": 6.701992369648156e-07, + "loss": 0.4057, + "mean_token_accuracy": 0.8606225252151489, + "num_tokens": 61513805.0, + "step": 1582 + }, + { + "epoch": 0.20137387100877752, + "grad_norm": 2.549431562423706, + "learning_rate": 6.706231454005934e-07, + "loss": 0.4017, + "mean_token_accuracy": 0.8661510348320007, + "num_tokens": 61560023.0, + "step": 1583 + }, + { + "epoch": 0.20150108128736802, + "grad_norm": 3.108609914779663, + "learning_rate": 6.710470538363713e-07, + "loss": 0.4159, + "mean_token_accuracy": 0.8638315200805664, + "num_tokens": 61595031.0, + "step": 1584 + }, + { + "epoch": 0.20162829156595852, + "grad_norm": 2.8130476474761963, + "learning_rate": 6.714709622721492e-07, + "loss": 0.3935, + "mean_token_accuracy": 0.8684980869293213, + "num_tokens": 61632390.0, + "step": 1585 + }, + { + "epoch": 0.20175550184454905, + "grad_norm": 3.118223190307617, + "learning_rate": 6.718948707079271e-07, + "loss": 0.4188, + "mean_token_accuracy": 0.8572244048118591, + "num_tokens": 61665670.0, + "step": 1586 + }, + { + "epoch": 0.20188271212313955, + "grad_norm": 3.147941827774048, + "learning_rate": 6.723187791437049e-07, + "loss": 0.454, + "mean_token_accuracy": 0.8464958071708679, + "num_tokens": 61699527.0, + "step": 1587 + }, + { + "epoch": 0.20200992240173005, + "grad_norm": 2.7589964866638184, + "learning_rate": 6.727426875794829e-07, + "loss": 0.3954, + "mean_token_accuracy": 0.866486132144928, + "num_tokens": 61742005.0, + "step": 1588 + }, + { + "epoch": 0.20213713268032057, + "grad_norm": 3.0198662281036377, + "learning_rate": 6.731665960152607e-07, + "loss": 0.3782, + "mean_token_accuracy": 0.8616875410079956, + "num_tokens": 61775200.0, + "step": 1589 + }, + { + "epoch": 0.20226434295891108, + "grad_norm": 2.7532615661621094, + "learning_rate": 6.735905044510385e-07, + "loss": 0.3975, + "mean_token_accuracy": 0.8653413653373718, + "num_tokens": 61816547.0, + "step": 1590 + }, + { + "epoch": 0.2023915532375016, + "grad_norm": 2.921649217605591, + "learning_rate": 6.740144128868164e-07, + "loss": 0.4383, + "mean_token_accuracy": 0.8512207269668579, + "num_tokens": 61855230.0, + "step": 1591 + }, + { + "epoch": 0.2025187635160921, + "grad_norm": 3.042079448699951, + "learning_rate": 6.744383213225942e-07, + "loss": 0.4015, + "mean_token_accuracy": 0.8628156185150146, + "num_tokens": 61895008.0, + "step": 1592 + }, + { + "epoch": 0.2026459737946826, + "grad_norm": 2.734459400177002, + "learning_rate": 6.748622297583722e-07, + "loss": 0.3515, + "mean_token_accuracy": 0.876914918422699, + "num_tokens": 61931515.0, + "step": 1593 + }, + { + "epoch": 0.20277318407327313, + "grad_norm": 2.7804884910583496, + "learning_rate": 6.7528613819415e-07, + "loss": 0.3659, + "mean_token_accuracy": 0.872584342956543, + "num_tokens": 61966642.0, + "step": 1594 + }, + { + "epoch": 0.20290039435186363, + "grad_norm": 2.8995749950408936, + "learning_rate": 6.757100466299279e-07, + "loss": 0.4247, + "mean_token_accuracy": 0.8590545654296875, + "num_tokens": 62003094.0, + "step": 1595 + }, + { + "epoch": 0.20302760463045413, + "grad_norm": 2.927093982696533, + "learning_rate": 6.761339550657058e-07, + "loss": 0.4193, + "mean_token_accuracy": 0.8558933734893799, + "num_tokens": 62039181.0, + "step": 1596 + }, + { + "epoch": 0.20315481490904466, + "grad_norm": 2.9332635402679443, + "learning_rate": 6.765578635014837e-07, + "loss": 0.3767, + "mean_token_accuracy": 0.8703289031982422, + "num_tokens": 62076111.0, + "step": 1597 + }, + { + "epoch": 0.20328202518763516, + "grad_norm": 2.766664981842041, + "learning_rate": 6.769817719372614e-07, + "loss": 0.3807, + "mean_token_accuracy": 0.8678364753723145, + "num_tokens": 62111913.0, + "step": 1598 + }, + { + "epoch": 0.20340923546622566, + "grad_norm": 2.5917985439300537, + "learning_rate": 6.774056803730394e-07, + "loss": 0.3673, + "mean_token_accuracy": 0.8737491369247437, + "num_tokens": 62151566.0, + "step": 1599 + }, + { + "epoch": 0.2035364457448162, + "grad_norm": 2.8614118099212646, + "learning_rate": 6.778295888088172e-07, + "loss": 0.3649, + "mean_token_accuracy": 0.8747466802597046, + "num_tokens": 62187480.0, + "step": 1600 + }, + { + "epoch": 0.2036636560234067, + "grad_norm": 2.901367425918579, + "learning_rate": 6.782534972445952e-07, + "loss": 0.431, + "mean_token_accuracy": 0.8541081547737122, + "num_tokens": 62225164.0, + "step": 1601 + }, + { + "epoch": 0.2037908663019972, + "grad_norm": 2.7232022285461426, + "learning_rate": 6.78677405680373e-07, + "loss": 0.3536, + "mean_token_accuracy": 0.8806852102279663, + "num_tokens": 62267222.0, + "step": 1602 + }, + { + "epoch": 0.20391807658058772, + "grad_norm": 2.9203579425811768, + "learning_rate": 6.791013141161509e-07, + "loss": 0.3838, + "mean_token_accuracy": 0.8694981932640076, + "num_tokens": 62305272.0, + "step": 1603 + }, + { + "epoch": 0.20404528685917822, + "grad_norm": 3.0464601516723633, + "learning_rate": 6.795252225519288e-07, + "loss": 0.4155, + "mean_token_accuracy": 0.8604795336723328, + "num_tokens": 62343282.0, + "step": 1604 + }, + { + "epoch": 0.20417249713776872, + "grad_norm": 2.9207680225372314, + "learning_rate": 6.799491309877067e-07, + "loss": 0.3667, + "mean_token_accuracy": 0.8738530278205872, + "num_tokens": 62379255.0, + "step": 1605 + }, + { + "epoch": 0.20429970741635925, + "grad_norm": 2.769946575164795, + "learning_rate": 6.803730394234844e-07, + "loss": 0.3413, + "mean_token_accuracy": 0.8840378522872925, + "num_tokens": 62417158.0, + "step": 1606 + }, + { + "epoch": 0.20442691769494975, + "grad_norm": 2.867483139038086, + "learning_rate": 6.807969478592624e-07, + "loss": 0.4248, + "mean_token_accuracy": 0.8580765724182129, + "num_tokens": 62458221.0, + "step": 1607 + }, + { + "epoch": 0.20455412797354025, + "grad_norm": 2.8011515140533447, + "learning_rate": 6.812208562950402e-07, + "loss": 0.3801, + "mean_token_accuracy": 0.8669263124465942, + "num_tokens": 62497654.0, + "step": 1608 + }, + { + "epoch": 0.20468133825213078, + "grad_norm": 2.8224680423736572, + "learning_rate": 6.816447647308182e-07, + "loss": 0.4112, + "mean_token_accuracy": 0.8601275086402893, + "num_tokens": 62536081.0, + "step": 1609 + }, + { + "epoch": 0.20480854853072128, + "grad_norm": 2.812797784805298, + "learning_rate": 6.82068673166596e-07, + "loss": 0.3387, + "mean_token_accuracy": 0.8848870396614075, + "num_tokens": 62570015.0, + "step": 1610 + }, + { + "epoch": 0.20493575880931178, + "grad_norm": 2.6459851264953613, + "learning_rate": 6.824925816023738e-07, + "loss": 0.3497, + "mean_token_accuracy": 0.8815080523490906, + "num_tokens": 62613195.0, + "step": 1611 + }, + { + "epoch": 0.2050629690879023, + "grad_norm": 2.712446689605713, + "learning_rate": 6.829164900381518e-07, + "loss": 0.3673, + "mean_token_accuracy": 0.8728386163711548, + "num_tokens": 62654077.0, + "step": 1612 + }, + { + "epoch": 0.2051901793664928, + "grad_norm": 2.6810545921325684, + "learning_rate": 6.833403984739295e-07, + "loss": 0.3757, + "mean_token_accuracy": 0.8723609447479248, + "num_tokens": 62696023.0, + "step": 1613 + }, + { + "epoch": 0.2053173896450833, + "grad_norm": 2.739680051803589, + "learning_rate": 6.837643069097074e-07, + "loss": 0.3743, + "mean_token_accuracy": 0.8727161288261414, + "num_tokens": 62732694.0, + "step": 1614 + }, + { + "epoch": 0.20544459992367384, + "grad_norm": 2.8838534355163574, + "learning_rate": 6.841882153454853e-07, + "loss": 0.3742, + "mean_token_accuracy": 0.8729307651519775, + "num_tokens": 62769414.0, + "step": 1615 + }, + { + "epoch": 0.20557181020226434, + "grad_norm": 2.876713991165161, + "learning_rate": 6.846121237812632e-07, + "loss": 0.3818, + "mean_token_accuracy": 0.8676989674568176, + "num_tokens": 62805872.0, + "step": 1616 + }, + { + "epoch": 0.20569902048085487, + "grad_norm": 3.7008769512176514, + "learning_rate": 6.850360322170411e-07, + "loss": 0.409, + "mean_token_accuracy": 0.8604952096939087, + "num_tokens": 62849539.0, + "step": 1617 + }, + { + "epoch": 0.20582623075944537, + "grad_norm": 2.7168397903442383, + "learning_rate": 6.85459940652819e-07, + "loss": 0.3359, + "mean_token_accuracy": 0.8822653889656067, + "num_tokens": 62886951.0, + "step": 1618 + }, + { + "epoch": 0.20595344103803587, + "grad_norm": 2.8299455642700195, + "learning_rate": 6.858838490885968e-07, + "loss": 0.4234, + "mean_token_accuracy": 0.8590827584266663, + "num_tokens": 62927938.0, + "step": 1619 + }, + { + "epoch": 0.2060806513166264, + "grad_norm": 2.8776347637176514, + "learning_rate": 6.863077575243748e-07, + "loss": 0.4384, + "mean_token_accuracy": 0.8510908484458923, + "num_tokens": 62969067.0, + "step": 1620 + }, + { + "epoch": 0.2062078615952169, + "grad_norm": 2.7095389366149902, + "learning_rate": 6.867316659601525e-07, + "loss": 0.4338, + "mean_token_accuracy": 0.8555942177772522, + "num_tokens": 63014350.0, + "step": 1621 + }, + { + "epoch": 0.2063350718738074, + "grad_norm": 2.8180418014526367, + "learning_rate": 6.871555743959304e-07, + "loss": 0.3881, + "mean_token_accuracy": 0.869007408618927, + "num_tokens": 63051567.0, + "step": 1622 + }, + { + "epoch": 0.20646228215239792, + "grad_norm": 2.8807969093322754, + "learning_rate": 6.875794828317083e-07, + "loss": 0.4513, + "mean_token_accuracy": 0.8470668196678162, + "num_tokens": 63092807.0, + "step": 1623 + }, + { + "epoch": 0.20658949243098843, + "grad_norm": 2.6674256324768066, + "learning_rate": 6.880033912674862e-07, + "loss": 0.3667, + "mean_token_accuracy": 0.8745942711830139, + "num_tokens": 63138002.0, + "step": 1624 + }, + { + "epoch": 0.20671670270957893, + "grad_norm": 3.0285768508911133, + "learning_rate": 6.884272997032641e-07, + "loss": 0.3987, + "mean_token_accuracy": 0.8637664318084717, + "num_tokens": 63173110.0, + "step": 1625 + }, + { + "epoch": 0.20684391298816945, + "grad_norm": 2.8550000190734863, + "learning_rate": 6.88851208139042e-07, + "loss": 0.4168, + "mean_token_accuracy": 0.859157919883728, + "num_tokens": 63213554.0, + "step": 1626 + }, + { + "epoch": 0.20697112326675995, + "grad_norm": 3.3173482418060303, + "learning_rate": 6.892751165748198e-07, + "loss": 0.4193, + "mean_token_accuracy": 0.8586397171020508, + "num_tokens": 63244476.0, + "step": 1627 + }, + { + "epoch": 0.20709833354535045, + "grad_norm": 2.989084482192993, + "learning_rate": 6.896990250105978e-07, + "loss": 0.3536, + "mean_token_accuracy": 0.8797235488891602, + "num_tokens": 63280206.0, + "step": 1628 + }, + { + "epoch": 0.20722554382394098, + "grad_norm": 2.613849639892578, + "learning_rate": 6.901229334463755e-07, + "loss": 0.3561, + "mean_token_accuracy": 0.8775327801704407, + "num_tokens": 63322023.0, + "step": 1629 + }, + { + "epoch": 0.20735275410253148, + "grad_norm": 2.990241050720215, + "learning_rate": 6.905468418821534e-07, + "loss": 0.3996, + "mean_token_accuracy": 0.8645368218421936, + "num_tokens": 63356621.0, + "step": 1630 + }, + { + "epoch": 0.20747996438112198, + "grad_norm": 2.83358097076416, + "learning_rate": 6.909707503179313e-07, + "loss": 0.3779, + "mean_token_accuracy": 0.8727742433547974, + "num_tokens": 63396848.0, + "step": 1631 + }, + { + "epoch": 0.2076071746597125, + "grad_norm": 2.9948208332061768, + "learning_rate": 6.913946587537091e-07, + "loss": 0.4219, + "mean_token_accuracy": 0.8540658354759216, + "num_tokens": 63434165.0, + "step": 1632 + }, + { + "epoch": 0.207734384938303, + "grad_norm": 2.621459484100342, + "learning_rate": 6.918185671894871e-07, + "loss": 0.3726, + "mean_token_accuracy": 0.8725093603134155, + "num_tokens": 63478121.0, + "step": 1633 + }, + { + "epoch": 0.2078615952168935, + "grad_norm": 2.8962836265563965, + "learning_rate": 6.922424756252649e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8663638830184937, + "num_tokens": 63516934.0, + "step": 1634 + }, + { + "epoch": 0.20798880549548404, + "grad_norm": 3.142472982406616, + "learning_rate": 6.926663840610428e-07, + "loss": 0.414, + "mean_token_accuracy": 0.858540952205658, + "num_tokens": 63549773.0, + "step": 1635 + }, + { + "epoch": 0.20811601577407454, + "grad_norm": 2.8374183177948, + "learning_rate": 6.930902924968206e-07, + "loss": 0.4446, + "mean_token_accuracy": 0.8507075905799866, + "num_tokens": 63592750.0, + "step": 1636 + }, + { + "epoch": 0.20824322605266504, + "grad_norm": 3.107678174972534, + "learning_rate": 6.935142009325985e-07, + "loss": 0.3685, + "mean_token_accuracy": 0.8716613054275513, + "num_tokens": 63624421.0, + "step": 1637 + }, + { + "epoch": 0.20837043633125557, + "grad_norm": 2.6772282123565674, + "learning_rate": 6.939381093683764e-07, + "loss": 0.4028, + "mean_token_accuracy": 0.8637841939926147, + "num_tokens": 63669041.0, + "step": 1638 + }, + { + "epoch": 0.20849764660984607, + "grad_norm": 2.856895923614502, + "learning_rate": 6.943620178041543e-07, + "loss": 0.3781, + "mean_token_accuracy": 0.8649431467056274, + "num_tokens": 63709125.0, + "step": 1639 + }, + { + "epoch": 0.2086248568884366, + "grad_norm": 2.757349729537964, + "learning_rate": 6.947859262399321e-07, + "loss": 0.3942, + "mean_token_accuracy": 0.8644680380821228, + "num_tokens": 63747247.0, + "step": 1640 + }, + { + "epoch": 0.2087520671670271, + "grad_norm": 2.8493833541870117, + "learning_rate": 6.952098346757101e-07, + "loss": 0.3805, + "mean_token_accuracy": 0.8710331916809082, + "num_tokens": 63785583.0, + "step": 1641 + }, + { + "epoch": 0.2088792774456176, + "grad_norm": 2.9730477333068848, + "learning_rate": 6.956337431114879e-07, + "loss": 0.4518, + "mean_token_accuracy": 0.8470944166183472, + "num_tokens": 63828532.0, + "step": 1642 + }, + { + "epoch": 0.20900648772420813, + "grad_norm": 2.9596283435821533, + "learning_rate": 6.960576515472658e-07, + "loss": 0.4157, + "mean_token_accuracy": 0.8570345640182495, + "num_tokens": 63867111.0, + "step": 1643 + }, + { + "epoch": 0.20913369800279863, + "grad_norm": 2.7482964992523193, + "learning_rate": 6.964815599830436e-07, + "loss": 0.3995, + "mean_token_accuracy": 0.8628671169281006, + "num_tokens": 63908292.0, + "step": 1644 + }, + { + "epoch": 0.20926090828138913, + "grad_norm": 3.1346242427825928, + "learning_rate": 6.969054684188215e-07, + "loss": 0.3879, + "mean_token_accuracy": 0.8657839298248291, + "num_tokens": 63943825.0, + "step": 1645 + }, + { + "epoch": 0.20938811855997966, + "grad_norm": 2.9588799476623535, + "learning_rate": 6.973293768545994e-07, + "loss": 0.3775, + "mean_token_accuracy": 0.8707753419876099, + "num_tokens": 63979204.0, + "step": 1646 + }, + { + "epoch": 0.20951532883857016, + "grad_norm": 2.6786911487579346, + "learning_rate": 6.977532852903773e-07, + "loss": 0.4029, + "mean_token_accuracy": 0.8640562891960144, + "num_tokens": 64023268.0, + "step": 1647 + }, + { + "epoch": 0.20964253911716066, + "grad_norm": 2.4385008811950684, + "learning_rate": 6.981771937261551e-07, + "loss": 0.3501, + "mean_token_accuracy": 0.8808528780937195, + "num_tokens": 64068506.0, + "step": 1648 + }, + { + "epoch": 0.2097697493957512, + "grad_norm": 2.6260123252868652, + "learning_rate": 6.986011021619331e-07, + "loss": 0.3872, + "mean_token_accuracy": 0.867950439453125, + "num_tokens": 64110171.0, + "step": 1649 + }, + { + "epoch": 0.2098969596743417, + "grad_norm": 3.0260887145996094, + "learning_rate": 6.990250105977109e-07, + "loss": 0.4018, + "mean_token_accuracy": 0.8640491962432861, + "num_tokens": 64147186.0, + "step": 1650 + }, + { + "epoch": 0.2100241699529322, + "grad_norm": 2.7470757961273193, + "learning_rate": 6.994489190334886e-07, + "loss": 0.3961, + "mean_token_accuracy": 0.8636707067489624, + "num_tokens": 64186187.0, + "step": 1651 + }, + { + "epoch": 0.21015138023152272, + "grad_norm": 3.0445125102996826, + "learning_rate": 6.998728274692666e-07, + "loss": 0.4252, + "mean_token_accuracy": 0.8546417951583862, + "num_tokens": 64220187.0, + "step": 1652 + }, + { + "epoch": 0.21027859051011322, + "grad_norm": 2.7351136207580566, + "learning_rate": 7.002967359050444e-07, + "loss": 0.4155, + "mean_token_accuracy": 0.8559510111808777, + "num_tokens": 64260818.0, + "step": 1653 + }, + { + "epoch": 0.21040580078870372, + "grad_norm": 2.873760223388672, + "learning_rate": 7.007206443408224e-07, + "loss": 0.4217, + "mean_token_accuracy": 0.858188271522522, + "num_tokens": 64300587.0, + "step": 1654 + }, + { + "epoch": 0.21053301106729425, + "grad_norm": 2.8469817638397217, + "learning_rate": 7.011445527766002e-07, + "loss": 0.3945, + "mean_token_accuracy": 0.8645734786987305, + "num_tokens": 64339780.0, + "step": 1655 + }, + { + "epoch": 0.21066022134588475, + "grad_norm": 2.9653403759002686, + "learning_rate": 7.015684612123781e-07, + "loss": 0.4457, + "mean_token_accuracy": 0.8480207920074463, + "num_tokens": 64380516.0, + "step": 1656 + }, + { + "epoch": 0.21078743162447525, + "grad_norm": 2.9307146072387695, + "learning_rate": 7.01992369648156e-07, + "loss": 0.3702, + "mean_token_accuracy": 0.8726892471313477, + "num_tokens": 64415625.0, + "step": 1657 + }, + { + "epoch": 0.21091464190306577, + "grad_norm": 3.1668214797973633, + "learning_rate": 7.024162780839339e-07, + "loss": 0.4401, + "mean_token_accuracy": 0.8550605773925781, + "num_tokens": 64449832.0, + "step": 1658 + }, + { + "epoch": 0.21104185218165628, + "grad_norm": 2.8955092430114746, + "learning_rate": 7.028401865197116e-07, + "loss": 0.4087, + "mean_token_accuracy": 0.8606757521629333, + "num_tokens": 64487286.0, + "step": 1659 + }, + { + "epoch": 0.21116906246024678, + "grad_norm": 2.999933958053589, + "learning_rate": 7.032640949554896e-07, + "loss": 0.3748, + "mean_token_accuracy": 0.8693369626998901, + "num_tokens": 64523056.0, + "step": 1660 + }, + { + "epoch": 0.2112962727388373, + "grad_norm": 2.936201333999634, + "learning_rate": 7.036880033912674e-07, + "loss": 0.3913, + "mean_token_accuracy": 0.8660566210746765, + "num_tokens": 64559075.0, + "step": 1661 + }, + { + "epoch": 0.2114234830174278, + "grad_norm": 2.8841116428375244, + "learning_rate": 7.041119118270454e-07, + "loss": 0.3844, + "mean_token_accuracy": 0.8686680197715759, + "num_tokens": 64596724.0, + "step": 1662 + }, + { + "epoch": 0.2115506932960183, + "grad_norm": 2.4601428508758545, + "learning_rate": 7.045358202628232e-07, + "loss": 0.3185, + "mean_token_accuracy": 0.8892714381217957, + "num_tokens": 64637277.0, + "step": 1663 + }, + { + "epoch": 0.21167790357460883, + "grad_norm": 3.03078293800354, + "learning_rate": 7.049597286986011e-07, + "loss": 0.4164, + "mean_token_accuracy": 0.8601011633872986, + "num_tokens": 64673832.0, + "step": 1664 + }, + { + "epoch": 0.21180511385319933, + "grad_norm": 2.6718828678131104, + "learning_rate": 7.05383637134379e-07, + "loss": 0.3268, + "mean_token_accuracy": 0.8847370147705078, + "num_tokens": 64710204.0, + "step": 1665 + }, + { + "epoch": 0.21193232413178986, + "grad_norm": 2.9436426162719727, + "learning_rate": 7.058075455701568e-07, + "loss": 0.3784, + "mean_token_accuracy": 0.8712456226348877, + "num_tokens": 64744493.0, + "step": 1666 + }, + { + "epoch": 0.21205953441038036, + "grad_norm": 3.182345151901245, + "learning_rate": 7.062314540059346e-07, + "loss": 0.4121, + "mean_token_accuracy": 0.8597014546394348, + "num_tokens": 64774269.0, + "step": 1667 + }, + { + "epoch": 0.21218674468897086, + "grad_norm": 2.5014798641204834, + "learning_rate": 7.066553624417126e-07, + "loss": 0.3747, + "mean_token_accuracy": 0.8706673383712769, + "num_tokens": 64824107.0, + "step": 1668 + }, + { + "epoch": 0.2123139549675614, + "grad_norm": 2.868630886077881, + "learning_rate": 7.070792708774904e-07, + "loss": 0.4295, + "mean_token_accuracy": 0.8579902648925781, + "num_tokens": 64863908.0, + "step": 1669 + }, + { + "epoch": 0.2124411652461519, + "grad_norm": 2.8382978439331055, + "learning_rate": 7.075031793132684e-07, + "loss": 0.3752, + "mean_token_accuracy": 0.8723562955856323, + "num_tokens": 64900783.0, + "step": 1670 + }, + { + "epoch": 0.2125683755247424, + "grad_norm": 2.803062915802002, + "learning_rate": 7.079270877490462e-07, + "loss": 0.3871, + "mean_token_accuracy": 0.8663909435272217, + "num_tokens": 64937812.0, + "step": 1671 + }, + { + "epoch": 0.21269558580333292, + "grad_norm": 2.9378817081451416, + "learning_rate": 7.08350996184824e-07, + "loss": 0.4147, + "mean_token_accuracy": 0.860302209854126, + "num_tokens": 64974118.0, + "step": 1672 + }, + { + "epoch": 0.21282279608192342, + "grad_norm": 2.898852825164795, + "learning_rate": 7.08774904620602e-07, + "loss": 0.3717, + "mean_token_accuracy": 0.8720723390579224, + "num_tokens": 65013664.0, + "step": 1673 + }, + { + "epoch": 0.21295000636051392, + "grad_norm": 2.8756446838378906, + "learning_rate": 7.091988130563797e-07, + "loss": 0.449, + "mean_token_accuracy": 0.8491711616516113, + "num_tokens": 65057372.0, + "step": 1674 + }, + { + "epoch": 0.21307721663910445, + "grad_norm": 2.665850877761841, + "learning_rate": 7.096227214921576e-07, + "loss": 0.396, + "mean_token_accuracy": 0.8663235902786255, + "num_tokens": 65097793.0, + "step": 1675 + }, + { + "epoch": 0.21320442691769495, + "grad_norm": 2.9902422428131104, + "learning_rate": 7.100466299279355e-07, + "loss": 0.4128, + "mean_token_accuracy": 0.858580470085144, + "num_tokens": 65132387.0, + "step": 1676 + }, + { + "epoch": 0.21333163719628545, + "grad_norm": 2.725498914718628, + "learning_rate": 7.104705383637134e-07, + "loss": 0.3317, + "mean_token_accuracy": 0.8851783275604248, + "num_tokens": 65172158.0, + "step": 1677 + }, + { + "epoch": 0.21345884747487598, + "grad_norm": 2.6575233936309814, + "learning_rate": 7.108944467994913e-07, + "loss": 0.3944, + "mean_token_accuracy": 0.8655024766921997, + "num_tokens": 65214816.0, + "step": 1678 + }, + { + "epoch": 0.21358605775346648, + "grad_norm": 2.841224431991577, + "learning_rate": 7.113183552352692e-07, + "loss": 0.4416, + "mean_token_accuracy": 0.8506426215171814, + "num_tokens": 65255049.0, + "step": 1679 + }, + { + "epoch": 0.21371326803205698, + "grad_norm": 3.250793695449829, + "learning_rate": 7.11742263671047e-07, + "loss": 0.3967, + "mean_token_accuracy": 0.8643525838851929, + "num_tokens": 65287104.0, + "step": 1680 + }, + { + "epoch": 0.2138404783106475, + "grad_norm": 2.5705485343933105, + "learning_rate": 7.12166172106825e-07, + "loss": 0.3706, + "mean_token_accuracy": 0.873249351978302, + "num_tokens": 65331440.0, + "step": 1681 + }, + { + "epoch": 0.213967688589238, + "grad_norm": 3.014464855194092, + "learning_rate": 7.125900805426027e-07, + "loss": 0.4366, + "mean_token_accuracy": 0.8542953729629517, + "num_tokens": 65372196.0, + "step": 1682 + }, + { + "epoch": 0.2140948988678285, + "grad_norm": 2.80458402633667, + "learning_rate": 7.130139889783806e-07, + "loss": 0.4234, + "mean_token_accuracy": 0.8564090728759766, + "num_tokens": 65413211.0, + "step": 1683 + }, + { + "epoch": 0.21422210914641904, + "grad_norm": 2.861233949661255, + "learning_rate": 7.134378974141585e-07, + "loss": 0.3591, + "mean_token_accuracy": 0.8753198385238647, + "num_tokens": 65452907.0, + "step": 1684 + }, + { + "epoch": 0.21434931942500954, + "grad_norm": 2.6926369667053223, + "learning_rate": 7.138618058499364e-07, + "loss": 0.3578, + "mean_token_accuracy": 0.8768976926803589, + "num_tokens": 65496583.0, + "step": 1685 + }, + { + "epoch": 0.21447652970360004, + "grad_norm": 2.5854742527008057, + "learning_rate": 7.142857142857143e-07, + "loss": 0.3727, + "mean_token_accuracy": 0.8725009560585022, + "num_tokens": 65539097.0, + "step": 1686 + }, + { + "epoch": 0.21460373998219057, + "grad_norm": 3.0181214809417725, + "learning_rate": 7.147096227214922e-07, + "loss": 0.4125, + "mean_token_accuracy": 0.8628984689712524, + "num_tokens": 65574011.0, + "step": 1687 + }, + { + "epoch": 0.21473095026078107, + "grad_norm": 2.7554049491882324, + "learning_rate": 7.1513353115727e-07, + "loss": 0.3698, + "mean_token_accuracy": 0.873211145401001, + "num_tokens": 65612198.0, + "step": 1688 + }, + { + "epoch": 0.21485816053937157, + "grad_norm": 2.703169107437134, + "learning_rate": 7.155574395930479e-07, + "loss": 0.4115, + "mean_token_accuracy": 0.8606124520301819, + "num_tokens": 65657097.0, + "step": 1689 + }, + { + "epoch": 0.2149853708179621, + "grad_norm": 2.7957210540771484, + "learning_rate": 7.159813480288257e-07, + "loss": 0.4199, + "mean_token_accuracy": 0.8603383302688599, + "num_tokens": 65699315.0, + "step": 1690 + }, + { + "epoch": 0.2151125810965526, + "grad_norm": 2.7705183029174805, + "learning_rate": 7.164052564646035e-07, + "loss": 0.4096, + "mean_token_accuracy": 0.8618450164794922, + "num_tokens": 65741079.0, + "step": 1691 + }, + { + "epoch": 0.21523979137514312, + "grad_norm": 2.60038685798645, + "learning_rate": 7.168291649003815e-07, + "loss": 0.4068, + "mean_token_accuracy": 0.8613095283508301, + "num_tokens": 65787776.0, + "step": 1692 + }, + { + "epoch": 0.21536700165373363, + "grad_norm": 2.7243237495422363, + "learning_rate": 7.172530733361593e-07, + "loss": 0.3872, + "mean_token_accuracy": 0.8680586218833923, + "num_tokens": 65830227.0, + "step": 1693 + }, + { + "epoch": 0.21549421193232413, + "grad_norm": 3.0145578384399414, + "learning_rate": 7.176769817719373e-07, + "loss": 0.3883, + "mean_token_accuracy": 0.867023229598999, + "num_tokens": 65867500.0, + "step": 1694 + }, + { + "epoch": 0.21562142221091465, + "grad_norm": 2.8149654865264893, + "learning_rate": 7.181008902077151e-07, + "loss": 0.3859, + "mean_token_accuracy": 0.8687260150909424, + "num_tokens": 65907133.0, + "step": 1695 + }, + { + "epoch": 0.21574863248950515, + "grad_norm": 2.873272180557251, + "learning_rate": 7.18524798643493e-07, + "loss": 0.4164, + "mean_token_accuracy": 0.8591423034667969, + "num_tokens": 65949376.0, + "step": 1696 + }, + { + "epoch": 0.21587584276809565, + "grad_norm": 2.7695398330688477, + "learning_rate": 7.189487070792708e-07, + "loss": 0.4217, + "mean_token_accuracy": 0.8606995344161987, + "num_tokens": 65992805.0, + "step": 1697 + }, + { + "epoch": 0.21600305304668618, + "grad_norm": 2.783555746078491, + "learning_rate": 7.193726155150487e-07, + "loss": 0.3819, + "mean_token_accuracy": 0.8692827820777893, + "num_tokens": 66032345.0, + "step": 1698 + }, + { + "epoch": 0.21613026332527668, + "grad_norm": 2.7752156257629395, + "learning_rate": 7.197965239508265e-07, + "loss": 0.4343, + "mean_token_accuracy": 0.8563629984855652, + "num_tokens": 66078222.0, + "step": 1699 + }, + { + "epoch": 0.21625747360386718, + "grad_norm": 2.8213865756988525, + "learning_rate": 7.202204323866045e-07, + "loss": 0.3824, + "mean_token_accuracy": 0.871171236038208, + "num_tokens": 66118739.0, + "step": 1700 + }, + { + "epoch": 0.2163846838824577, + "grad_norm": 2.821634531021118, + "learning_rate": 7.206443408223823e-07, + "loss": 0.4214, + "mean_token_accuracy": 0.8561519384384155, + "num_tokens": 66158939.0, + "step": 1701 + }, + { + "epoch": 0.2165118941610482, + "grad_norm": 2.9069366455078125, + "learning_rate": 7.210682492581603e-07, + "loss": 0.4065, + "mean_token_accuracy": 0.8600172996520996, + "num_tokens": 66195906.0, + "step": 1702 + }, + { + "epoch": 0.2166391044396387, + "grad_norm": 3.0037853717803955, + "learning_rate": 7.214921576939381e-07, + "loss": 0.4421, + "mean_token_accuracy": 0.851334810256958, + "num_tokens": 66237453.0, + "step": 1703 + }, + { + "epoch": 0.21676631471822924, + "grad_norm": 3.164578676223755, + "learning_rate": 7.219160661297159e-07, + "loss": 0.4421, + "mean_token_accuracy": 0.8514508008956909, + "num_tokens": 66272631.0, + "step": 1704 + }, + { + "epoch": 0.21689352499681974, + "grad_norm": 2.6105902194976807, + "learning_rate": 7.223399745654938e-07, + "loss": 0.4081, + "mean_token_accuracy": 0.8603034615516663, + "num_tokens": 66317287.0, + "step": 1705 + }, + { + "epoch": 0.21702073527541024, + "grad_norm": 3.509967088699341, + "learning_rate": 7.227638830012717e-07, + "loss": 0.4908, + "mean_token_accuracy": 0.8406081795692444, + "num_tokens": 66348323.0, + "step": 1706 + }, + { + "epoch": 0.21714794555400077, + "grad_norm": 3.0637006759643555, + "learning_rate": 7.231877914370495e-07, + "loss": 0.3976, + "mean_token_accuracy": 0.8623054623603821, + "num_tokens": 66380378.0, + "step": 1707 + }, + { + "epoch": 0.21727515583259127, + "grad_norm": 2.77451753616333, + "learning_rate": 7.236116998728275e-07, + "loss": 0.3624, + "mean_token_accuracy": 0.875178337097168, + "num_tokens": 66415985.0, + "step": 1708 + }, + { + "epoch": 0.21740236611118177, + "grad_norm": 2.838965654373169, + "learning_rate": 7.240356083086053e-07, + "loss": 0.4231, + "mean_token_accuracy": 0.8548758029937744, + "num_tokens": 66456284.0, + "step": 1709 + }, + { + "epoch": 0.2175295763897723, + "grad_norm": 2.822211980819702, + "learning_rate": 7.244595167443833e-07, + "loss": 0.4138, + "mean_token_accuracy": 0.8605492115020752, + "num_tokens": 66506162.0, + "step": 1710 + }, + { + "epoch": 0.2176567866683628, + "grad_norm": 3.1161770820617676, + "learning_rate": 7.248834251801611e-07, + "loss": 0.4749, + "mean_token_accuracy": 0.8444551229476929, + "num_tokens": 66542229.0, + "step": 1711 + }, + { + "epoch": 0.2177839969469533, + "grad_norm": 2.865063190460205, + "learning_rate": 7.253073336159388e-07, + "loss": 0.377, + "mean_token_accuracy": 0.8706309795379639, + "num_tokens": 66577203.0, + "step": 1712 + }, + { + "epoch": 0.21791120722554383, + "grad_norm": 2.480746030807495, + "learning_rate": 7.257312420517168e-07, + "loss": 0.385, + "mean_token_accuracy": 0.8694037199020386, + "num_tokens": 66623279.0, + "step": 1713 + }, + { + "epoch": 0.21803841750413433, + "grad_norm": 2.943871259689331, + "learning_rate": 7.261551504874946e-07, + "loss": 0.4123, + "mean_token_accuracy": 0.8574927449226379, + "num_tokens": 66662264.0, + "step": 1714 + }, + { + "epoch": 0.21816562778272486, + "grad_norm": 2.8377389907836914, + "learning_rate": 7.265790589232725e-07, + "loss": 0.337, + "mean_token_accuracy": 0.8813046216964722, + "num_tokens": 66697183.0, + "step": 1715 + }, + { + "epoch": 0.21829283806131536, + "grad_norm": 3.4315950870513916, + "learning_rate": 7.270029673590504e-07, + "loss": 0.4211, + "mean_token_accuracy": 0.85758376121521, + "num_tokens": 66725077.0, + "step": 1716 + }, + { + "epoch": 0.21842004833990586, + "grad_norm": 2.8407633304595947, + "learning_rate": 7.274268757948283e-07, + "loss": 0.3949, + "mean_token_accuracy": 0.8655071258544922, + "num_tokens": 66765539.0, + "step": 1717 + }, + { + "epoch": 0.2185472586184964, + "grad_norm": 2.6274447441101074, + "learning_rate": 7.278507842306062e-07, + "loss": 0.3847, + "mean_token_accuracy": 0.8654242753982544, + "num_tokens": 66810050.0, + "step": 1718 + }, + { + "epoch": 0.2186744688970869, + "grad_norm": 2.8045144081115723, + "learning_rate": 7.282746926663841e-07, + "loss": 0.3699, + "mean_token_accuracy": 0.8724331855773926, + "num_tokens": 66849632.0, + "step": 1719 + }, + { + "epoch": 0.2188016791756774, + "grad_norm": 2.745021104812622, + "learning_rate": 7.286986011021618e-07, + "loss": 0.3816, + "mean_token_accuracy": 0.8696978092193604, + "num_tokens": 66890700.0, + "step": 1720 + }, + { + "epoch": 0.21892888945426792, + "grad_norm": 2.9289424419403076, + "learning_rate": 7.291225095379398e-07, + "loss": 0.3934, + "mean_token_accuracy": 0.862880527973175, + "num_tokens": 66929508.0, + "step": 1721 + }, + { + "epoch": 0.21905609973285842, + "grad_norm": 2.855755090713501, + "learning_rate": 7.295464179737176e-07, + "loss": 0.3927, + "mean_token_accuracy": 0.8634256720542908, + "num_tokens": 66966592.0, + "step": 1722 + }, + { + "epoch": 0.21918331001144892, + "grad_norm": 3.052853584289551, + "learning_rate": 7.299703264094955e-07, + "loss": 0.4114, + "mean_token_accuracy": 0.8588145971298218, + "num_tokens": 67003514.0, + "step": 1723 + }, + { + "epoch": 0.21931052029003945, + "grad_norm": 2.9004647731781006, + "learning_rate": 7.303942348452734e-07, + "loss": 0.3605, + "mean_token_accuracy": 0.8753383159637451, + "num_tokens": 67037837.0, + "step": 1724 + }, + { + "epoch": 0.21943773056862995, + "grad_norm": 2.7600204944610596, + "learning_rate": 7.308181432810513e-07, + "loss": 0.3822, + "mean_token_accuracy": 0.8682026863098145, + "num_tokens": 67078636.0, + "step": 1725 + }, + { + "epoch": 0.21956494084722045, + "grad_norm": 2.609543561935425, + "learning_rate": 7.312420517168292e-07, + "loss": 0.377, + "mean_token_accuracy": 0.8697746992111206, + "num_tokens": 67123641.0, + "step": 1726 + }, + { + "epoch": 0.21969215112581097, + "grad_norm": 2.908892869949341, + "learning_rate": 7.31665960152607e-07, + "loss": 0.3768, + "mean_token_accuracy": 0.8733299970626831, + "num_tokens": 67158976.0, + "step": 1727 + }, + { + "epoch": 0.21981936140440148, + "grad_norm": 2.759530544281006, + "learning_rate": 7.320898685883848e-07, + "loss": 0.3918, + "mean_token_accuracy": 0.8647848963737488, + "num_tokens": 67197774.0, + "step": 1728 + }, + { + "epoch": 0.21994657168299198, + "grad_norm": 2.659126043319702, + "learning_rate": 7.325137770241628e-07, + "loss": 0.3849, + "mean_token_accuracy": 0.8682438135147095, + "num_tokens": 67240689.0, + "step": 1729 + }, + { + "epoch": 0.2200737819615825, + "grad_norm": 2.8045527935028076, + "learning_rate": 7.329376854599406e-07, + "loss": 0.4004, + "mean_token_accuracy": 0.8651727437973022, + "num_tokens": 67280043.0, + "step": 1730 + }, + { + "epoch": 0.220200992240173, + "grad_norm": 2.533953905105591, + "learning_rate": 7.333615938957184e-07, + "loss": 0.3998, + "mean_token_accuracy": 0.8640854358673096, + "num_tokens": 67326836.0, + "step": 1731 + }, + { + "epoch": 0.2203282025187635, + "grad_norm": 2.6212451457977295, + "learning_rate": 7.337855023314964e-07, + "loss": 0.3496, + "mean_token_accuracy": 0.8779970407485962, + "num_tokens": 67366428.0, + "step": 1732 + }, + { + "epoch": 0.22045541279735403, + "grad_norm": 2.893343687057495, + "learning_rate": 7.342094107672742e-07, + "loss": 0.3736, + "mean_token_accuracy": 0.8688091039657593, + "num_tokens": 67401543.0, + "step": 1733 + }, + { + "epoch": 0.22058262307594453, + "grad_norm": 2.774749279022217, + "learning_rate": 7.346333192030522e-07, + "loss": 0.3657, + "mean_token_accuracy": 0.874329686164856, + "num_tokens": 67439941.0, + "step": 1734 + }, + { + "epoch": 0.22070983335453503, + "grad_norm": 3.0209221839904785, + "learning_rate": 7.350572276388299e-07, + "loss": 0.4706, + "mean_token_accuracy": 0.8443974256515503, + "num_tokens": 67480421.0, + "step": 1735 + }, + { + "epoch": 0.22083704363312556, + "grad_norm": 2.929190158843994, + "learning_rate": 7.354811360746078e-07, + "loss": 0.4365, + "mean_token_accuracy": 0.8561975955963135, + "num_tokens": 67523240.0, + "step": 1736 + }, + { + "epoch": 0.22096425391171606, + "grad_norm": 2.8107430934906006, + "learning_rate": 7.359050445103857e-07, + "loss": 0.4162, + "mean_token_accuracy": 0.8581206798553467, + "num_tokens": 67563823.0, + "step": 1737 + }, + { + "epoch": 0.22109146419030656, + "grad_norm": 3.339571714401245, + "learning_rate": 7.363289529461636e-07, + "loss": 0.4232, + "mean_token_accuracy": 0.8589251637458801, + "num_tokens": 67592969.0, + "step": 1738 + }, + { + "epoch": 0.2212186744688971, + "grad_norm": 2.7098805904388428, + "learning_rate": 7.367528613819415e-07, + "loss": 0.3853, + "mean_token_accuracy": 0.8689723014831543, + "num_tokens": 67633938.0, + "step": 1739 + }, + { + "epoch": 0.2213458847474876, + "grad_norm": 2.7914879322052, + "learning_rate": 7.371767698177194e-07, + "loss": 0.4231, + "mean_token_accuracy": 0.8580451607704163, + "num_tokens": 67675376.0, + "step": 1740 + }, + { + "epoch": 0.22147309502607812, + "grad_norm": 2.5700292587280273, + "learning_rate": 7.376006782534972e-07, + "loss": 0.3398, + "mean_token_accuracy": 0.8833301067352295, + "num_tokens": 67720847.0, + "step": 1741 + }, + { + "epoch": 0.22160030530466862, + "grad_norm": 2.705442190170288, + "learning_rate": 7.380245866892751e-07, + "loss": 0.3807, + "mean_token_accuracy": 0.8697561025619507, + "num_tokens": 67760602.0, + "step": 1742 + }, + { + "epoch": 0.22172751558325912, + "grad_norm": 2.569376230239868, + "learning_rate": 7.384484951250529e-07, + "loss": 0.3721, + "mean_token_accuracy": 0.8738936185836792, + "num_tokens": 67804177.0, + "step": 1743 + }, + { + "epoch": 0.22185472586184965, + "grad_norm": 3.0387346744537354, + "learning_rate": 7.388724035608308e-07, + "loss": 0.3901, + "mean_token_accuracy": 0.8667597770690918, + "num_tokens": 67837747.0, + "step": 1744 + }, + { + "epoch": 0.22198193614044015, + "grad_norm": 2.9018428325653076, + "learning_rate": 7.392963119966087e-07, + "loss": 0.409, + "mean_token_accuracy": 0.8622400760650635, + "num_tokens": 67876621.0, + "step": 1745 + }, + { + "epoch": 0.22210914641903065, + "grad_norm": 3.0011520385742188, + "learning_rate": 7.397202204323866e-07, + "loss": 0.3421, + "mean_token_accuracy": 0.8819466829299927, + "num_tokens": 67911318.0, + "step": 1746 + }, + { + "epoch": 0.22223635669762118, + "grad_norm": 2.722123146057129, + "learning_rate": 7.401441288681645e-07, + "loss": 0.3476, + "mean_token_accuracy": 0.8777608275413513, + "num_tokens": 67949235.0, + "step": 1747 + }, + { + "epoch": 0.22236356697621168, + "grad_norm": 2.8356447219848633, + "learning_rate": 7.405680373039424e-07, + "loss": 0.3915, + "mean_token_accuracy": 0.8686478137969971, + "num_tokens": 67988607.0, + "step": 1748 + }, + { + "epoch": 0.22249077725480218, + "grad_norm": 2.683566093444824, + "learning_rate": 7.409919457397202e-07, + "loss": 0.3535, + "mean_token_accuracy": 0.8800961971282959, + "num_tokens": 68028474.0, + "step": 1749 + }, + { + "epoch": 0.2226179875333927, + "grad_norm": 3.1072564125061035, + "learning_rate": 7.414158541754981e-07, + "loss": 0.3868, + "mean_token_accuracy": 0.8680986166000366, + "num_tokens": 68063363.0, + "step": 1750 + }, + { + "epoch": 0.2227451978119832, + "grad_norm": 2.976207733154297, + "learning_rate": 7.418397626112759e-07, + "loss": 0.44, + "mean_token_accuracy": 0.8559488654136658, + "num_tokens": 68101369.0, + "step": 1751 + }, + { + "epoch": 0.2228724080905737, + "grad_norm": 2.932405948638916, + "learning_rate": 7.422636710470537e-07, + "loss": 0.3939, + "mean_token_accuracy": 0.8649497628211975, + "num_tokens": 68139672.0, + "step": 1752 + }, + { + "epoch": 0.22299961836916424, + "grad_norm": 3.0175726413726807, + "learning_rate": 7.426875794828317e-07, + "loss": 0.3913, + "mean_token_accuracy": 0.8650310039520264, + "num_tokens": 68174831.0, + "step": 1753 + }, + { + "epoch": 0.22312682864775474, + "grad_norm": 2.7186529636383057, + "learning_rate": 7.431114879186095e-07, + "loss": 0.375, + "mean_token_accuracy": 0.8734915256500244, + "num_tokens": 68214337.0, + "step": 1754 + }, + { + "epoch": 0.22325403892634524, + "grad_norm": 2.6637461185455322, + "learning_rate": 7.435353963543875e-07, + "loss": 0.3737, + "mean_token_accuracy": 0.8718650937080383, + "num_tokens": 68257082.0, + "step": 1755 + }, + { + "epoch": 0.22338124920493577, + "grad_norm": 2.6690118312835693, + "learning_rate": 7.439593047901653e-07, + "loss": 0.3578, + "mean_token_accuracy": 0.8766175508499146, + "num_tokens": 68300864.0, + "step": 1756 + }, + { + "epoch": 0.22350845948352627, + "grad_norm": 2.624969005584717, + "learning_rate": 7.443832132259431e-07, + "loss": 0.3716, + "mean_token_accuracy": 0.8695794939994812, + "num_tokens": 68342483.0, + "step": 1757 + }, + { + "epoch": 0.22363566976211677, + "grad_norm": 2.6662750244140625, + "learning_rate": 7.44807121661721e-07, + "loss": 0.3832, + "mean_token_accuracy": 0.8679865002632141, + "num_tokens": 68386077.0, + "step": 1758 + }, + { + "epoch": 0.2237628800407073, + "grad_norm": 2.523250102996826, + "learning_rate": 7.452310300974989e-07, + "loss": 0.3589, + "mean_token_accuracy": 0.8772316575050354, + "num_tokens": 68429287.0, + "step": 1759 + }, + { + "epoch": 0.2238900903192978, + "grad_norm": 2.990875244140625, + "learning_rate": 7.456549385332767e-07, + "loss": 0.4058, + "mean_token_accuracy": 0.8576586246490479, + "num_tokens": 68466325.0, + "step": 1760 + }, + { + "epoch": 0.2240173005978883, + "grad_norm": 3.0855319499969482, + "learning_rate": 7.460788469690547e-07, + "loss": 0.4474, + "mean_token_accuracy": 0.8515853881835938, + "num_tokens": 68504129.0, + "step": 1761 + }, + { + "epoch": 0.22414451087647883, + "grad_norm": 3.0330495834350586, + "learning_rate": 7.465027554048325e-07, + "loss": 0.3679, + "mean_token_accuracy": 0.8723500967025757, + "num_tokens": 68541010.0, + "step": 1762 + }, + { + "epoch": 0.22427172115506933, + "grad_norm": 3.1764681339263916, + "learning_rate": 7.469266638406105e-07, + "loss": 0.4082, + "mean_token_accuracy": 0.8601372241973877, + "num_tokens": 68572759.0, + "step": 1763 + }, + { + "epoch": 0.22439893143365983, + "grad_norm": 2.9185760021209717, + "learning_rate": 7.473505722763883e-07, + "loss": 0.3779, + "mean_token_accuracy": 0.8667859435081482, + "num_tokens": 68608332.0, + "step": 1764 + }, + { + "epoch": 0.22452614171225035, + "grad_norm": 3.078636646270752, + "learning_rate": 7.477744807121661e-07, + "loss": 0.3823, + "mean_token_accuracy": 0.8716993927955627, + "num_tokens": 68649535.0, + "step": 1765 + }, + { + "epoch": 0.22465335199084085, + "grad_norm": 2.795527935028076, + "learning_rate": 7.48198389147944e-07, + "loss": 0.4053, + "mean_token_accuracy": 0.8604768514633179, + "num_tokens": 68690956.0, + "step": 1766 + }, + { + "epoch": 0.22478056226943138, + "grad_norm": 2.8415606021881104, + "learning_rate": 7.486222975837219e-07, + "loss": 0.3902, + "mean_token_accuracy": 0.8660374283790588, + "num_tokens": 68728254.0, + "step": 1767 + }, + { + "epoch": 0.22490777254802188, + "grad_norm": 2.8715317249298096, + "learning_rate": 7.490462060194997e-07, + "loss": 0.4206, + "mean_token_accuracy": 0.8564321398735046, + "num_tokens": 68768860.0, + "step": 1768 + }, + { + "epoch": 0.22503498282661238, + "grad_norm": 2.9837353229522705, + "learning_rate": 7.494701144552777e-07, + "loss": 0.3816, + "mean_token_accuracy": 0.8689303994178772, + "num_tokens": 68803237.0, + "step": 1769 + }, + { + "epoch": 0.2251621931052029, + "grad_norm": 2.7742018699645996, + "learning_rate": 7.498940228910555e-07, + "loss": 0.4088, + "mean_token_accuracy": 0.8630954027175903, + "num_tokens": 68843407.0, + "step": 1770 + }, + { + "epoch": 0.2252894033837934, + "grad_norm": 3.15116810798645, + "learning_rate": 7.503179313268335e-07, + "loss": 0.3912, + "mean_token_accuracy": 0.866652250289917, + "num_tokens": 68880811.0, + "step": 1771 + }, + { + "epoch": 0.2254166136623839, + "grad_norm": 3.0048739910125732, + "learning_rate": 7.507418397626113e-07, + "loss": 0.3975, + "mean_token_accuracy": 0.861541748046875, + "num_tokens": 68916947.0, + "step": 1772 + }, + { + "epoch": 0.22554382394097444, + "grad_norm": 2.660377264022827, + "learning_rate": 7.51165748198389e-07, + "loss": 0.3918, + "mean_token_accuracy": 0.8640377521514893, + "num_tokens": 68960463.0, + "step": 1773 + }, + { + "epoch": 0.22567103421956494, + "grad_norm": 2.6597979068756104, + "learning_rate": 7.51589656634167e-07, + "loss": 0.3666, + "mean_token_accuracy": 0.8737189769744873, + "num_tokens": 69003751.0, + "step": 1774 + }, + { + "epoch": 0.22579824449815544, + "grad_norm": 3.027491807937622, + "learning_rate": 7.520135650699448e-07, + "loss": 0.4096, + "mean_token_accuracy": 0.8631210923194885, + "num_tokens": 69040394.0, + "step": 1775 + }, + { + "epoch": 0.22592545477674597, + "grad_norm": 2.792980194091797, + "learning_rate": 7.524374735057227e-07, + "loss": 0.377, + "mean_token_accuracy": 0.8697757720947266, + "num_tokens": 69080203.0, + "step": 1776 + }, + { + "epoch": 0.22605266505533647, + "grad_norm": 2.941863775253296, + "learning_rate": 7.528613819415006e-07, + "loss": 0.398, + "mean_token_accuracy": 0.8646987080574036, + "num_tokens": 69116891.0, + "step": 1777 + }, + { + "epoch": 0.22617987533392697, + "grad_norm": 2.8366293907165527, + "learning_rate": 7.532852903772785e-07, + "loss": 0.4241, + "mean_token_accuracy": 0.8547126054763794, + "num_tokens": 69159177.0, + "step": 1778 + }, + { + "epoch": 0.2263070856125175, + "grad_norm": 3.049227714538574, + "learning_rate": 7.537091988130564e-07, + "loss": 0.4718, + "mean_token_accuracy": 0.8449692726135254, + "num_tokens": 69195758.0, + "step": 1779 + }, + { + "epoch": 0.226434295891108, + "grad_norm": 3.204362392425537, + "learning_rate": 7.541331072488342e-07, + "loss": 0.4049, + "mean_token_accuracy": 0.863358736038208, + "num_tokens": 69227982.0, + "step": 1780 + }, + { + "epoch": 0.2265615061696985, + "grad_norm": 3.020324468612671, + "learning_rate": 7.54557015684612e-07, + "loss": 0.4274, + "mean_token_accuracy": 0.8530038595199585, + "num_tokens": 69261048.0, + "step": 1781 + }, + { + "epoch": 0.22668871644828903, + "grad_norm": 2.611207962036133, + "learning_rate": 7.5498092412039e-07, + "loss": 0.3475, + "mean_token_accuracy": 0.8796250820159912, + "num_tokens": 69300940.0, + "step": 1782 + }, + { + "epoch": 0.22681592672687953, + "grad_norm": 2.7823262214660645, + "learning_rate": 7.554048325561678e-07, + "loss": 0.4029, + "mean_token_accuracy": 0.8603762984275818, + "num_tokens": 69345596.0, + "step": 1783 + }, + { + "epoch": 0.22694313700547003, + "grad_norm": 2.889867067337036, + "learning_rate": 7.558287409919457e-07, + "loss": 0.4097, + "mean_token_accuracy": 0.8602892160415649, + "num_tokens": 69385205.0, + "step": 1784 + }, + { + "epoch": 0.22707034728406056, + "grad_norm": 2.613910436630249, + "learning_rate": 7.562526494277236e-07, + "loss": 0.3491, + "mean_token_accuracy": 0.8786802291870117, + "num_tokens": 69425908.0, + "step": 1785 + }, + { + "epoch": 0.22719755756265106, + "grad_norm": 2.7920913696289062, + "learning_rate": 7.566765578635015e-07, + "loss": 0.4085, + "mean_token_accuracy": 0.8610711097717285, + "num_tokens": 69465375.0, + "step": 1786 + }, + { + "epoch": 0.22732476784124156, + "grad_norm": 2.9400529861450195, + "learning_rate": 7.571004662992794e-07, + "loss": 0.4105, + "mean_token_accuracy": 0.8596174716949463, + "num_tokens": 69501293.0, + "step": 1787 + }, + { + "epoch": 0.2274519781198321, + "grad_norm": 2.6839535236358643, + "learning_rate": 7.575243747350572e-07, + "loss": 0.3747, + "mean_token_accuracy": 0.8730767965316772, + "num_tokens": 69542386.0, + "step": 1788 + }, + { + "epoch": 0.2275791883984226, + "grad_norm": 2.613703966140747, + "learning_rate": 7.57948283170835e-07, + "loss": 0.3412, + "mean_token_accuracy": 0.8828247785568237, + "num_tokens": 69579191.0, + "step": 1789 + }, + { + "epoch": 0.2277063986770131, + "grad_norm": 2.6208465099334717, + "learning_rate": 7.58372191606613e-07, + "loss": 0.3817, + "mean_token_accuracy": 0.8716787099838257, + "num_tokens": 69623479.0, + "step": 1790 + }, + { + "epoch": 0.22783360895560362, + "grad_norm": 2.9992995262145996, + "learning_rate": 7.587961000423908e-07, + "loss": 0.4452, + "mean_token_accuracy": 0.8483337163925171, + "num_tokens": 69665379.0, + "step": 1791 + }, + { + "epoch": 0.22796081923419412, + "grad_norm": 2.9846889972686768, + "learning_rate": 7.592200084781686e-07, + "loss": 0.3826, + "mean_token_accuracy": 0.8706393241882324, + "num_tokens": 69700366.0, + "step": 1792 + }, + { + "epoch": 0.22808802951278465, + "grad_norm": 2.831772804260254, + "learning_rate": 7.596439169139466e-07, + "loss": 0.4163, + "mean_token_accuracy": 0.8584717512130737, + "num_tokens": 69740868.0, + "step": 1793 + }, + { + "epoch": 0.22821523979137515, + "grad_norm": 3.1015288829803467, + "learning_rate": 7.600678253497244e-07, + "loss": 0.3903, + "mean_token_accuracy": 0.8672797679901123, + "num_tokens": 69774978.0, + "step": 1794 + }, + { + "epoch": 0.22834245006996565, + "grad_norm": 2.9491796493530273, + "learning_rate": 7.604917337855023e-07, + "loss": 0.3981, + "mean_token_accuracy": 0.8664374947547913, + "num_tokens": 69814488.0, + "step": 1795 + }, + { + "epoch": 0.22846966034855618, + "grad_norm": 2.66658878326416, + "learning_rate": 7.609156422212801e-07, + "loss": 0.359, + "mean_token_accuracy": 0.8764264583587646, + "num_tokens": 69857317.0, + "step": 1796 + }, + { + "epoch": 0.22859687062714668, + "grad_norm": 2.7501440048217773, + "learning_rate": 7.61339550657058e-07, + "loss": 0.382, + "mean_token_accuracy": 0.8712995052337646, + "num_tokens": 69899524.0, + "step": 1797 + }, + { + "epoch": 0.22872408090573718, + "grad_norm": 2.9245195388793945, + "learning_rate": 7.617634590928359e-07, + "loss": 0.4491, + "mean_token_accuracy": 0.8460878133773804, + "num_tokens": 69942365.0, + "step": 1798 + }, + { + "epoch": 0.2288512911843277, + "grad_norm": 2.861355781555176, + "learning_rate": 7.621873675286138e-07, + "loss": 0.377, + "mean_token_accuracy": 0.8737715482711792, + "num_tokens": 69977926.0, + "step": 1799 + }, + { + "epoch": 0.2289785014629182, + "grad_norm": 3.028424024581909, + "learning_rate": 7.626112759643916e-07, + "loss": 0.3802, + "mean_token_accuracy": 0.869407057762146, + "num_tokens": 70013220.0, + "step": 1800 + }, + { + "epoch": 0.2291057117415087, + "grad_norm": 2.736903190612793, + "learning_rate": 7.630351844001696e-07, + "loss": 0.3939, + "mean_token_accuracy": 0.8657987117767334, + "num_tokens": 70052538.0, + "step": 1801 + }, + { + "epoch": 0.22923292202009923, + "grad_norm": 2.6300086975097656, + "learning_rate": 7.634590928359474e-07, + "loss": 0.3209, + "mean_token_accuracy": 0.8874047994613647, + "num_tokens": 70089270.0, + "step": 1802 + }, + { + "epoch": 0.22936013229868973, + "grad_norm": 2.8221771717071533, + "learning_rate": 7.638830012717253e-07, + "loss": 0.4026, + "mean_token_accuracy": 0.8608574271202087, + "num_tokens": 70128183.0, + "step": 1803 + }, + { + "epoch": 0.22948734257728023, + "grad_norm": 3.0764379501342773, + "learning_rate": 7.643069097075031e-07, + "loss": 0.4007, + "mean_token_accuracy": 0.8634796142578125, + "num_tokens": 70159019.0, + "step": 1804 + }, + { + "epoch": 0.22961455285587076, + "grad_norm": 3.0240302085876465, + "learning_rate": 7.64730818143281e-07, + "loss": 0.4302, + "mean_token_accuracy": 0.8593253493309021, + "num_tokens": 70196572.0, + "step": 1805 + }, + { + "epoch": 0.22974176313446126, + "grad_norm": 2.73022723197937, + "learning_rate": 7.651547265790589e-07, + "loss": 0.3515, + "mean_token_accuracy": 0.8742070198059082, + "num_tokens": 70232381.0, + "step": 1806 + }, + { + "epoch": 0.22986897341305176, + "grad_norm": 2.9177441596984863, + "learning_rate": 7.655786350148368e-07, + "loss": 0.3784, + "mean_token_accuracy": 0.8714219331741333, + "num_tokens": 70269114.0, + "step": 1807 + }, + { + "epoch": 0.2299961836916423, + "grad_norm": 2.898745059967041, + "learning_rate": 7.660025434506146e-07, + "loss": 0.4296, + "mean_token_accuracy": 0.8489365577697754, + "num_tokens": 70308765.0, + "step": 1808 + }, + { + "epoch": 0.2301233939702328, + "grad_norm": 2.960167646408081, + "learning_rate": 7.664264518863926e-07, + "loss": 0.434, + "mean_token_accuracy": 0.8572535514831543, + "num_tokens": 70347917.0, + "step": 1809 + }, + { + "epoch": 0.2302506042488233, + "grad_norm": 2.71694016456604, + "learning_rate": 7.668503603221704e-07, + "loss": 0.3734, + "mean_token_accuracy": 0.8742997646331787, + "num_tokens": 70386933.0, + "step": 1810 + }, + { + "epoch": 0.23037781452741382, + "grad_norm": 3.0113823413848877, + "learning_rate": 7.672742687579483e-07, + "loss": 0.3628, + "mean_token_accuracy": 0.8729735612869263, + "num_tokens": 70417531.0, + "step": 1811 + }, + { + "epoch": 0.23050502480600432, + "grad_norm": 2.78861403465271, + "learning_rate": 7.676981771937261e-07, + "loss": 0.3687, + "mean_token_accuracy": 0.8692247867584229, + "num_tokens": 70455553.0, + "step": 1812 + }, + { + "epoch": 0.23063223508459482, + "grad_norm": 2.697225332260132, + "learning_rate": 7.681220856295039e-07, + "loss": 0.3564, + "mean_token_accuracy": 0.877768874168396, + "num_tokens": 70496423.0, + "step": 1813 + }, + { + "epoch": 0.23075944536318535, + "grad_norm": 2.777573585510254, + "learning_rate": 7.685459940652819e-07, + "loss": 0.351, + "mean_token_accuracy": 0.875769853591919, + "num_tokens": 70530886.0, + "step": 1814 + }, + { + "epoch": 0.23088665564177585, + "grad_norm": 2.6799676418304443, + "learning_rate": 7.689699025010597e-07, + "loss": 0.3589, + "mean_token_accuracy": 0.8769513964653015, + "num_tokens": 70569656.0, + "step": 1815 + }, + { + "epoch": 0.23101386592036638, + "grad_norm": 2.9873666763305664, + "learning_rate": 7.693938109368376e-07, + "loss": 0.4112, + "mean_token_accuracy": 0.8563897013664246, + "num_tokens": 70606495.0, + "step": 1816 + }, + { + "epoch": 0.23114107619895688, + "grad_norm": 3.024700403213501, + "learning_rate": 7.698177193726155e-07, + "loss": 0.3629, + "mean_token_accuracy": 0.877422571182251, + "num_tokens": 70641230.0, + "step": 1817 + }, + { + "epoch": 0.23126828647754738, + "grad_norm": 2.5205726623535156, + "learning_rate": 7.702416278083933e-07, + "loss": 0.3792, + "mean_token_accuracy": 0.868195116519928, + "num_tokens": 70687760.0, + "step": 1818 + }, + { + "epoch": 0.2313954967561379, + "grad_norm": 2.783719539642334, + "learning_rate": 7.706655362441712e-07, + "loss": 0.3868, + "mean_token_accuracy": 0.8687853217124939, + "num_tokens": 70726799.0, + "step": 1819 + }, + { + "epoch": 0.2315227070347284, + "grad_norm": 2.679182529449463, + "learning_rate": 7.710894446799491e-07, + "loss": 0.3665, + "mean_token_accuracy": 0.8735589385032654, + "num_tokens": 70765962.0, + "step": 1820 + }, + { + "epoch": 0.2316499173133189, + "grad_norm": 2.798351287841797, + "learning_rate": 7.715133531157269e-07, + "loss": 0.3928, + "mean_token_accuracy": 0.8661843538284302, + "num_tokens": 70807474.0, + "step": 1821 + }, + { + "epoch": 0.23177712759190944, + "grad_norm": 2.9240360260009766, + "learning_rate": 7.719372615515049e-07, + "loss": 0.4429, + "mean_token_accuracy": 0.8499187231063843, + "num_tokens": 70849344.0, + "step": 1822 + }, + { + "epoch": 0.23190433787049994, + "grad_norm": 2.8896989822387695, + "learning_rate": 7.723611699872827e-07, + "loss": 0.4176, + "mean_token_accuracy": 0.8577767610549927, + "num_tokens": 70890723.0, + "step": 1823 + }, + { + "epoch": 0.23203154814909044, + "grad_norm": 3.1100244522094727, + "learning_rate": 7.727850784230606e-07, + "loss": 0.4211, + "mean_token_accuracy": 0.8568181991577148, + "num_tokens": 70924939.0, + "step": 1824 + }, + { + "epoch": 0.23215875842768097, + "grad_norm": 2.5271153450012207, + "learning_rate": 7.732089868588385e-07, + "loss": 0.3257, + "mean_token_accuracy": 0.8857852220535278, + "num_tokens": 70965880.0, + "step": 1825 + }, + { + "epoch": 0.23228596870627147, + "grad_norm": 2.757824420928955, + "learning_rate": 7.736328952946163e-07, + "loss": 0.3695, + "mean_token_accuracy": 0.8721189498901367, + "num_tokens": 71004449.0, + "step": 1826 + }, + { + "epoch": 0.23241317898486197, + "grad_norm": 3.073126792907715, + "learning_rate": 7.740568037303942e-07, + "loss": 0.4265, + "mean_token_accuracy": 0.8536704182624817, + "num_tokens": 71043041.0, + "step": 1827 + }, + { + "epoch": 0.2325403892634525, + "grad_norm": 2.9470605850219727, + "learning_rate": 7.744807121661721e-07, + "loss": 0.4248, + "mean_token_accuracy": 0.8568342328071594, + "num_tokens": 71080991.0, + "step": 1828 + }, + { + "epoch": 0.232667599542043, + "grad_norm": 2.614344835281372, + "learning_rate": 7.749046206019499e-07, + "loss": 0.3845, + "mean_token_accuracy": 0.8659958839416504, + "num_tokens": 71126023.0, + "step": 1829 + }, + { + "epoch": 0.2327948098206335, + "grad_norm": 2.8934566974639893, + "learning_rate": 7.753285290377279e-07, + "loss": 0.4182, + "mean_token_accuracy": 0.8597373366355896, + "num_tokens": 71164580.0, + "step": 1830 + }, + { + "epoch": 0.23292202009922403, + "grad_norm": 3.31156063079834, + "learning_rate": 7.757524374735057e-07, + "loss": 0.3988, + "mean_token_accuracy": 0.8632115125656128, + "num_tokens": 71195183.0, + "step": 1831 + }, + { + "epoch": 0.23304923037781453, + "grad_norm": 2.7712273597717285, + "learning_rate": 7.761763459092836e-07, + "loss": 0.438, + "mean_token_accuracy": 0.8522025346755981, + "num_tokens": 71241311.0, + "step": 1832 + }, + { + "epoch": 0.23317644065640503, + "grad_norm": 2.583967685699463, + "learning_rate": 7.766002543450614e-07, + "loss": 0.3635, + "mean_token_accuracy": 0.8778215646743774, + "num_tokens": 71283751.0, + "step": 1833 + }, + { + "epoch": 0.23330365093499555, + "grad_norm": 2.713562250137329, + "learning_rate": 7.770241627808392e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8665282726287842, + "num_tokens": 71322560.0, + "step": 1834 + }, + { + "epoch": 0.23343086121358606, + "grad_norm": 2.803021192550659, + "learning_rate": 7.774480712166172e-07, + "loss": 0.3287, + "mean_token_accuracy": 0.8858771324157715, + "num_tokens": 71357021.0, + "step": 1835 + }, + { + "epoch": 0.23355807149217656, + "grad_norm": 2.8462538719177246, + "learning_rate": 7.77871979652395e-07, + "loss": 0.4076, + "mean_token_accuracy": 0.8595286011695862, + "num_tokens": 71394786.0, + "step": 1836 + }, + { + "epoch": 0.23368528177076708, + "grad_norm": 2.772392988204956, + "learning_rate": 7.782958880881729e-07, + "loss": 0.412, + "mean_token_accuracy": 0.8622748851776123, + "num_tokens": 71439644.0, + "step": 1837 + }, + { + "epoch": 0.23381249204935758, + "grad_norm": 2.9282562732696533, + "learning_rate": 7.787197965239508e-07, + "loss": 0.3753, + "mean_token_accuracy": 0.868423581123352, + "num_tokens": 71474784.0, + "step": 1838 + }, + { + "epoch": 0.23393970232794808, + "grad_norm": 2.563732385635376, + "learning_rate": 7.791437049597287e-07, + "loss": 0.3724, + "mean_token_accuracy": 0.8681581616401672, + "num_tokens": 71516342.0, + "step": 1839 + }, + { + "epoch": 0.2340669126065386, + "grad_norm": 2.88539719581604, + "learning_rate": 7.795676133955065e-07, + "loss": 0.3989, + "mean_token_accuracy": 0.8668862581253052, + "num_tokens": 71551950.0, + "step": 1840 + }, + { + "epoch": 0.2341941228851291, + "grad_norm": 3.2943267822265625, + "learning_rate": 7.799915218312844e-07, + "loss": 0.3844, + "mean_token_accuracy": 0.8684974312782288, + "num_tokens": 71582159.0, + "step": 1841 + }, + { + "epoch": 0.23432133316371964, + "grad_norm": 2.783418655395508, + "learning_rate": 7.804154302670622e-07, + "loss": 0.4229, + "mean_token_accuracy": 0.8584375381469727, + "num_tokens": 71627003.0, + "step": 1842 + }, + { + "epoch": 0.23444854344231014, + "grad_norm": 2.6444404125213623, + "learning_rate": 7.808393387028402e-07, + "loss": 0.3605, + "mean_token_accuracy": 0.8730612397193909, + "num_tokens": 71667149.0, + "step": 1843 + }, + { + "epoch": 0.23457575372090064, + "grad_norm": 2.899057388305664, + "learning_rate": 7.81263247138618e-07, + "loss": 0.3778, + "mean_token_accuracy": 0.8705747127532959, + "num_tokens": 71702841.0, + "step": 1844 + }, + { + "epoch": 0.23470296399949117, + "grad_norm": 2.714299440383911, + "learning_rate": 7.816871555743959e-07, + "loss": 0.4006, + "mean_token_accuracy": 0.8622262477874756, + "num_tokens": 71745043.0, + "step": 1845 + }, + { + "epoch": 0.23483017427808167, + "grad_norm": 2.8459854125976562, + "learning_rate": 7.821110640101738e-07, + "loss": 0.4182, + "mean_token_accuracy": 0.8583024740219116, + "num_tokens": 71787282.0, + "step": 1846 + }, + { + "epoch": 0.23495738455667217, + "grad_norm": 2.7212061882019043, + "learning_rate": 7.825349724459517e-07, + "loss": 0.372, + "mean_token_accuracy": 0.8688668608665466, + "num_tokens": 71824605.0, + "step": 1847 + }, + { + "epoch": 0.2350845948352627, + "grad_norm": 2.6001808643341064, + "learning_rate": 7.829588808817294e-07, + "loss": 0.3849, + "mean_token_accuracy": 0.8681199550628662, + "num_tokens": 71869198.0, + "step": 1848 + }, + { + "epoch": 0.2352118051138532, + "grad_norm": 3.020299196243286, + "learning_rate": 7.833827893175074e-07, + "loss": 0.4318, + "mean_token_accuracy": 0.8548188209533691, + "num_tokens": 71907615.0, + "step": 1849 + }, + { + "epoch": 0.2353390153924437, + "grad_norm": 2.695849657058716, + "learning_rate": 7.838066977532852e-07, + "loss": 0.3403, + "mean_token_accuracy": 0.8832306861877441, + "num_tokens": 71945588.0, + "step": 1850 + }, + { + "epoch": 0.23546622567103423, + "grad_norm": 2.974705457687378, + "learning_rate": 7.842306061890632e-07, + "loss": 0.4142, + "mean_token_accuracy": 0.8572577834129333, + "num_tokens": 71981560.0, + "step": 1851 + }, + { + "epoch": 0.23559343594962473, + "grad_norm": 2.7735867500305176, + "learning_rate": 7.84654514624841e-07, + "loss": 0.3455, + "mean_token_accuracy": 0.8776079416275024, + "num_tokens": 72018573.0, + "step": 1852 + }, + { + "epoch": 0.23572064622821523, + "grad_norm": 2.811687469482422, + "learning_rate": 7.850784230606188e-07, + "loss": 0.4056, + "mean_token_accuracy": 0.8611825704574585, + "num_tokens": 72059630.0, + "step": 1853 + }, + { + "epoch": 0.23584785650680576, + "grad_norm": 3.0168161392211914, + "learning_rate": 7.855023314963968e-07, + "loss": 0.3954, + "mean_token_accuracy": 0.8663385510444641, + "num_tokens": 72093658.0, + "step": 1854 + }, + { + "epoch": 0.23597506678539626, + "grad_norm": 2.8296477794647217, + "learning_rate": 7.859262399321746e-07, + "loss": 0.3934, + "mean_token_accuracy": 0.8669779300689697, + "num_tokens": 72135743.0, + "step": 1855 + }, + { + "epoch": 0.23610227706398676, + "grad_norm": 2.9551122188568115, + "learning_rate": 7.863501483679524e-07, + "loss": 0.3758, + "mean_token_accuracy": 0.8725106120109558, + "num_tokens": 72175532.0, + "step": 1856 + }, + { + "epoch": 0.2362294873425773, + "grad_norm": 2.792598009109497, + "learning_rate": 7.867740568037303e-07, + "loss": 0.4128, + "mean_token_accuracy": 0.8611942529678345, + "num_tokens": 72217451.0, + "step": 1857 + }, + { + "epoch": 0.2363566976211678, + "grad_norm": 2.8562278747558594, + "learning_rate": 7.871979652395082e-07, + "loss": 0.3523, + "mean_token_accuracy": 0.8742619752883911, + "num_tokens": 72254758.0, + "step": 1858 + }, + { + "epoch": 0.2364839078997583, + "grad_norm": 2.6124770641326904, + "learning_rate": 7.876218736752861e-07, + "loss": 0.3693, + "mean_token_accuracy": 0.8720213770866394, + "num_tokens": 72294587.0, + "step": 1859 + }, + { + "epoch": 0.23661111817834882, + "grad_norm": 2.829223871231079, + "learning_rate": 7.88045782111064e-07, + "loss": 0.3805, + "mean_token_accuracy": 0.8706479072570801, + "num_tokens": 72336578.0, + "step": 1860 + }, + { + "epoch": 0.23673832845693932, + "grad_norm": 2.882211208343506, + "learning_rate": 7.884696905468418e-07, + "loss": 0.4068, + "mean_token_accuracy": 0.8610963821411133, + "num_tokens": 72378170.0, + "step": 1861 + }, + { + "epoch": 0.23686553873552982, + "grad_norm": 2.542160749435425, + "learning_rate": 7.888935989826198e-07, + "loss": 0.3785, + "mean_token_accuracy": 0.8711131811141968, + "num_tokens": 72425226.0, + "step": 1862 + }, + { + "epoch": 0.23699274901412035, + "grad_norm": 2.737902879714966, + "learning_rate": 7.893175074183976e-07, + "loss": 0.3814, + "mean_token_accuracy": 0.8691531419754028, + "num_tokens": 72466586.0, + "step": 1863 + }, + { + "epoch": 0.23711995929271085, + "grad_norm": 2.5751616954803467, + "learning_rate": 7.897414158541754e-07, + "loss": 0.3793, + "mean_token_accuracy": 0.8719178438186646, + "num_tokens": 72509595.0, + "step": 1864 + }, + { + "epoch": 0.23724716957130135, + "grad_norm": 2.795928478240967, + "learning_rate": 7.901653242899533e-07, + "loss": 0.3676, + "mean_token_accuracy": 0.8730107545852661, + "num_tokens": 72546522.0, + "step": 1865 + }, + { + "epoch": 0.23737437984989188, + "grad_norm": 2.9414172172546387, + "learning_rate": 7.905892327257312e-07, + "loss": 0.4323, + "mean_token_accuracy": 0.8539097309112549, + "num_tokens": 72586502.0, + "step": 1866 + }, + { + "epoch": 0.23750159012848238, + "grad_norm": 2.9720427989959717, + "learning_rate": 7.910131411615091e-07, + "loss": 0.4252, + "mean_token_accuracy": 0.8565196990966797, + "num_tokens": 72625214.0, + "step": 1867 + }, + { + "epoch": 0.2376288004070729, + "grad_norm": 2.6259677410125732, + "learning_rate": 7.91437049597287e-07, + "loss": 0.3399, + "mean_token_accuracy": 0.8824438452720642, + "num_tokens": 72664770.0, + "step": 1868 + }, + { + "epoch": 0.2377560106856634, + "grad_norm": 3.306872606277466, + "learning_rate": 7.918609580330648e-07, + "loss": 0.349, + "mean_token_accuracy": 0.8800094127655029, + "num_tokens": 72699706.0, + "step": 1869 + }, + { + "epoch": 0.2378832209642539, + "grad_norm": 4.085806846618652, + "learning_rate": 7.922848664688428e-07, + "loss": 0.3908, + "mean_token_accuracy": 0.8660739660263062, + "num_tokens": 72737135.0, + "step": 1870 + }, + { + "epoch": 0.23801043124284443, + "grad_norm": 2.7986373901367188, + "learning_rate": 7.927087749046205e-07, + "loss": 0.4421, + "mean_token_accuracy": 0.8476828932762146, + "num_tokens": 72780178.0, + "step": 1871 + }, + { + "epoch": 0.23813764152143493, + "grad_norm": 2.9435269832611084, + "learning_rate": 7.931326833403983e-07, + "loss": 0.434, + "mean_token_accuracy": 0.8539656400680542, + "num_tokens": 72817230.0, + "step": 1872 + }, + { + "epoch": 0.23826485180002543, + "grad_norm": 2.7444097995758057, + "learning_rate": 7.935565917761763e-07, + "loss": 0.3781, + "mean_token_accuracy": 0.8681749105453491, + "num_tokens": 72855893.0, + "step": 1873 + }, + { + "epoch": 0.23839206207861596, + "grad_norm": 2.6817214488983154, + "learning_rate": 7.939805002119541e-07, + "loss": 0.3673, + "mean_token_accuracy": 0.8725267648696899, + "num_tokens": 72894428.0, + "step": 1874 + }, + { + "epoch": 0.23851927235720646, + "grad_norm": 2.935535430908203, + "learning_rate": 7.944044086477321e-07, + "loss": 0.4449, + "mean_token_accuracy": 0.8504565954208374, + "num_tokens": 72936115.0, + "step": 1875 + }, + { + "epoch": 0.23864648263579696, + "grad_norm": 2.7501986026763916, + "learning_rate": 7.948283170835099e-07, + "loss": 0.3649, + "mean_token_accuracy": 0.8744341135025024, + "num_tokens": 72977115.0, + "step": 1876 + }, + { + "epoch": 0.2387736929143875, + "grad_norm": 2.8125996589660645, + "learning_rate": 7.952522255192878e-07, + "loss": 0.3947, + "mean_token_accuracy": 0.8640556335449219, + "num_tokens": 73017065.0, + "step": 1877 + }, + { + "epoch": 0.238900903192978, + "grad_norm": 2.938926935195923, + "learning_rate": 7.956761339550657e-07, + "loss": 0.4078, + "mean_token_accuracy": 0.8651566505432129, + "num_tokens": 73054644.0, + "step": 1878 + }, + { + "epoch": 0.2390281134715685, + "grad_norm": 2.905210018157959, + "learning_rate": 7.961000423908435e-07, + "loss": 0.3467, + "mean_token_accuracy": 0.876672089099884, + "num_tokens": 73090769.0, + "step": 1879 + }, + { + "epoch": 0.23915532375015902, + "grad_norm": 2.4467484951019287, + "learning_rate": 7.965239508266214e-07, + "loss": 0.3518, + "mean_token_accuracy": 0.8791942596435547, + "num_tokens": 73134025.0, + "step": 1880 + }, + { + "epoch": 0.23928253402874952, + "grad_norm": 2.6479036808013916, + "learning_rate": 7.969478592623993e-07, + "loss": 0.3571, + "mean_token_accuracy": 0.8777003884315491, + "num_tokens": 73175871.0, + "step": 1881 + }, + { + "epoch": 0.23940974430734002, + "grad_norm": 2.8195273876190186, + "learning_rate": 7.973717676981771e-07, + "loss": 0.416, + "mean_token_accuracy": 0.8551164865493774, + "num_tokens": 73216970.0, + "step": 1882 + }, + { + "epoch": 0.23953695458593055, + "grad_norm": 2.706282615661621, + "learning_rate": 7.977956761339551e-07, + "loss": 0.3814, + "mean_token_accuracy": 0.8668115139007568, + "num_tokens": 73256483.0, + "step": 1883 + }, + { + "epoch": 0.23966416486452105, + "grad_norm": 2.886357069015503, + "learning_rate": 7.982195845697329e-07, + "loss": 0.4015, + "mean_token_accuracy": 0.862208902835846, + "num_tokens": 73296086.0, + "step": 1884 + }, + { + "epoch": 0.23979137514311155, + "grad_norm": 2.9809587001800537, + "learning_rate": 7.986434930055108e-07, + "loss": 0.4637, + "mean_token_accuracy": 0.8416345119476318, + "num_tokens": 73334994.0, + "step": 1885 + }, + { + "epoch": 0.23991858542170208, + "grad_norm": 3.111365795135498, + "learning_rate": 7.990674014412886e-07, + "loss": 0.4108, + "mean_token_accuracy": 0.8620800375938416, + "num_tokens": 73366069.0, + "step": 1886 + }, + { + "epoch": 0.24004579570029258, + "grad_norm": 3.02980899810791, + "learning_rate": 7.994913098770665e-07, + "loss": 0.4089, + "mean_token_accuracy": 0.8588168025016785, + "num_tokens": 73404831.0, + "step": 1887 + }, + { + "epoch": 0.24017300597888308, + "grad_norm": 2.6134231090545654, + "learning_rate": 7.999152183128444e-07, + "loss": 0.364, + "mean_token_accuracy": 0.873831033706665, + "num_tokens": 73448033.0, + "step": 1888 + }, + { + "epoch": 0.2403002162574736, + "grad_norm": 2.8885750770568848, + "learning_rate": 8.003391267486223e-07, + "loss": 0.3747, + "mean_token_accuracy": 0.8726134896278381, + "num_tokens": 73484239.0, + "step": 1889 + }, + { + "epoch": 0.2404274265360641, + "grad_norm": 2.5510332584381104, + "learning_rate": 8.007630351844001e-07, + "loss": 0.316, + "mean_token_accuracy": 0.8914175033569336, + "num_tokens": 73523324.0, + "step": 1890 + }, + { + "epoch": 0.24055463681465464, + "grad_norm": 2.8397655487060547, + "learning_rate": 8.011869436201781e-07, + "loss": 0.4109, + "mean_token_accuracy": 0.860582172870636, + "num_tokens": 73562130.0, + "step": 1891 + }, + { + "epoch": 0.24068184709324514, + "grad_norm": 2.7635228633880615, + "learning_rate": 8.016108520559559e-07, + "loss": 0.3717, + "mean_token_accuracy": 0.870650589466095, + "num_tokens": 73598631.0, + "step": 1892 + }, + { + "epoch": 0.24080905737183564, + "grad_norm": 2.7019155025482178, + "learning_rate": 8.020347604917338e-07, + "loss": 0.3657, + "mean_token_accuracy": 0.8745185732841492, + "num_tokens": 73638567.0, + "step": 1893 + }, + { + "epoch": 0.24093626765042617, + "grad_norm": 2.8979742527008057, + "learning_rate": 8.024586689275116e-07, + "loss": 0.3748, + "mean_token_accuracy": 0.8684722781181335, + "num_tokens": 73673969.0, + "step": 1894 + }, + { + "epoch": 0.24106347792901667, + "grad_norm": 2.8204030990600586, + "learning_rate": 8.028825773632894e-07, + "loss": 0.3913, + "mean_token_accuracy": 0.8672354221343994, + "num_tokens": 73714354.0, + "step": 1895 + }, + { + "epoch": 0.24119068820760717, + "grad_norm": 2.9899401664733887, + "learning_rate": 8.033064857990674e-07, + "loss": 0.4743, + "mean_token_accuracy": 0.8414931297302246, + "num_tokens": 73754711.0, + "step": 1896 + }, + { + "epoch": 0.2413178984861977, + "grad_norm": 3.2978343963623047, + "learning_rate": 8.037303942348452e-07, + "loss": 0.4335, + "mean_token_accuracy": 0.8540897369384766, + "num_tokens": 73786547.0, + "step": 1897 + }, + { + "epoch": 0.2414451087647882, + "grad_norm": 2.7298941612243652, + "learning_rate": 8.041543026706231e-07, + "loss": 0.3635, + "mean_token_accuracy": 0.8765116930007935, + "num_tokens": 73823442.0, + "step": 1898 + }, + { + "epoch": 0.2415723190433787, + "grad_norm": 2.4949066638946533, + "learning_rate": 8.04578211106401e-07, + "loss": 0.3573, + "mean_token_accuracy": 0.8781011700630188, + "num_tokens": 73866820.0, + "step": 1899 + }, + { + "epoch": 0.24169952932196923, + "grad_norm": 2.7827863693237305, + "learning_rate": 8.050021195421789e-07, + "loss": 0.3781, + "mean_token_accuracy": 0.8704317212104797, + "num_tokens": 73907449.0, + "step": 1900 + }, + { + "epoch": 0.24182673960055973, + "grad_norm": 2.691837787628174, + "learning_rate": 8.054260279779567e-07, + "loss": 0.4183, + "mean_token_accuracy": 0.8570725917816162, + "num_tokens": 73952953.0, + "step": 1901 + }, + { + "epoch": 0.24195394987915023, + "grad_norm": 2.9038031101226807, + "learning_rate": 8.058499364137346e-07, + "loss": 0.4117, + "mean_token_accuracy": 0.8634049892425537, + "num_tokens": 73996114.0, + "step": 1902 + }, + { + "epoch": 0.24208116015774075, + "grad_norm": 2.745800018310547, + "learning_rate": 8.062738448495124e-07, + "loss": 0.3712, + "mean_token_accuracy": 0.8713580369949341, + "num_tokens": 74032663.0, + "step": 1903 + }, + { + "epoch": 0.24220837043633126, + "grad_norm": 2.6744987964630127, + "learning_rate": 8.066977532852904e-07, + "loss": 0.3832, + "mean_token_accuracy": 0.8695821166038513, + "num_tokens": 74073604.0, + "step": 1904 + }, + { + "epoch": 0.24233558071492176, + "grad_norm": 2.6007258892059326, + "learning_rate": 8.071216617210682e-07, + "loss": 0.3807, + "mean_token_accuracy": 0.8695422410964966, + "num_tokens": 74116583.0, + "step": 1905 + }, + { + "epoch": 0.24246279099351228, + "grad_norm": 2.7447946071624756, + "learning_rate": 8.075455701568461e-07, + "loss": 0.3977, + "mean_token_accuracy": 0.8640434741973877, + "num_tokens": 74159310.0, + "step": 1906 + }, + { + "epoch": 0.24259000127210278, + "grad_norm": 3.0329253673553467, + "learning_rate": 8.07969478592624e-07, + "loss": 0.4144, + "mean_token_accuracy": 0.856924295425415, + "num_tokens": 74194225.0, + "step": 1907 + }, + { + "epoch": 0.24271721155069328, + "grad_norm": 2.7611083984375, + "learning_rate": 8.083933870284019e-07, + "loss": 0.3713, + "mean_token_accuracy": 0.8696119785308838, + "num_tokens": 74234208.0, + "step": 1908 + }, + { + "epoch": 0.2428444218292838, + "grad_norm": 2.902132511138916, + "learning_rate": 8.088172954641796e-07, + "loss": 0.4257, + "mean_token_accuracy": 0.8554674983024597, + "num_tokens": 74274103.0, + "step": 1909 + }, + { + "epoch": 0.2429716321078743, + "grad_norm": 3.053511381149292, + "learning_rate": 8.092412038999576e-07, + "loss": 0.4835, + "mean_token_accuracy": 0.8370586037635803, + "num_tokens": 74314678.0, + "step": 1910 + }, + { + "epoch": 0.24309884238646481, + "grad_norm": 2.9548916816711426, + "learning_rate": 8.096651123357354e-07, + "loss": 0.3952, + "mean_token_accuracy": 0.863817572593689, + "num_tokens": 74350775.0, + "step": 1911 + }, + { + "epoch": 0.24322605266505534, + "grad_norm": 2.6004762649536133, + "learning_rate": 8.100890207715134e-07, + "loss": 0.3512, + "mean_token_accuracy": 0.8787002563476562, + "num_tokens": 74391634.0, + "step": 1912 + }, + { + "epoch": 0.24335326294364584, + "grad_norm": 2.865469455718994, + "learning_rate": 8.105129292072912e-07, + "loss": 0.3819, + "mean_token_accuracy": 0.8715255260467529, + "num_tokens": 74433816.0, + "step": 1913 + }, + { + "epoch": 0.24348047322223634, + "grad_norm": 2.965078592300415, + "learning_rate": 8.10936837643069e-07, + "loss": 0.3943, + "mean_token_accuracy": 0.8624683618545532, + "num_tokens": 74470220.0, + "step": 1914 + }, + { + "epoch": 0.24360768350082687, + "grad_norm": 3.11607027053833, + "learning_rate": 8.11360746078847e-07, + "loss": 0.4544, + "mean_token_accuracy": 0.8472959995269775, + "num_tokens": 74513853.0, + "step": 1915 + }, + { + "epoch": 0.24373489377941737, + "grad_norm": 2.6797828674316406, + "learning_rate": 8.117846545146248e-07, + "loss": 0.3551, + "mean_token_accuracy": 0.878173828125, + "num_tokens": 74553577.0, + "step": 1916 + }, + { + "epoch": 0.2438621040580079, + "grad_norm": 3.1708672046661377, + "learning_rate": 8.122085629504026e-07, + "loss": 0.4643, + "mean_token_accuracy": 0.8422728776931763, + "num_tokens": 74593252.0, + "step": 1917 + }, + { + "epoch": 0.2439893143365984, + "grad_norm": 2.6223433017730713, + "learning_rate": 8.126324713861805e-07, + "loss": 0.3493, + "mean_token_accuracy": 0.8770164251327515, + "num_tokens": 74634091.0, + "step": 1918 + }, + { + "epoch": 0.2441165246151889, + "grad_norm": 2.5472965240478516, + "learning_rate": 8.130563798219584e-07, + "loss": 0.3825, + "mean_token_accuracy": 0.8686429262161255, + "num_tokens": 74678029.0, + "step": 1919 + }, + { + "epoch": 0.24424373489377943, + "grad_norm": 2.5408012866973877, + "learning_rate": 8.134802882577363e-07, + "loss": 0.3568, + "mean_token_accuracy": 0.8766818046569824, + "num_tokens": 74721046.0, + "step": 1920 + }, + { + "epoch": 0.24437094517236993, + "grad_norm": 2.557577133178711, + "learning_rate": 8.139041966935142e-07, + "loss": 0.3618, + "mean_token_accuracy": 0.8735425472259521, + "num_tokens": 74761617.0, + "step": 1921 + }, + { + "epoch": 0.24449815545096043, + "grad_norm": 2.861938238143921, + "learning_rate": 8.14328105129292e-07, + "loss": 0.4402, + "mean_token_accuracy": 0.8539429903030396, + "num_tokens": 74800967.0, + "step": 1922 + }, + { + "epoch": 0.24462536572955096, + "grad_norm": 2.832426071166992, + "learning_rate": 8.1475201356507e-07, + "loss": 0.3717, + "mean_token_accuracy": 0.8748279809951782, + "num_tokens": 74840038.0, + "step": 1923 + }, + { + "epoch": 0.24475257600814146, + "grad_norm": 2.9641356468200684, + "learning_rate": 8.151759220008477e-07, + "loss": 0.4143, + "mean_token_accuracy": 0.8606150150299072, + "num_tokens": 74878541.0, + "step": 1924 + }, + { + "epoch": 0.24487978628673196, + "grad_norm": 2.768460988998413, + "learning_rate": 8.155998304366256e-07, + "loss": 0.3897, + "mean_token_accuracy": 0.8648632764816284, + "num_tokens": 74921289.0, + "step": 1925 + }, + { + "epoch": 0.2450069965653225, + "grad_norm": 2.5784361362457275, + "learning_rate": 8.160237388724035e-07, + "loss": 0.3606, + "mean_token_accuracy": 0.8767349720001221, + "num_tokens": 74963289.0, + "step": 1926 + }, + { + "epoch": 0.245134206843913, + "grad_norm": 2.711505651473999, + "learning_rate": 8.164476473081814e-07, + "loss": 0.3306, + "mean_token_accuracy": 0.8796488046646118, + "num_tokens": 75000244.0, + "step": 1927 + }, + { + "epoch": 0.2452614171225035, + "grad_norm": 2.944617986679077, + "learning_rate": 8.168715557439593e-07, + "loss": 0.3966, + "mean_token_accuracy": 0.867341160774231, + "num_tokens": 75036517.0, + "step": 1928 + }, + { + "epoch": 0.24538862740109402, + "grad_norm": 2.9910075664520264, + "learning_rate": 8.172954641797372e-07, + "loss": 0.4232, + "mean_token_accuracy": 0.8571149706840515, + "num_tokens": 75083530.0, + "step": 1929 + }, + { + "epoch": 0.24551583767968452, + "grad_norm": 2.769895076751709, + "learning_rate": 8.17719372615515e-07, + "loss": 0.3419, + "mean_token_accuracy": 0.8800500631332397, + "num_tokens": 75117817.0, + "step": 1930 + }, + { + "epoch": 0.24564304795827502, + "grad_norm": 2.7934842109680176, + "learning_rate": 8.18143281051293e-07, + "loss": 0.3934, + "mean_token_accuracy": 0.8659838438034058, + "num_tokens": 75157277.0, + "step": 1931 + }, + { + "epoch": 0.24577025823686555, + "grad_norm": 2.742497205734253, + "learning_rate": 8.185671894870707e-07, + "loss": 0.4034, + "mean_token_accuracy": 0.8643475770950317, + "num_tokens": 75199692.0, + "step": 1932 + }, + { + "epoch": 0.24589746851545605, + "grad_norm": 2.6083807945251465, + "learning_rate": 8.189910979228485e-07, + "loss": 0.3496, + "mean_token_accuracy": 0.8781763315200806, + "num_tokens": 75242416.0, + "step": 1933 + }, + { + "epoch": 0.24602467879404655, + "grad_norm": 2.805994749069214, + "learning_rate": 8.194150063586265e-07, + "loss": 0.3586, + "mean_token_accuracy": 0.8734875321388245, + "num_tokens": 75279858.0, + "step": 1934 + }, + { + "epoch": 0.24615188907263708, + "grad_norm": 3.0048394203186035, + "learning_rate": 8.198389147944043e-07, + "loss": 0.4182, + "mean_token_accuracy": 0.8593697547912598, + "num_tokens": 75318837.0, + "step": 1935 + }, + { + "epoch": 0.24627909935122758, + "grad_norm": 2.8835206031799316, + "learning_rate": 8.202628232301823e-07, + "loss": 0.3925, + "mean_token_accuracy": 0.8645670413970947, + "num_tokens": 75354371.0, + "step": 1936 + }, + { + "epoch": 0.24640630962981808, + "grad_norm": 2.7726857662200928, + "learning_rate": 8.206867316659601e-07, + "loss": 0.3971, + "mean_token_accuracy": 0.8629037141799927, + "num_tokens": 75396487.0, + "step": 1937 + }, + { + "epoch": 0.2465335199084086, + "grad_norm": 2.7790541648864746, + "learning_rate": 8.21110640101738e-07, + "loss": 0.4112, + "mean_token_accuracy": 0.8611136674880981, + "num_tokens": 75438250.0, + "step": 1938 + }, + { + "epoch": 0.2466607301869991, + "grad_norm": 2.654904365539551, + "learning_rate": 8.215345485375159e-07, + "loss": 0.3687, + "mean_token_accuracy": 0.8749738931655884, + "num_tokens": 75478720.0, + "step": 1939 + }, + { + "epoch": 0.2467879404655896, + "grad_norm": 2.702582359313965, + "learning_rate": 8.219584569732937e-07, + "loss": 0.3925, + "mean_token_accuracy": 0.8709431886672974, + "num_tokens": 75523432.0, + "step": 1940 + }, + { + "epoch": 0.24691515074418013, + "grad_norm": 2.9177050590515137, + "learning_rate": 8.223823654090715e-07, + "loss": 0.3828, + "mean_token_accuracy": 0.8677908778190613, + "num_tokens": 75558544.0, + "step": 1941 + }, + { + "epoch": 0.24704236102277063, + "grad_norm": 2.834207057952881, + "learning_rate": 8.228062738448495e-07, + "loss": 0.4117, + "mean_token_accuracy": 0.8611790537834167, + "num_tokens": 75600480.0, + "step": 1942 + }, + { + "epoch": 0.24716957130136116, + "grad_norm": 2.8460912704467773, + "learning_rate": 8.232301822806273e-07, + "loss": 0.4022, + "mean_token_accuracy": 0.8634393811225891, + "num_tokens": 75641239.0, + "step": 1943 + }, + { + "epoch": 0.24729678157995166, + "grad_norm": 2.851249933242798, + "learning_rate": 8.236540907164053e-07, + "loss": 0.3506, + "mean_token_accuracy": 0.8771956562995911, + "num_tokens": 75679524.0, + "step": 1944 + }, + { + "epoch": 0.24742399185854216, + "grad_norm": 2.605285882949829, + "learning_rate": 8.240779991521831e-07, + "loss": 0.3512, + "mean_token_accuracy": 0.8805195093154907, + "num_tokens": 75720861.0, + "step": 1945 + }, + { + "epoch": 0.2475512021371327, + "grad_norm": 2.7879042625427246, + "learning_rate": 8.24501907587961e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8594284057617188, + "num_tokens": 75763219.0, + "step": 1946 + }, + { + "epoch": 0.2476784124157232, + "grad_norm": 2.8482964038848877, + "learning_rate": 8.249258160237388e-07, + "loss": 0.3364, + "mean_token_accuracy": 0.884212851524353, + "num_tokens": 75796870.0, + "step": 1947 + }, + { + "epoch": 0.2478056226943137, + "grad_norm": 3.0398499965667725, + "learning_rate": 8.253497244595167e-07, + "loss": 0.4142, + "mean_token_accuracy": 0.8569568395614624, + "num_tokens": 75831198.0, + "step": 1948 + }, + { + "epoch": 0.24793283297290422, + "grad_norm": 2.885227918624878, + "learning_rate": 8.257736328952945e-07, + "loss": 0.4063, + "mean_token_accuracy": 0.8606394529342651, + "num_tokens": 75869109.0, + "step": 1949 + }, + { + "epoch": 0.24806004325149472, + "grad_norm": 2.718848466873169, + "learning_rate": 8.261975413310725e-07, + "loss": 0.3635, + "mean_token_accuracy": 0.8763401508331299, + "num_tokens": 75912271.0, + "step": 1950 + }, + { + "epoch": 0.24818725353008522, + "grad_norm": 3.1835150718688965, + "learning_rate": 8.266214497668503e-07, + "loss": 0.3781, + "mean_token_accuracy": 0.870934009552002, + "num_tokens": 75944136.0, + "step": 1951 + }, + { + "epoch": 0.24831446380867575, + "grad_norm": 2.781543493270874, + "learning_rate": 8.270453582026283e-07, + "loss": 0.396, + "mean_token_accuracy": 0.8641746044158936, + "num_tokens": 75982652.0, + "step": 1952 + }, + { + "epoch": 0.24844167408726625, + "grad_norm": 2.459155797958374, + "learning_rate": 8.274692666384061e-07, + "loss": 0.3721, + "mean_token_accuracy": 0.8709493279457092, + "num_tokens": 76028543.0, + "step": 1953 + }, + { + "epoch": 0.24856888436585675, + "grad_norm": 2.837693929672241, + "learning_rate": 8.27893175074184e-07, + "loss": 0.3926, + "mean_token_accuracy": 0.8685317039489746, + "num_tokens": 76069821.0, + "step": 1954 + }, + { + "epoch": 0.24869609464444728, + "grad_norm": 2.8643295764923096, + "learning_rate": 8.283170835099618e-07, + "loss": 0.4086, + "mean_token_accuracy": 0.8620285391807556, + "num_tokens": 76109050.0, + "step": 1955 + }, + { + "epoch": 0.24882330492303778, + "grad_norm": 2.745835065841675, + "learning_rate": 8.287409919457396e-07, + "loss": 0.4048, + "mean_token_accuracy": 0.8638485670089722, + "num_tokens": 76154603.0, + "step": 1956 + }, + { + "epoch": 0.24895051520162828, + "grad_norm": 2.614252805709839, + "learning_rate": 8.291649003815175e-07, + "loss": 0.3613, + "mean_token_accuracy": 0.8736894130706787, + "num_tokens": 76196335.0, + "step": 1957 + }, + { + "epoch": 0.2490777254802188, + "grad_norm": 2.993217706680298, + "learning_rate": 8.295888088172954e-07, + "loss": 0.4053, + "mean_token_accuracy": 0.8629827499389648, + "num_tokens": 76232929.0, + "step": 1958 + }, + { + "epoch": 0.2492049357588093, + "grad_norm": 2.8623673915863037, + "learning_rate": 8.300127172530733e-07, + "loss": 0.3866, + "mean_token_accuracy": 0.8640893697738647, + "num_tokens": 76272312.0, + "step": 1959 + }, + { + "epoch": 0.2493321460373998, + "grad_norm": 2.8938941955566406, + "learning_rate": 8.304366256888512e-07, + "loss": 0.3796, + "mean_token_accuracy": 0.8681848049163818, + "num_tokens": 76311615.0, + "step": 1960 + }, + { + "epoch": 0.24945935631599034, + "grad_norm": 2.8024532794952393, + "learning_rate": 8.308605341246291e-07, + "loss": 0.3878, + "mean_token_accuracy": 0.8663082122802734, + "num_tokens": 76350998.0, + "step": 1961 + }, + { + "epoch": 0.24958656659458084, + "grad_norm": 2.811701774597168, + "learning_rate": 8.312844425604068e-07, + "loss": 0.4103, + "mean_token_accuracy": 0.8590038418769836, + "num_tokens": 76392449.0, + "step": 1962 + }, + { + "epoch": 0.24971377687317134, + "grad_norm": 2.8246870040893555, + "learning_rate": 8.317083509961848e-07, + "loss": 0.3666, + "mean_token_accuracy": 0.8725686073303223, + "num_tokens": 76426285.0, + "step": 1963 + }, + { + "epoch": 0.24984098715176187, + "grad_norm": 2.6755106449127197, + "learning_rate": 8.321322594319626e-07, + "loss": 0.41, + "mean_token_accuracy": 0.8596105575561523, + "num_tokens": 76471823.0, + "step": 1964 + }, + { + "epoch": 0.24996819743035237, + "grad_norm": 2.6504967212677, + "learning_rate": 8.325561678677405e-07, + "loss": 0.3589, + "mean_token_accuracy": 0.8742411732673645, + "num_tokens": 76514863.0, + "step": 1965 + }, + { + "epoch": 0.2500954077089429, + "grad_norm": 2.8518495559692383, + "learning_rate": 8.329800763035184e-07, + "loss": 0.4229, + "mean_token_accuracy": 0.8572553992271423, + "num_tokens": 76555475.0, + "step": 1966 + }, + { + "epoch": 0.25022261798753337, + "grad_norm": 2.825714349746704, + "learning_rate": 8.334039847392963e-07, + "loss": 0.381, + "mean_token_accuracy": 0.8692147135734558, + "num_tokens": 76594663.0, + "step": 1967 + }, + { + "epoch": 0.2503498282661239, + "grad_norm": 2.999242067337036, + "learning_rate": 8.338278931750742e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8617984056472778, + "num_tokens": 76633214.0, + "step": 1968 + }, + { + "epoch": 0.2504770385447144, + "grad_norm": 3.0435824394226074, + "learning_rate": 8.342518016108521e-07, + "loss": 0.3921, + "mean_token_accuracy": 0.869027853012085, + "num_tokens": 76665598.0, + "step": 1969 + }, + { + "epoch": 0.2506042488233049, + "grad_norm": 2.8102030754089355, + "learning_rate": 8.346757100466298e-07, + "loss": 0.4149, + "mean_token_accuracy": 0.8610889911651611, + "num_tokens": 76704581.0, + "step": 1970 + }, + { + "epoch": 0.2507314591018954, + "grad_norm": 2.581123113632202, + "learning_rate": 8.350996184824078e-07, + "loss": 0.3716, + "mean_token_accuracy": 0.871427595615387, + "num_tokens": 76746742.0, + "step": 1971 + }, + { + "epoch": 0.25085866938048595, + "grad_norm": 2.9764649868011475, + "learning_rate": 8.355235269181856e-07, + "loss": 0.4067, + "mean_token_accuracy": 0.8607329726219177, + "num_tokens": 76780795.0, + "step": 1972 + }, + { + "epoch": 0.2509858796590764, + "grad_norm": 2.9013097286224365, + "learning_rate": 8.359474353539635e-07, + "loss": 0.3961, + "mean_token_accuracy": 0.8619641065597534, + "num_tokens": 76818991.0, + "step": 1973 + }, + { + "epoch": 0.25111308993766696, + "grad_norm": 3.189831256866455, + "learning_rate": 8.363713437897414e-07, + "loss": 0.397, + "mean_token_accuracy": 0.8628956079483032, + "num_tokens": 76849867.0, + "step": 1974 + }, + { + "epoch": 0.2512403002162575, + "grad_norm": 2.846440076828003, + "learning_rate": 8.367952522255193e-07, + "loss": 0.4046, + "mean_token_accuracy": 0.8630658984184265, + "num_tokens": 76891058.0, + "step": 1975 + }, + { + "epoch": 0.25136751049484796, + "grad_norm": 2.5946505069732666, + "learning_rate": 8.372191606612972e-07, + "loss": 0.37, + "mean_token_accuracy": 0.8729994893074036, + "num_tokens": 76933788.0, + "step": 1976 + }, + { + "epoch": 0.2514947207734385, + "grad_norm": 2.6623854637145996, + "learning_rate": 8.376430690970749e-07, + "loss": 0.3865, + "mean_token_accuracy": 0.86818528175354, + "num_tokens": 76973757.0, + "step": 1977 + }, + { + "epoch": 0.251621931052029, + "grad_norm": 2.6523046493530273, + "learning_rate": 8.380669775328528e-07, + "loss": 0.3712, + "mean_token_accuracy": 0.8707442283630371, + "num_tokens": 77014746.0, + "step": 1978 + }, + { + "epoch": 0.25174914133061954, + "grad_norm": 2.674192428588867, + "learning_rate": 8.384908859686307e-07, + "loss": 0.3633, + "mean_token_accuracy": 0.8743735551834106, + "num_tokens": 77055635.0, + "step": 1979 + }, + { + "epoch": 0.25187635160921, + "grad_norm": 2.842479705810547, + "learning_rate": 8.389147944044086e-07, + "loss": 0.4043, + "mean_token_accuracy": 0.8612582683563232, + "num_tokens": 77096550.0, + "step": 1980 + }, + { + "epoch": 0.25200356188780054, + "grad_norm": 2.9079768657684326, + "learning_rate": 8.393387028401864e-07, + "loss": 0.4274, + "mean_token_accuracy": 0.8561785817146301, + "num_tokens": 77135716.0, + "step": 1981 + }, + { + "epoch": 0.25213077216639107, + "grad_norm": 2.7765893936157227, + "learning_rate": 8.397626112759644e-07, + "loss": 0.3867, + "mean_token_accuracy": 0.8692731857299805, + "num_tokens": 77175494.0, + "step": 1982 + }, + { + "epoch": 0.25225798244498154, + "grad_norm": 2.8534014225006104, + "learning_rate": 8.401865197117422e-07, + "loss": 0.3522, + "mean_token_accuracy": 0.8795019388198853, + "num_tokens": 77211002.0, + "step": 1983 + }, + { + "epoch": 0.25238519272357207, + "grad_norm": 2.7217512130737305, + "learning_rate": 8.406104281475202e-07, + "loss": 0.3455, + "mean_token_accuracy": 0.8797668814659119, + "num_tokens": 77248694.0, + "step": 1984 + }, + { + "epoch": 0.2525124030021626, + "grad_norm": 2.8045434951782227, + "learning_rate": 8.410343365832979e-07, + "loss": 0.4063, + "mean_token_accuracy": 0.8609760999679565, + "num_tokens": 77291486.0, + "step": 1985 + }, + { + "epoch": 0.2526396132807531, + "grad_norm": 2.56463623046875, + "learning_rate": 8.414582450190758e-07, + "loss": 0.3614, + "mean_token_accuracy": 0.8753562569618225, + "num_tokens": 77339460.0, + "step": 1986 + }, + { + "epoch": 0.2527668235593436, + "grad_norm": 2.7835609912872314, + "learning_rate": 8.418821534548537e-07, + "loss": 0.3726, + "mean_token_accuracy": 0.8713055849075317, + "num_tokens": 77379549.0, + "step": 1987 + }, + { + "epoch": 0.25289403383793413, + "grad_norm": 2.758821487426758, + "learning_rate": 8.423060618906316e-07, + "loss": 0.3719, + "mean_token_accuracy": 0.8708251714706421, + "num_tokens": 77418321.0, + "step": 1988 + }, + { + "epoch": 0.2530212441165246, + "grad_norm": 3.08085560798645, + "learning_rate": 8.427299703264095e-07, + "loss": 0.3871, + "mean_token_accuracy": 0.8689168691635132, + "num_tokens": 77452615.0, + "step": 1989 + }, + { + "epoch": 0.25314845439511513, + "grad_norm": 2.8162448406219482, + "learning_rate": 8.431538787621874e-07, + "loss": 0.4001, + "mean_token_accuracy": 0.8630104064941406, + "num_tokens": 77492764.0, + "step": 1990 + }, + { + "epoch": 0.25327566467370566, + "grad_norm": 2.833956003189087, + "learning_rate": 8.435777871979652e-07, + "loss": 0.4123, + "mean_token_accuracy": 0.859032154083252, + "num_tokens": 77533013.0, + "step": 1991 + }, + { + "epoch": 0.25340287495229613, + "grad_norm": 2.855241537094116, + "learning_rate": 8.440016956337432e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8667932748794556, + "num_tokens": 77573572.0, + "step": 1992 + }, + { + "epoch": 0.25353008523088666, + "grad_norm": 2.9164416790008545, + "learning_rate": 8.444256040695209e-07, + "loss": 0.3658, + "mean_token_accuracy": 0.8722946643829346, + "num_tokens": 77613190.0, + "step": 1993 + }, + { + "epoch": 0.2536572955094772, + "grad_norm": 2.8598313331604004, + "learning_rate": 8.448495125052988e-07, + "loss": 0.4062, + "mean_token_accuracy": 0.8630771040916443, + "num_tokens": 77652446.0, + "step": 1994 + }, + { + "epoch": 0.25378450578806766, + "grad_norm": 2.6412813663482666, + "learning_rate": 8.452734209410767e-07, + "loss": 0.3873, + "mean_token_accuracy": 0.8664774894714355, + "num_tokens": 77693400.0, + "step": 1995 + }, + { + "epoch": 0.2539117160666582, + "grad_norm": 2.799030303955078, + "learning_rate": 8.456973293768545e-07, + "loss": 0.3461, + "mean_token_accuracy": 0.882064700126648, + "num_tokens": 77730929.0, + "step": 1996 + }, + { + "epoch": 0.2540389263452487, + "grad_norm": 2.689455986022949, + "learning_rate": 8.461212378126325e-07, + "loss": 0.3625, + "mean_token_accuracy": 0.8720536231994629, + "num_tokens": 77772150.0, + "step": 1997 + }, + { + "epoch": 0.2541661366238392, + "grad_norm": 2.952857255935669, + "learning_rate": 8.465451462484103e-07, + "loss": 0.3751, + "mean_token_accuracy": 0.8699911832809448, + "num_tokens": 77809317.0, + "step": 1998 + }, + { + "epoch": 0.2542933469024297, + "grad_norm": 2.856563091278076, + "learning_rate": 8.469690546841882e-07, + "loss": 0.3733, + "mean_token_accuracy": 0.8724005222320557, + "num_tokens": 77848626.0, + "step": 1999 + }, + { + "epoch": 0.25442055718102025, + "grad_norm": 2.9606103897094727, + "learning_rate": 8.47392963119966e-07, + "loss": 0.4351, + "mean_token_accuracy": 0.8542348146438599, + "num_tokens": 77888077.0, + "step": 2000 + }, + { + "epoch": 0.2545477674596107, + "grad_norm": 2.796250104904175, + "learning_rate": 8.478168715557439e-07, + "loss": 0.3572, + "mean_token_accuracy": 0.8783526420593262, + "num_tokens": 77922507.0, + "step": 2001 + }, + { + "epoch": 0.25467497773820125, + "grad_norm": 2.79866361618042, + "learning_rate": 8.482407799915217e-07, + "loss": 0.413, + "mean_token_accuracy": 0.8582611083984375, + "num_tokens": 77961508.0, + "step": 2002 + }, + { + "epoch": 0.2548021880167918, + "grad_norm": 2.9287328720092773, + "learning_rate": 8.486646884272997e-07, + "loss": 0.3604, + "mean_token_accuracy": 0.8776934146881104, + "num_tokens": 77997774.0, + "step": 2003 + }, + { + "epoch": 0.25492939829538225, + "grad_norm": 2.6403627395629883, + "learning_rate": 8.490885968630775e-07, + "loss": 0.3923, + "mean_token_accuracy": 0.8670674562454224, + "num_tokens": 78041998.0, + "step": 2004 + }, + { + "epoch": 0.2550566085739728, + "grad_norm": 2.8348546028137207, + "learning_rate": 8.495125052988555e-07, + "loss": 0.4204, + "mean_token_accuracy": 0.8592039346694946, + "num_tokens": 78080820.0, + "step": 2005 + }, + { + "epoch": 0.2551838188525633, + "grad_norm": 2.821854829788208, + "learning_rate": 8.499364137346333e-07, + "loss": 0.3869, + "mean_token_accuracy": 0.8664181232452393, + "num_tokens": 78118112.0, + "step": 2006 + }, + { + "epoch": 0.2553110291311538, + "grad_norm": 2.5473570823669434, + "learning_rate": 8.503603221704112e-07, + "loss": 0.3592, + "mean_token_accuracy": 0.8730501532554626, + "num_tokens": 78160004.0, + "step": 2007 + }, + { + "epoch": 0.2554382394097443, + "grad_norm": 2.5416812896728516, + "learning_rate": 8.50784230606189e-07, + "loss": 0.3509, + "mean_token_accuracy": 0.8775013089179993, + "num_tokens": 78203192.0, + "step": 2008 + }, + { + "epoch": 0.25556544968833483, + "grad_norm": 2.774766683578491, + "learning_rate": 8.512081390419669e-07, + "loss": 0.3979, + "mean_token_accuracy": 0.8632523417472839, + "num_tokens": 78242881.0, + "step": 2009 + }, + { + "epoch": 0.2556926599669253, + "grad_norm": 2.872434139251709, + "learning_rate": 8.516320474777447e-07, + "loss": 0.4003, + "mean_token_accuracy": 0.8613821268081665, + "num_tokens": 78282267.0, + "step": 2010 + }, + { + "epoch": 0.25581987024551583, + "grad_norm": 2.5989301204681396, + "learning_rate": 8.520559559135227e-07, + "loss": 0.3717, + "mean_token_accuracy": 0.8728250861167908, + "num_tokens": 78327168.0, + "step": 2011 + }, + { + "epoch": 0.25594708052410636, + "grad_norm": 2.663383960723877, + "learning_rate": 8.524798643493005e-07, + "loss": 0.4228, + "mean_token_accuracy": 0.8558956980705261, + "num_tokens": 78370911.0, + "step": 2012 + }, + { + "epoch": 0.25607429080269684, + "grad_norm": 2.9621965885162354, + "learning_rate": 8.529037727850785e-07, + "loss": 0.4078, + "mean_token_accuracy": 0.8665370941162109, + "num_tokens": 78408324.0, + "step": 2013 + }, + { + "epoch": 0.25620150108128736, + "grad_norm": 2.866762161254883, + "learning_rate": 8.533276812208563e-07, + "loss": 0.3931, + "mean_token_accuracy": 0.8633432984352112, + "num_tokens": 78449396.0, + "step": 2014 + }, + { + "epoch": 0.2563287113598779, + "grad_norm": 2.7415058612823486, + "learning_rate": 8.53751589656634e-07, + "loss": 0.3575, + "mean_token_accuracy": 0.875433087348938, + "num_tokens": 78485085.0, + "step": 2015 + }, + { + "epoch": 0.25645592163846836, + "grad_norm": 2.294630765914917, + "learning_rate": 8.54175498092412e-07, + "loss": 0.3192, + "mean_token_accuracy": 0.8893948197364807, + "num_tokens": 78530789.0, + "step": 2016 + }, + { + "epoch": 0.2565831319170589, + "grad_norm": 2.6470229625701904, + "learning_rate": 8.545994065281898e-07, + "loss": 0.3593, + "mean_token_accuracy": 0.8766627311706543, + "num_tokens": 78571172.0, + "step": 2017 + }, + { + "epoch": 0.2567103421956494, + "grad_norm": 2.91422176361084, + "learning_rate": 8.550233149639677e-07, + "loss": 0.4071, + "mean_token_accuracy": 0.8584271669387817, + "num_tokens": 78606746.0, + "step": 2018 + }, + { + "epoch": 0.2568375524742399, + "grad_norm": 2.96437931060791, + "learning_rate": 8.554472233997456e-07, + "loss": 0.3786, + "mean_token_accuracy": 0.8695470094680786, + "num_tokens": 78644765.0, + "step": 2019 + }, + { + "epoch": 0.2569647627528304, + "grad_norm": 2.8340587615966797, + "learning_rate": 8.558711318355235e-07, + "loss": 0.3792, + "mean_token_accuracy": 0.8677006363868713, + "num_tokens": 78683039.0, + "step": 2020 + }, + { + "epoch": 0.25709197303142095, + "grad_norm": 3.0208706855773926, + "learning_rate": 8.562950402713014e-07, + "loss": 0.378, + "mean_token_accuracy": 0.8734067678451538, + "num_tokens": 78716250.0, + "step": 2021 + }, + { + "epoch": 0.2572191833100114, + "grad_norm": 2.6324586868286133, + "learning_rate": 8.567189487070793e-07, + "loss": 0.3486, + "mean_token_accuracy": 0.8781471252441406, + "num_tokens": 78755675.0, + "step": 2022 + }, + { + "epoch": 0.25734639358860195, + "grad_norm": 2.644721746444702, + "learning_rate": 8.57142857142857e-07, + "loss": 0.4156, + "mean_token_accuracy": 0.8597230911254883, + "num_tokens": 78802037.0, + "step": 2023 + }, + { + "epoch": 0.2574736038671925, + "grad_norm": 2.586787700653076, + "learning_rate": 8.57566765578635e-07, + "loss": 0.4002, + "mean_token_accuracy": 0.8641095161437988, + "num_tokens": 78846095.0, + "step": 2024 + }, + { + "epoch": 0.25760081414578295, + "grad_norm": 2.5398991107940674, + "learning_rate": 8.579906740144128e-07, + "loss": 0.3292, + "mean_token_accuracy": 0.880527913570404, + "num_tokens": 78885119.0, + "step": 2025 + }, + { + "epoch": 0.2577280244243735, + "grad_norm": 2.719867706298828, + "learning_rate": 8.584145824501907e-07, + "loss": 0.3459, + "mean_token_accuracy": 0.8772707581520081, + "num_tokens": 78922094.0, + "step": 2026 + }, + { + "epoch": 0.257855234702964, + "grad_norm": 2.960968017578125, + "learning_rate": 8.588384908859686e-07, + "loss": 0.3905, + "mean_token_accuracy": 0.8652055263519287, + "num_tokens": 78963478.0, + "step": 2027 + }, + { + "epoch": 0.25798244498155454, + "grad_norm": 3.1789469718933105, + "learning_rate": 8.592623993217465e-07, + "loss": 0.4271, + "mean_token_accuracy": 0.8550801277160645, + "num_tokens": 79000418.0, + "step": 2028 + }, + { + "epoch": 0.258109655260145, + "grad_norm": 2.832589626312256, + "learning_rate": 8.596863077575244e-07, + "loss": 0.3981, + "mean_token_accuracy": 0.8676002025604248, + "num_tokens": 79038198.0, + "step": 2029 + }, + { + "epoch": 0.25823686553873554, + "grad_norm": 2.8157479763031006, + "learning_rate": 8.601102161933023e-07, + "loss": 0.4436, + "mean_token_accuracy": 0.8526480793952942, + "num_tokens": 79083016.0, + "step": 2030 + }, + { + "epoch": 0.25836407581732607, + "grad_norm": 2.971334934234619, + "learning_rate": 8.6053412462908e-07, + "loss": 0.413, + "mean_token_accuracy": 0.8602681756019592, + "num_tokens": 79118712.0, + "step": 2031 + }, + { + "epoch": 0.25849128609591654, + "grad_norm": 2.6493136882781982, + "learning_rate": 8.60958033064858e-07, + "loss": 0.4011, + "mean_token_accuracy": 0.8647481799125671, + "num_tokens": 79162334.0, + "step": 2032 + }, + { + "epoch": 0.25861849637450707, + "grad_norm": 2.696855068206787, + "learning_rate": 8.613819415006358e-07, + "loss": 0.3511, + "mean_token_accuracy": 0.8792988657951355, + "num_tokens": 79202242.0, + "step": 2033 + }, + { + "epoch": 0.2587457066530976, + "grad_norm": 3.0391347408294678, + "learning_rate": 8.618058499364137e-07, + "loss": 0.384, + "mean_token_accuracy": 0.8681479692459106, + "num_tokens": 79241836.0, + "step": 2034 + }, + { + "epoch": 0.25887291693168807, + "grad_norm": 3.1943655014038086, + "learning_rate": 8.622297583721916e-07, + "loss": 0.4045, + "mean_token_accuracy": 0.8598902225494385, + "num_tokens": 79274294.0, + "step": 2035 + }, + { + "epoch": 0.2590001272102786, + "grad_norm": 2.734198808670044, + "learning_rate": 8.626536668079695e-07, + "loss": 0.3532, + "mean_token_accuracy": 0.8767634034156799, + "num_tokens": 79315797.0, + "step": 2036 + }, + { + "epoch": 0.2591273374888691, + "grad_norm": 2.9513635635375977, + "learning_rate": 8.630775752437474e-07, + "loss": 0.3903, + "mean_token_accuracy": 0.8674859404563904, + "num_tokens": 79353591.0, + "step": 2037 + }, + { + "epoch": 0.2592545477674596, + "grad_norm": 2.737537384033203, + "learning_rate": 8.635014836795251e-07, + "loss": 0.3912, + "mean_token_accuracy": 0.8657433390617371, + "num_tokens": 79395757.0, + "step": 2038 + }, + { + "epoch": 0.2593817580460501, + "grad_norm": 2.738978624343872, + "learning_rate": 8.63925392115303e-07, + "loss": 0.3782, + "mean_token_accuracy": 0.8730517029762268, + "num_tokens": 79435370.0, + "step": 2039 + }, + { + "epoch": 0.25950896832464065, + "grad_norm": 2.622053623199463, + "learning_rate": 8.643493005510809e-07, + "loss": 0.3355, + "mean_token_accuracy": 0.8793830871582031, + "num_tokens": 79471849.0, + "step": 2040 + }, + { + "epoch": 0.2596361786032311, + "grad_norm": 2.955509662628174, + "learning_rate": 8.647732089868588e-07, + "loss": 0.4392, + "mean_token_accuracy": 0.8555036783218384, + "num_tokens": 79513538.0, + "step": 2041 + }, + { + "epoch": 0.25976338888182166, + "grad_norm": 2.6582889556884766, + "learning_rate": 8.651971174226366e-07, + "loss": 0.3761, + "mean_token_accuracy": 0.8752036094665527, + "num_tokens": 79555187.0, + "step": 2042 + }, + { + "epoch": 0.2598905991604122, + "grad_norm": 2.940049409866333, + "learning_rate": 8.656210258584146e-07, + "loss": 0.3665, + "mean_token_accuracy": 0.874488115310669, + "num_tokens": 79590529.0, + "step": 2043 + }, + { + "epoch": 0.26001780943900266, + "grad_norm": 3.1039958000183105, + "learning_rate": 8.660449342941924e-07, + "loss": 0.381, + "mean_token_accuracy": 0.8662409782409668, + "num_tokens": 79622697.0, + "step": 2044 + }, + { + "epoch": 0.2601450197175932, + "grad_norm": 2.8878610134124756, + "learning_rate": 8.664688427299704e-07, + "loss": 0.3712, + "mean_token_accuracy": 0.872780442237854, + "num_tokens": 79662996.0, + "step": 2045 + }, + { + "epoch": 0.2602722299961837, + "grad_norm": 2.7286956310272217, + "learning_rate": 8.668927511657481e-07, + "loss": 0.3782, + "mean_token_accuracy": 0.8685981035232544, + "num_tokens": 79703572.0, + "step": 2046 + }, + { + "epoch": 0.2603994402747742, + "grad_norm": 2.6533613204956055, + "learning_rate": 8.67316659601526e-07, + "loss": 0.3698, + "mean_token_accuracy": 0.8686768412590027, + "num_tokens": 79742050.0, + "step": 2047 + }, + { + "epoch": 0.2605266505533647, + "grad_norm": 2.6818671226501465, + "learning_rate": 8.677405680373039e-07, + "loss": 0.3565, + "mean_token_accuracy": 0.8771766424179077, + "num_tokens": 79779365.0, + "step": 2048 + }, + { + "epoch": 0.26065386083195524, + "grad_norm": 2.9274561405181885, + "learning_rate": 8.681644764730818e-07, + "loss": 0.3955, + "mean_token_accuracy": 0.8669350147247314, + "num_tokens": 79818479.0, + "step": 2049 + }, + { + "epoch": 0.2607810711105457, + "grad_norm": 2.743727207183838, + "learning_rate": 8.685883849088596e-07, + "loss": 0.3688, + "mean_token_accuracy": 0.8704466819763184, + "num_tokens": 79859038.0, + "step": 2050 + }, + { + "epoch": 0.26090828138913624, + "grad_norm": 3.039748191833496, + "learning_rate": 8.690122933446376e-07, + "loss": 0.3582, + "mean_token_accuracy": 0.8735067844390869, + "num_tokens": 79892756.0, + "step": 2051 + }, + { + "epoch": 0.26103549166772677, + "grad_norm": 2.75395131111145, + "learning_rate": 8.694362017804154e-07, + "loss": 0.3745, + "mean_token_accuracy": 0.8730296492576599, + "num_tokens": 79929768.0, + "step": 2052 + }, + { + "epoch": 0.26116270194631724, + "grad_norm": 2.683089256286621, + "learning_rate": 8.698601102161933e-07, + "loss": 0.3567, + "mean_token_accuracy": 0.8749446868896484, + "num_tokens": 79966180.0, + "step": 2053 + }, + { + "epoch": 0.26128991222490777, + "grad_norm": 2.8103857040405273, + "learning_rate": 8.702840186519711e-07, + "loss": 0.3411, + "mean_token_accuracy": 0.8824130296707153, + "num_tokens": 80004519.0, + "step": 2054 + }, + { + "epoch": 0.2614171225034983, + "grad_norm": 3.012829065322876, + "learning_rate": 8.70707927087749e-07, + "loss": 0.4025, + "mean_token_accuracy": 0.8604378700256348, + "num_tokens": 80039552.0, + "step": 2055 + }, + { + "epoch": 0.2615443327820888, + "grad_norm": 2.860438585281372, + "learning_rate": 8.711318355235269e-07, + "loss": 0.35, + "mean_token_accuracy": 0.8796157836914062, + "num_tokens": 80075701.0, + "step": 2056 + }, + { + "epoch": 0.2616715430606793, + "grad_norm": 3.0046863555908203, + "learning_rate": 8.715557439593047e-07, + "loss": 0.3603, + "mean_token_accuracy": 0.8718186020851135, + "num_tokens": 80108178.0, + "step": 2057 + }, + { + "epoch": 0.26179875333926983, + "grad_norm": 2.615196466445923, + "learning_rate": 8.719796523950826e-07, + "loss": 0.3326, + "mean_token_accuracy": 0.8835597038269043, + "num_tokens": 80148695.0, + "step": 2058 + }, + { + "epoch": 0.2619259636178603, + "grad_norm": 2.811826467514038, + "learning_rate": 8.724035608308605e-07, + "loss": 0.3708, + "mean_token_accuracy": 0.8690648674964905, + "num_tokens": 80185981.0, + "step": 2059 + }, + { + "epoch": 0.26205317389645083, + "grad_norm": 2.8308303356170654, + "learning_rate": 8.728274692666384e-07, + "loss": 0.4007, + "mean_token_accuracy": 0.8629072904586792, + "num_tokens": 80228107.0, + "step": 2060 + }, + { + "epoch": 0.26218038417504136, + "grad_norm": 3.0637946128845215, + "learning_rate": 8.732513777024162e-07, + "loss": 0.3318, + "mean_token_accuracy": 0.8819433450698853, + "num_tokens": 80258597.0, + "step": 2061 + }, + { + "epoch": 0.26230759445363183, + "grad_norm": 2.9282424449920654, + "learning_rate": 8.736752861381941e-07, + "loss": 0.3728, + "mean_token_accuracy": 0.8721215724945068, + "num_tokens": 80293724.0, + "step": 2062 + }, + { + "epoch": 0.26243480473222236, + "grad_norm": 2.315805673599243, + "learning_rate": 8.740991945739719e-07, + "loss": 0.3198, + "mean_token_accuracy": 0.8895423412322998, + "num_tokens": 80339761.0, + "step": 2063 + }, + { + "epoch": 0.2625620150108129, + "grad_norm": 2.784471273422241, + "learning_rate": 8.745231030097499e-07, + "loss": 0.3566, + "mean_token_accuracy": 0.8770672082901001, + "num_tokens": 80377339.0, + "step": 2064 + }, + { + "epoch": 0.26268922528940336, + "grad_norm": 2.7173783779144287, + "learning_rate": 8.749470114455277e-07, + "loss": 0.3499, + "mean_token_accuracy": 0.8769596219062805, + "num_tokens": 80413831.0, + "step": 2065 + }, + { + "epoch": 0.2628164355679939, + "grad_norm": 2.874335289001465, + "learning_rate": 8.753709198813056e-07, + "loss": 0.4154, + "mean_token_accuracy": 0.8587141036987305, + "num_tokens": 80457507.0, + "step": 2066 + }, + { + "epoch": 0.2629436458465844, + "grad_norm": 2.7226338386535645, + "learning_rate": 8.757948283170835e-07, + "loss": 0.387, + "mean_token_accuracy": 0.8682845830917358, + "num_tokens": 80499606.0, + "step": 2067 + }, + { + "epoch": 0.2630708561251749, + "grad_norm": 2.651334524154663, + "learning_rate": 8.762187367528613e-07, + "loss": 0.3818, + "mean_token_accuracy": 0.8705698251724243, + "num_tokens": 80540436.0, + "step": 2068 + }, + { + "epoch": 0.2631980664037654, + "grad_norm": 2.889064311981201, + "learning_rate": 8.766426451886392e-07, + "loss": 0.3432, + "mean_token_accuracy": 0.8809942007064819, + "num_tokens": 80572813.0, + "step": 2069 + }, + { + "epoch": 0.26332527668235595, + "grad_norm": 2.977776288986206, + "learning_rate": 8.770665536244171e-07, + "loss": 0.3898, + "mean_token_accuracy": 0.8669899106025696, + "num_tokens": 80606411.0, + "step": 2070 + }, + { + "epoch": 0.2634524869609464, + "grad_norm": 2.916684865951538, + "learning_rate": 8.774904620601949e-07, + "loss": 0.3604, + "mean_token_accuracy": 0.8725727796554565, + "num_tokens": 80641881.0, + "step": 2071 + }, + { + "epoch": 0.26357969723953695, + "grad_norm": 2.6238250732421875, + "learning_rate": 8.779143704959729e-07, + "loss": 0.3478, + "mean_token_accuracy": 0.8791371583938599, + "num_tokens": 80679884.0, + "step": 2072 + }, + { + "epoch": 0.2637069075181275, + "grad_norm": 2.754915714263916, + "learning_rate": 8.783382789317507e-07, + "loss": 0.4337, + "mean_token_accuracy": 0.8554898500442505, + "num_tokens": 80725440.0, + "step": 2073 + }, + { + "epoch": 0.26383411779671795, + "grad_norm": 2.897850513458252, + "learning_rate": 8.787621873675286e-07, + "loss": 0.3755, + "mean_token_accuracy": 0.869105875492096, + "num_tokens": 80761743.0, + "step": 2074 + }, + { + "epoch": 0.2639613280753085, + "grad_norm": 3.132392168045044, + "learning_rate": 8.791860958033065e-07, + "loss": 0.3745, + "mean_token_accuracy": 0.8698011636734009, + "num_tokens": 80794250.0, + "step": 2075 + }, + { + "epoch": 0.264088538353899, + "grad_norm": 2.6350269317626953, + "learning_rate": 8.796100042390842e-07, + "loss": 0.4188, + "mean_token_accuracy": 0.857957124710083, + "num_tokens": 80839303.0, + "step": 2076 + }, + { + "epoch": 0.2642157486324895, + "grad_norm": 2.9121692180633545, + "learning_rate": 8.800339126748622e-07, + "loss": 0.3932, + "mean_token_accuracy": 0.86431485414505, + "num_tokens": 80877245.0, + "step": 2077 + }, + { + "epoch": 0.26434295891108, + "grad_norm": 3.0185210704803467, + "learning_rate": 8.8045782111064e-07, + "loss": 0.433, + "mean_token_accuracy": 0.8524705171585083, + "num_tokens": 80918583.0, + "step": 2078 + }, + { + "epoch": 0.26447016918967053, + "grad_norm": 2.7893500328063965, + "learning_rate": 8.808817295464179e-07, + "loss": 0.4344, + "mean_token_accuracy": 0.8536980152130127, + "num_tokens": 80963837.0, + "step": 2079 + }, + { + "epoch": 0.26459737946826106, + "grad_norm": 2.8926479816436768, + "learning_rate": 8.813056379821958e-07, + "loss": 0.3464, + "mean_token_accuracy": 0.8764057159423828, + "num_tokens": 81001324.0, + "step": 2080 + }, + { + "epoch": 0.26472458974685154, + "grad_norm": 2.937933921813965, + "learning_rate": 8.817295464179737e-07, + "loss": 0.3922, + "mean_token_accuracy": 0.865559458732605, + "num_tokens": 81040936.0, + "step": 2081 + }, + { + "epoch": 0.26485180002544206, + "grad_norm": 2.665858507156372, + "learning_rate": 8.821534548537515e-07, + "loss": 0.3786, + "mean_token_accuracy": 0.8694182634353638, + "num_tokens": 81085090.0, + "step": 2082 + }, + { + "epoch": 0.2649790103040326, + "grad_norm": 2.778468370437622, + "learning_rate": 8.825773632895295e-07, + "loss": 0.4194, + "mean_token_accuracy": 0.8575786352157593, + "num_tokens": 81127869.0, + "step": 2083 + }, + { + "epoch": 0.26510622058262306, + "grad_norm": 2.568136692047119, + "learning_rate": 8.830012717253072e-07, + "loss": 0.3416, + "mean_token_accuracy": 0.8792167901992798, + "num_tokens": 81170634.0, + "step": 2084 + }, + { + "epoch": 0.2652334308612136, + "grad_norm": 2.8631153106689453, + "learning_rate": 8.834251801610852e-07, + "loss": 0.3975, + "mean_token_accuracy": 0.8646731972694397, + "num_tokens": 81206848.0, + "step": 2085 + }, + { + "epoch": 0.2653606411398041, + "grad_norm": 2.9157044887542725, + "learning_rate": 8.83849088596863e-07, + "loss": 0.3817, + "mean_token_accuracy": 0.8695815801620483, + "num_tokens": 81244236.0, + "step": 2086 + }, + { + "epoch": 0.2654878514183946, + "grad_norm": 2.8109424114227295, + "learning_rate": 8.842729970326409e-07, + "loss": 0.3932, + "mean_token_accuracy": 0.8677513003349304, + "num_tokens": 81281661.0, + "step": 2087 + }, + { + "epoch": 0.2656150616969851, + "grad_norm": 2.829383134841919, + "learning_rate": 8.846969054684188e-07, + "loss": 0.3742, + "mean_token_accuracy": 0.8702150583267212, + "num_tokens": 81317940.0, + "step": 2088 + }, + { + "epoch": 0.26574227197557565, + "grad_norm": 2.8350305557250977, + "learning_rate": 8.851208139041967e-07, + "loss": 0.3276, + "mean_token_accuracy": 0.8843860626220703, + "num_tokens": 81354462.0, + "step": 2089 + }, + { + "epoch": 0.2658694822541661, + "grad_norm": 2.7996926307678223, + "learning_rate": 8.855447223399745e-07, + "loss": 0.3797, + "mean_token_accuracy": 0.8688466548919678, + "num_tokens": 81392755.0, + "step": 2090 + }, + { + "epoch": 0.26599669253275665, + "grad_norm": 2.7644734382629395, + "learning_rate": 8.859686307757524e-07, + "loss": 0.3873, + "mean_token_accuracy": 0.8674077987670898, + "num_tokens": 81437328.0, + "step": 2091 + }, + { + "epoch": 0.2661239028113472, + "grad_norm": 2.674178123474121, + "learning_rate": 8.863925392115302e-07, + "loss": 0.3448, + "mean_token_accuracy": 0.8796916007995605, + "num_tokens": 81477736.0, + "step": 2092 + }, + { + "epoch": 0.26625111308993765, + "grad_norm": 2.838210105895996, + "learning_rate": 8.868164476473082e-07, + "loss": 0.3884, + "mean_token_accuracy": 0.8687949180603027, + "num_tokens": 81514915.0, + "step": 2093 + }, + { + "epoch": 0.2663783233685282, + "grad_norm": 3.0919861793518066, + "learning_rate": 8.87240356083086e-07, + "loss": 0.3634, + "mean_token_accuracy": 0.8762456178665161, + "num_tokens": 81555272.0, + "step": 2094 + }, + { + "epoch": 0.2665055336471187, + "grad_norm": 2.938711404800415, + "learning_rate": 8.876642645188639e-07, + "loss": 0.363, + "mean_token_accuracy": 0.8758647441864014, + "num_tokens": 81591663.0, + "step": 2095 + }, + { + "epoch": 0.2666327439257092, + "grad_norm": 2.8784854412078857, + "learning_rate": 8.880881729546418e-07, + "loss": 0.3809, + "mean_token_accuracy": 0.8666239380836487, + "num_tokens": 81625789.0, + "step": 2096 + }, + { + "epoch": 0.2667599542042997, + "grad_norm": 2.7716031074523926, + "learning_rate": 8.885120813904197e-07, + "loss": 0.3851, + "mean_token_accuracy": 0.867830216884613, + "num_tokens": 81667831.0, + "step": 2097 + }, + { + "epoch": 0.26688716448289024, + "grad_norm": 2.7263007164001465, + "learning_rate": 8.889359898261976e-07, + "loss": 0.401, + "mean_token_accuracy": 0.8640309572219849, + "num_tokens": 81710333.0, + "step": 2098 + }, + { + "epoch": 0.2670143747614807, + "grad_norm": 2.7236266136169434, + "learning_rate": 8.893598982619753e-07, + "loss": 0.3725, + "mean_token_accuracy": 0.8699980974197388, + "num_tokens": 81748386.0, + "step": 2099 + }, + { + "epoch": 0.26714158504007124, + "grad_norm": 2.8319828510284424, + "learning_rate": 8.897838066977532e-07, + "loss": 0.3663, + "mean_token_accuracy": 0.8749226331710815, + "num_tokens": 81781589.0, + "step": 2100 + }, + { + "epoch": 0.26726879531866177, + "grad_norm": 2.955942153930664, + "learning_rate": 8.902077151335311e-07, + "loss": 0.3983, + "mean_token_accuracy": 0.8613096475601196, + "num_tokens": 81817967.0, + "step": 2101 + }, + { + "epoch": 0.26739600559725224, + "grad_norm": 2.4676997661590576, + "learning_rate": 8.90631623569309e-07, + "loss": 0.3293, + "mean_token_accuracy": 0.8867565989494324, + "num_tokens": 81860211.0, + "step": 2102 + }, + { + "epoch": 0.26752321587584277, + "grad_norm": 2.6932497024536133, + "learning_rate": 8.910555320050868e-07, + "loss": 0.3295, + "mean_token_accuracy": 0.8823249340057373, + "num_tokens": 81899292.0, + "step": 2103 + }, + { + "epoch": 0.2676504261544333, + "grad_norm": 2.6292288303375244, + "learning_rate": 8.914794404408648e-07, + "loss": 0.3423, + "mean_token_accuracy": 0.8814494609832764, + "num_tokens": 81935423.0, + "step": 2104 + }, + { + "epoch": 0.26777763643302377, + "grad_norm": 3.0060558319091797, + "learning_rate": 8.919033488766426e-07, + "loss": 0.416, + "mean_token_accuracy": 0.860608696937561, + "num_tokens": 81970151.0, + "step": 2105 + }, + { + "epoch": 0.2679048467116143, + "grad_norm": 2.6658968925476074, + "learning_rate": 8.923272573124204e-07, + "loss": 0.3591, + "mean_token_accuracy": 0.8771024346351624, + "num_tokens": 82012261.0, + "step": 2106 + }, + { + "epoch": 0.2680320569902048, + "grad_norm": 2.7478041648864746, + "learning_rate": 8.927511657481983e-07, + "loss": 0.3953, + "mean_token_accuracy": 0.8655158877372742, + "num_tokens": 82051413.0, + "step": 2107 + }, + { + "epoch": 0.2681592672687953, + "grad_norm": 2.6929755210876465, + "learning_rate": 8.931750741839762e-07, + "loss": 0.3457, + "mean_token_accuracy": 0.8806098699569702, + "num_tokens": 82091651.0, + "step": 2108 + }, + { + "epoch": 0.2682864775473858, + "grad_norm": 2.663073778152466, + "learning_rate": 8.935989826197541e-07, + "loss": 0.4032, + "mean_token_accuracy": 0.862045407295227, + "num_tokens": 82136692.0, + "step": 2109 + }, + { + "epoch": 0.26841368782597635, + "grad_norm": 2.7282707691192627, + "learning_rate": 8.94022891055532e-07, + "loss": 0.4439, + "mean_token_accuracy": 0.849873423576355, + "num_tokens": 82183030.0, + "step": 2110 + }, + { + "epoch": 0.2685408981045668, + "grad_norm": 2.6828267574310303, + "learning_rate": 8.944467994913098e-07, + "loss": 0.3465, + "mean_token_accuracy": 0.8827779293060303, + "num_tokens": 82225552.0, + "step": 2111 + }, + { + "epoch": 0.26866810838315736, + "grad_norm": 2.535750150680542, + "learning_rate": 8.948707079270878e-07, + "loss": 0.3539, + "mean_token_accuracy": 0.8794369697570801, + "num_tokens": 82266088.0, + "step": 2112 + }, + { + "epoch": 0.2687953186617479, + "grad_norm": 2.750142812728882, + "learning_rate": 8.952946163628656e-07, + "loss": 0.3637, + "mean_token_accuracy": 0.8738253116607666, + "num_tokens": 82303391.0, + "step": 2113 + }, + { + "epoch": 0.26892252894033836, + "grad_norm": 2.6467623710632324, + "learning_rate": 8.957185247986434e-07, + "loss": 0.3809, + "mean_token_accuracy": 0.867538571357727, + "num_tokens": 82346744.0, + "step": 2114 + }, + { + "epoch": 0.2690497392189289, + "grad_norm": 2.708775281906128, + "learning_rate": 8.961424332344213e-07, + "loss": 0.3527, + "mean_token_accuracy": 0.8786988258361816, + "num_tokens": 82385064.0, + "step": 2115 + }, + { + "epoch": 0.2691769494975194, + "grad_norm": 2.8618242740631104, + "learning_rate": 8.965663416701992e-07, + "loss": 0.3402, + "mean_token_accuracy": 0.882393479347229, + "num_tokens": 82418538.0, + "step": 2116 + }, + { + "epoch": 0.2693041597761099, + "grad_norm": 2.5618560314178467, + "learning_rate": 8.969902501059771e-07, + "loss": 0.3822, + "mean_token_accuracy": 0.8711058497428894, + "num_tokens": 82463231.0, + "step": 2117 + }, + { + "epoch": 0.2694313700547004, + "grad_norm": 2.7734084129333496, + "learning_rate": 8.97414158541755e-07, + "loss": 0.3856, + "mean_token_accuracy": 0.8679318428039551, + "num_tokens": 82502791.0, + "step": 2118 + }, + { + "epoch": 0.26955858033329094, + "grad_norm": 2.686786413192749, + "learning_rate": 8.978380669775328e-07, + "loss": 0.3852, + "mean_token_accuracy": 0.8686906099319458, + "num_tokens": 82545462.0, + "step": 2119 + }, + { + "epoch": 0.2696857906118814, + "grad_norm": 2.761702299118042, + "learning_rate": 8.982619754133107e-07, + "loss": 0.41, + "mean_token_accuracy": 0.8617647886276245, + "num_tokens": 82587263.0, + "step": 2120 + }, + { + "epoch": 0.26981300089047194, + "grad_norm": 2.584014415740967, + "learning_rate": 8.986858838490886e-07, + "loss": 0.325, + "mean_token_accuracy": 0.8861125707626343, + "num_tokens": 82627173.0, + "step": 2121 + }, + { + "epoch": 0.26994021116906247, + "grad_norm": 2.802504777908325, + "learning_rate": 8.991097922848663e-07, + "loss": 0.3659, + "mean_token_accuracy": 0.8691765666007996, + "num_tokens": 82662310.0, + "step": 2122 + }, + { + "epoch": 0.27006742144765294, + "grad_norm": 2.976222515106201, + "learning_rate": 8.995337007206443e-07, + "loss": 0.4429, + "mean_token_accuracy": 0.8523437976837158, + "num_tokens": 82699446.0, + "step": 2123 + }, + { + "epoch": 0.2701946317262435, + "grad_norm": 2.873825788497925, + "learning_rate": 8.999576091564221e-07, + "loss": 0.4169, + "mean_token_accuracy": 0.8566732406616211, + "num_tokens": 82740066.0, + "step": 2124 + }, + { + "epoch": 0.270321842004834, + "grad_norm": 3.5971055030822754, + "learning_rate": 9.003815175922001e-07, + "loss": 0.3772, + "mean_token_accuracy": 0.8729192018508911, + "num_tokens": 82777696.0, + "step": 2125 + }, + { + "epoch": 0.2704490522834245, + "grad_norm": 2.6589980125427246, + "learning_rate": 9.008054260279779e-07, + "loss": 0.3487, + "mean_token_accuracy": 0.8793445229530334, + "num_tokens": 82815242.0, + "step": 2126 + }, + { + "epoch": 0.270576262562015, + "grad_norm": 2.8297111988067627, + "learning_rate": 9.012293344637558e-07, + "loss": 0.4069, + "mean_token_accuracy": 0.8608638048171997, + "num_tokens": 82858538.0, + "step": 2127 + }, + { + "epoch": 0.27070347284060553, + "grad_norm": 2.821908712387085, + "learning_rate": 9.016532428995337e-07, + "loss": 0.3522, + "mean_token_accuracy": 0.8789530992507935, + "num_tokens": 82893602.0, + "step": 2128 + }, + { + "epoch": 0.27083068311919606, + "grad_norm": 2.9074459075927734, + "learning_rate": 9.020771513353115e-07, + "loss": 0.4075, + "mean_token_accuracy": 0.8640227317810059, + "num_tokens": 82930973.0, + "step": 2129 + }, + { + "epoch": 0.27095789339778653, + "grad_norm": 2.5268783569335938, + "learning_rate": 9.025010597710894e-07, + "loss": 0.348, + "mean_token_accuracy": 0.8785733580589294, + "num_tokens": 82971476.0, + "step": 2130 + }, + { + "epoch": 0.27108510367637706, + "grad_norm": 2.8952279090881348, + "learning_rate": 9.029249682068673e-07, + "loss": 0.3866, + "mean_token_accuracy": 0.8678514957427979, + "num_tokens": 83006925.0, + "step": 2131 + }, + { + "epoch": 0.2712123139549676, + "grad_norm": 3.033874750137329, + "learning_rate": 9.033488766426451e-07, + "loss": 0.374, + "mean_token_accuracy": 0.86857008934021, + "num_tokens": 83039663.0, + "step": 2132 + }, + { + "epoch": 0.27133952423355806, + "grad_norm": 2.907989501953125, + "learning_rate": 9.037727850784231e-07, + "loss": 0.3616, + "mean_token_accuracy": 0.8701077699661255, + "num_tokens": 83072212.0, + "step": 2133 + }, + { + "epoch": 0.2714667345121486, + "grad_norm": 2.7370550632476807, + "learning_rate": 9.041966935142009e-07, + "loss": 0.3935, + "mean_token_accuracy": 0.8685286641120911, + "num_tokens": 83113397.0, + "step": 2134 + }, + { + "epoch": 0.2715939447907391, + "grad_norm": 2.897282600402832, + "learning_rate": 9.046206019499788e-07, + "loss": 0.3347, + "mean_token_accuracy": 0.8818957209587097, + "num_tokens": 83144580.0, + "step": 2135 + }, + { + "epoch": 0.2717211550693296, + "grad_norm": 2.6989245414733887, + "learning_rate": 9.050445103857567e-07, + "loss": 0.4171, + "mean_token_accuracy": 0.8555472493171692, + "num_tokens": 83187671.0, + "step": 2136 + }, + { + "epoch": 0.2718483653479201, + "grad_norm": 2.8490078449249268, + "learning_rate": 9.054684188215344e-07, + "loss": 0.3981, + "mean_token_accuracy": 0.8635212182998657, + "num_tokens": 83229420.0, + "step": 2137 + }, + { + "epoch": 0.27197557562651065, + "grad_norm": 2.8820576667785645, + "learning_rate": 9.058923272573124e-07, + "loss": 0.4057, + "mean_token_accuracy": 0.861065685749054, + "num_tokens": 83271641.0, + "step": 2138 + }, + { + "epoch": 0.2721027859051011, + "grad_norm": 2.572725772857666, + "learning_rate": 9.063162356930902e-07, + "loss": 0.3835, + "mean_token_accuracy": 0.8663460612297058, + "num_tokens": 83315357.0, + "step": 2139 + }, + { + "epoch": 0.27222999618369165, + "grad_norm": 3.2066640853881836, + "learning_rate": 9.067401441288681e-07, + "loss": 0.3954, + "mean_token_accuracy": 0.8678029179573059, + "num_tokens": 83349221.0, + "step": 2140 + }, + { + "epoch": 0.2723572064622822, + "grad_norm": 2.9243547916412354, + "learning_rate": 9.07164052564646e-07, + "loss": 0.4137, + "mean_token_accuracy": 0.8596506118774414, + "num_tokens": 83390281.0, + "step": 2141 + }, + { + "epoch": 0.27248441674087265, + "grad_norm": 2.9147439002990723, + "learning_rate": 9.075879610004239e-07, + "loss": 0.394, + "mean_token_accuracy": 0.8669009208679199, + "num_tokens": 83428809.0, + "step": 2142 + }, + { + "epoch": 0.2726116270194632, + "grad_norm": 2.663212537765503, + "learning_rate": 9.080118694362017e-07, + "loss": 0.4397, + "mean_token_accuracy": 0.8562077283859253, + "num_tokens": 83474273.0, + "step": 2143 + }, + { + "epoch": 0.2727388372980537, + "grad_norm": 2.7410471439361572, + "learning_rate": 9.084357778719796e-07, + "loss": 0.3876, + "mean_token_accuracy": 0.8688539266586304, + "num_tokens": 83517386.0, + "step": 2144 + }, + { + "epoch": 0.2728660475766442, + "grad_norm": 2.7859463691711426, + "learning_rate": 9.088596863077574e-07, + "loss": 0.388, + "mean_token_accuracy": 0.8656392097473145, + "num_tokens": 83555712.0, + "step": 2145 + }, + { + "epoch": 0.2729932578552347, + "grad_norm": 3.053969144821167, + "learning_rate": 9.092835947435354e-07, + "loss": 0.4299, + "mean_token_accuracy": 0.8539001941680908, + "num_tokens": 83593113.0, + "step": 2146 + }, + { + "epoch": 0.27312046813382523, + "grad_norm": 2.803906202316284, + "learning_rate": 9.097075031793132e-07, + "loss": 0.3385, + "mean_token_accuracy": 0.8830772638320923, + "num_tokens": 83628905.0, + "step": 2147 + }, + { + "epoch": 0.2732476784124157, + "grad_norm": 2.6457269191741943, + "learning_rate": 9.101314116150911e-07, + "loss": 0.3932, + "mean_token_accuracy": 0.8649936318397522, + "num_tokens": 83671615.0, + "step": 2148 + }, + { + "epoch": 0.27337488869100623, + "grad_norm": 3.041264057159424, + "learning_rate": 9.10555320050869e-07, + "loss": 0.455, + "mean_token_accuracy": 0.8493392467498779, + "num_tokens": 83708172.0, + "step": 2149 + }, + { + "epoch": 0.27350209896959676, + "grad_norm": 2.8700263500213623, + "learning_rate": 9.109792284866469e-07, + "loss": 0.3941, + "mean_token_accuracy": 0.8654197454452515, + "num_tokens": 83747030.0, + "step": 2150 + }, + { + "epoch": 0.27362930924818724, + "grad_norm": 2.985110282897949, + "learning_rate": 9.114031369224247e-07, + "loss": 0.4181, + "mean_token_accuracy": 0.8632515668869019, + "num_tokens": 83783628.0, + "step": 2151 + }, + { + "epoch": 0.27375651952677776, + "grad_norm": 2.692819595336914, + "learning_rate": 9.118270453582026e-07, + "loss": 0.3265, + "mean_token_accuracy": 0.8858504295349121, + "num_tokens": 83819455.0, + "step": 2152 + }, + { + "epoch": 0.2738837298053683, + "grad_norm": 2.5502967834472656, + "learning_rate": 9.122509537939804e-07, + "loss": 0.3585, + "mean_token_accuracy": 0.8743770718574524, + "num_tokens": 83860561.0, + "step": 2153 + }, + { + "epoch": 0.27401094008395877, + "grad_norm": 2.6161704063415527, + "learning_rate": 9.126748622297584e-07, + "loss": 0.3531, + "mean_token_accuracy": 0.8771992325782776, + "num_tokens": 83898349.0, + "step": 2154 + }, + { + "epoch": 0.2741381503625493, + "grad_norm": 2.957636594772339, + "learning_rate": 9.130987706655362e-07, + "loss": 0.3791, + "mean_token_accuracy": 0.8721709847450256, + "num_tokens": 83933453.0, + "step": 2155 + }, + { + "epoch": 0.2742653606411398, + "grad_norm": 2.7745580673217773, + "learning_rate": 9.135226791013141e-07, + "loss": 0.3757, + "mean_token_accuracy": 0.8686031103134155, + "num_tokens": 83971851.0, + "step": 2156 + }, + { + "epoch": 0.2743925709197303, + "grad_norm": 2.7563459873199463, + "learning_rate": 9.13946587537092e-07, + "loss": 0.3494, + "mean_token_accuracy": 0.8790875673294067, + "num_tokens": 84006480.0, + "step": 2157 + }, + { + "epoch": 0.2745197811983208, + "grad_norm": 2.9021520614624023, + "learning_rate": 9.143704959728699e-07, + "loss": 0.3832, + "mean_token_accuracy": 0.8705764412879944, + "num_tokens": 84044023.0, + "step": 2158 + }, + { + "epoch": 0.27464699147691135, + "grad_norm": 2.839266061782837, + "learning_rate": 9.147944044086476e-07, + "loss": 0.3348, + "mean_token_accuracy": 0.880793571472168, + "num_tokens": 84077447.0, + "step": 2159 + }, + { + "epoch": 0.2747742017555018, + "grad_norm": 2.758436918258667, + "learning_rate": 9.152183128444255e-07, + "loss": 0.3818, + "mean_token_accuracy": 0.8701038360595703, + "num_tokens": 84116201.0, + "step": 2160 + }, + { + "epoch": 0.27490141203409235, + "grad_norm": 2.8749589920043945, + "learning_rate": 9.156422212802034e-07, + "loss": 0.3531, + "mean_token_accuracy": 0.8748533725738525, + "num_tokens": 84150903.0, + "step": 2161 + }, + { + "epoch": 0.2750286223126829, + "grad_norm": 2.7963356971740723, + "learning_rate": 9.160661297159813e-07, + "loss": 0.3808, + "mean_token_accuracy": 0.8736042976379395, + "num_tokens": 84190202.0, + "step": 2162 + }, + { + "epoch": 0.27515583259127335, + "grad_norm": 2.886669635772705, + "learning_rate": 9.164900381517592e-07, + "loss": 0.3886, + "mean_token_accuracy": 0.8663618564605713, + "num_tokens": 84222330.0, + "step": 2163 + }, + { + "epoch": 0.2752830428698639, + "grad_norm": 2.748683214187622, + "learning_rate": 9.16913946587537e-07, + "loss": 0.3794, + "mean_token_accuracy": 0.8702361583709717, + "num_tokens": 84263408.0, + "step": 2164 + }, + { + "epoch": 0.2754102531484544, + "grad_norm": 2.7353508472442627, + "learning_rate": 9.17337855023315e-07, + "loss": 0.3638, + "mean_token_accuracy": 0.8765815496444702, + "num_tokens": 84306703.0, + "step": 2165 + }, + { + "epoch": 0.2755374634270449, + "grad_norm": 2.6122844219207764, + "learning_rate": 9.177617634590928e-07, + "loss": 0.4139, + "mean_token_accuracy": 0.8553216457366943, + "num_tokens": 84350699.0, + "step": 2166 + }, + { + "epoch": 0.2756646737056354, + "grad_norm": 2.8125956058502197, + "learning_rate": 9.181856718948706e-07, + "loss": 0.3902, + "mean_token_accuracy": 0.8666671514511108, + "num_tokens": 84391169.0, + "step": 2167 + }, + { + "epoch": 0.27579188398422594, + "grad_norm": 2.7996978759765625, + "learning_rate": 9.186095803306485e-07, + "loss": 0.401, + "mean_token_accuracy": 0.8644680976867676, + "num_tokens": 84429322.0, + "step": 2168 + }, + { + "epoch": 0.2759190942628164, + "grad_norm": 2.7440764904022217, + "learning_rate": 9.190334887664264e-07, + "loss": 0.401, + "mean_token_accuracy": 0.8625486493110657, + "num_tokens": 84466526.0, + "step": 2169 + }, + { + "epoch": 0.27604630454140694, + "grad_norm": 2.5467686653137207, + "learning_rate": 9.194573972022043e-07, + "loss": 0.3608, + "mean_token_accuracy": 0.8759178519248962, + "num_tokens": 84509956.0, + "step": 2170 + }, + { + "epoch": 0.27617351481999747, + "grad_norm": 3.0585861206054688, + "learning_rate": 9.198813056379822e-07, + "loss": 0.3835, + "mean_token_accuracy": 0.8642005324363708, + "num_tokens": 84542752.0, + "step": 2171 + }, + { + "epoch": 0.27630072509858794, + "grad_norm": 2.638568162918091, + "learning_rate": 9.2030521407376e-07, + "loss": 0.3647, + "mean_token_accuracy": 0.8751203417778015, + "num_tokens": 84583948.0, + "step": 2172 + }, + { + "epoch": 0.27642793537717847, + "grad_norm": 2.433164358139038, + "learning_rate": 9.20729122509538e-07, + "loss": 0.3785, + "mean_token_accuracy": 0.8713346719741821, + "num_tokens": 84629676.0, + "step": 2173 + }, + { + "epoch": 0.276555145655769, + "grad_norm": 2.740623712539673, + "learning_rate": 9.211530309453158e-07, + "loss": 0.352, + "mean_token_accuracy": 0.8776246309280396, + "num_tokens": 84668744.0, + "step": 2174 + }, + { + "epoch": 0.27668235593435947, + "grad_norm": 2.8516557216644287, + "learning_rate": 9.215769393810936e-07, + "loss": 0.3646, + "mean_token_accuracy": 0.8720400333404541, + "num_tokens": 84703521.0, + "step": 2175 + }, + { + "epoch": 0.27680956621295, + "grad_norm": 2.8026487827301025, + "learning_rate": 9.220008478168715e-07, + "loss": 0.3512, + "mean_token_accuracy": 0.8791394233703613, + "num_tokens": 84742029.0, + "step": 2176 + }, + { + "epoch": 0.2769367764915405, + "grad_norm": 2.9213786125183105, + "learning_rate": 9.224247562526494e-07, + "loss": 0.3876, + "mean_token_accuracy": 0.8689250946044922, + "num_tokens": 84778207.0, + "step": 2177 + }, + { + "epoch": 0.277063986770131, + "grad_norm": 2.6903910636901855, + "learning_rate": 9.228486646884273e-07, + "loss": 0.3798, + "mean_token_accuracy": 0.8693406581878662, + "num_tokens": 84819359.0, + "step": 2178 + }, + { + "epoch": 0.2771911970487215, + "grad_norm": 2.900958776473999, + "learning_rate": 9.232725731242052e-07, + "loss": 0.4002, + "mean_token_accuracy": 0.862533688545227, + "num_tokens": 84857891.0, + "step": 2179 + }, + { + "epoch": 0.27731840732731206, + "grad_norm": 2.7643039226531982, + "learning_rate": 9.23696481559983e-07, + "loss": 0.3462, + "mean_token_accuracy": 0.8808859586715698, + "num_tokens": 84896935.0, + "step": 2180 + }, + { + "epoch": 0.2774456176059026, + "grad_norm": 2.922741174697876, + "learning_rate": 9.24120389995761e-07, + "loss": 0.4071, + "mean_token_accuracy": 0.8590711355209351, + "num_tokens": 84934751.0, + "step": 2181 + }, + { + "epoch": 0.27757282788449306, + "grad_norm": 2.7622110843658447, + "learning_rate": 9.245442984315387e-07, + "loss": 0.4284, + "mean_token_accuracy": 0.8525047898292542, + "num_tokens": 84978089.0, + "step": 2182 + }, + { + "epoch": 0.2777000381630836, + "grad_norm": 2.8745336532592773, + "learning_rate": 9.249682068673165e-07, + "loss": 0.4475, + "mean_token_accuracy": 0.8473780155181885, + "num_tokens": 85018463.0, + "step": 2183 + }, + { + "epoch": 0.2778272484416741, + "grad_norm": 2.852092981338501, + "learning_rate": 9.253921153030945e-07, + "loss": 0.3791, + "mean_token_accuracy": 0.8696328401565552, + "num_tokens": 85055304.0, + "step": 2184 + }, + { + "epoch": 0.2779544587202646, + "grad_norm": 3.071129083633423, + "learning_rate": 9.258160237388723e-07, + "loss": 0.3618, + "mean_token_accuracy": 0.876464307308197, + "num_tokens": 85087283.0, + "step": 2185 + }, + { + "epoch": 0.2780816689988551, + "grad_norm": 2.711784601211548, + "learning_rate": 9.262399321746503e-07, + "loss": 0.3835, + "mean_token_accuracy": 0.8653265237808228, + "num_tokens": 85127632.0, + "step": 2186 + }, + { + "epoch": 0.27820887927744564, + "grad_norm": 2.793914556503296, + "learning_rate": 9.266638406104281e-07, + "loss": 0.4121, + "mean_token_accuracy": 0.8582903146743774, + "num_tokens": 85167176.0, + "step": 2187 + }, + { + "epoch": 0.2783360895560361, + "grad_norm": 2.644357919692993, + "learning_rate": 9.27087749046206e-07, + "loss": 0.3847, + "mean_token_accuracy": 0.8661141395568848, + "num_tokens": 85209514.0, + "step": 2188 + }, + { + "epoch": 0.27846329983462664, + "grad_norm": 2.8310022354125977, + "learning_rate": 9.275116574819839e-07, + "loss": 0.3777, + "mean_token_accuracy": 0.8685463070869446, + "num_tokens": 85249048.0, + "step": 2189 + }, + { + "epoch": 0.27859051011321717, + "grad_norm": 2.6931023597717285, + "learning_rate": 9.279355659177617e-07, + "loss": 0.3703, + "mean_token_accuracy": 0.8709932565689087, + "num_tokens": 85288480.0, + "step": 2190 + }, + { + "epoch": 0.27871772039180764, + "grad_norm": 3.1262013912200928, + "learning_rate": 9.283594743535395e-07, + "loss": 0.428, + "mean_token_accuracy": 0.8551761507987976, + "num_tokens": 85323207.0, + "step": 2191 + }, + { + "epoch": 0.2788449306703982, + "grad_norm": 2.7301011085510254, + "learning_rate": 9.287833827893175e-07, + "loss": 0.4059, + "mean_token_accuracy": 0.8619102239608765, + "num_tokens": 85366335.0, + "step": 2192 + }, + { + "epoch": 0.2789721409489887, + "grad_norm": 2.8274037837982178, + "learning_rate": 9.292072912250953e-07, + "loss": 0.3888, + "mean_token_accuracy": 0.8653806447982788, + "num_tokens": 85407725.0, + "step": 2193 + }, + { + "epoch": 0.2790993512275792, + "grad_norm": 2.752108573913574, + "learning_rate": 9.296311996608733e-07, + "loss": 0.4033, + "mean_token_accuracy": 0.8622045516967773, + "num_tokens": 85448290.0, + "step": 2194 + }, + { + "epoch": 0.2792265615061697, + "grad_norm": 2.7319140434265137, + "learning_rate": 9.300551080966511e-07, + "loss": 0.3443, + "mean_token_accuracy": 0.8814617991447449, + "num_tokens": 85486876.0, + "step": 2195 + }, + { + "epoch": 0.27935377178476023, + "grad_norm": 3.047508955001831, + "learning_rate": 9.30479016532429e-07, + "loss": 0.4207, + "mean_token_accuracy": 0.8572857975959778, + "num_tokens": 85520377.0, + "step": 2196 + }, + { + "epoch": 0.2794809820633507, + "grad_norm": 2.9634757041931152, + "learning_rate": 9.309029249682068e-07, + "loss": 0.4305, + "mean_token_accuracy": 0.8542983531951904, + "num_tokens": 85559064.0, + "step": 2197 + }, + { + "epoch": 0.27960819234194123, + "grad_norm": 2.7376232147216797, + "learning_rate": 9.313268334039847e-07, + "loss": 0.3869, + "mean_token_accuracy": 0.865700364112854, + "num_tokens": 85599409.0, + "step": 2198 + }, + { + "epoch": 0.27973540262053176, + "grad_norm": 2.5869855880737305, + "learning_rate": 9.317507418397625e-07, + "loss": 0.3741, + "mean_token_accuracy": 0.8722156882286072, + "num_tokens": 85643555.0, + "step": 2199 + }, + { + "epoch": 0.27986261289912223, + "grad_norm": 2.9848599433898926, + "learning_rate": 9.321746502755404e-07, + "loss": 0.3867, + "mean_token_accuracy": 0.8642228841781616, + "num_tokens": 85674667.0, + "step": 2200 + }, + { + "epoch": 0.27998982317771276, + "grad_norm": 2.9308621883392334, + "learning_rate": 9.325985587113183e-07, + "loss": 0.3917, + "mean_token_accuracy": 0.8664067983627319, + "num_tokens": 85708715.0, + "step": 2201 + }, + { + "epoch": 0.2801170334563033, + "grad_norm": 2.5560178756713867, + "learning_rate": 9.330224671470962e-07, + "loss": 0.4166, + "mean_token_accuracy": 0.8581894636154175, + "num_tokens": 85754874.0, + "step": 2202 + }, + { + "epoch": 0.28024424373489376, + "grad_norm": 3.1868093013763428, + "learning_rate": 9.334463755828741e-07, + "loss": 0.3967, + "mean_token_accuracy": 0.864963948726654, + "num_tokens": 85786904.0, + "step": 2203 + }, + { + "epoch": 0.2803714540134843, + "grad_norm": 2.8387458324432373, + "learning_rate": 9.338702840186519e-07, + "loss": 0.3771, + "mean_token_accuracy": 0.8681018352508545, + "num_tokens": 85824719.0, + "step": 2204 + }, + { + "epoch": 0.2804986642920748, + "grad_norm": 2.5950562953948975, + "learning_rate": 9.342941924544298e-07, + "loss": 0.3487, + "mean_token_accuracy": 0.8788443207740784, + "num_tokens": 85863751.0, + "step": 2205 + }, + { + "epoch": 0.2806258745706653, + "grad_norm": 3.050027847290039, + "learning_rate": 9.347181008902076e-07, + "loss": 0.3576, + "mean_token_accuracy": 0.8726655840873718, + "num_tokens": 85895824.0, + "step": 2206 + }, + { + "epoch": 0.2807530848492558, + "grad_norm": 2.800826072692871, + "learning_rate": 9.351420093259855e-07, + "loss": 0.3556, + "mean_token_accuracy": 0.8769884705543518, + "num_tokens": 85934500.0, + "step": 2207 + }, + { + "epoch": 0.28088029512784635, + "grad_norm": 2.985473394393921, + "learning_rate": 9.355659177617634e-07, + "loss": 0.4514, + "mean_token_accuracy": 0.8435949683189392, + "num_tokens": 85973193.0, + "step": 2208 + }, + { + "epoch": 0.2810075054064368, + "grad_norm": 2.5927700996398926, + "learning_rate": 9.359898261975413e-07, + "loss": 0.3918, + "mean_token_accuracy": 0.8654505610466003, + "num_tokens": 86016927.0, + "step": 2209 + }, + { + "epoch": 0.28113471568502735, + "grad_norm": 3.070535898208618, + "learning_rate": 9.364137346333192e-07, + "loss": 0.429, + "mean_token_accuracy": 0.8521740436553955, + "num_tokens": 86051252.0, + "step": 2210 + }, + { + "epoch": 0.2812619259636179, + "grad_norm": 2.438981294631958, + "learning_rate": 9.368376430690971e-07, + "loss": 0.2685, + "mean_token_accuracy": 0.9045612215995789, + "num_tokens": 86088665.0, + "step": 2211 + }, + { + "epoch": 0.28138913624220835, + "grad_norm": 2.7678093910217285, + "learning_rate": 9.372615515048749e-07, + "loss": 0.3667, + "mean_token_accuracy": 0.8721146583557129, + "num_tokens": 86128944.0, + "step": 2212 + }, + { + "epoch": 0.2815163465207989, + "grad_norm": 2.745492458343506, + "learning_rate": 9.376854599406528e-07, + "loss": 0.3548, + "mean_token_accuracy": 0.8776040077209473, + "num_tokens": 86166015.0, + "step": 2213 + }, + { + "epoch": 0.2816435567993894, + "grad_norm": 3.047013282775879, + "learning_rate": 9.381093683764306e-07, + "loss": 0.4238, + "mean_token_accuracy": 0.8548010587692261, + "num_tokens": 86205119.0, + "step": 2214 + }, + { + "epoch": 0.2817707670779799, + "grad_norm": 3.0176286697387695, + "learning_rate": 9.385332768122085e-07, + "loss": 0.4161, + "mean_token_accuracy": 0.8594348430633545, + "num_tokens": 86238154.0, + "step": 2215 + }, + { + "epoch": 0.2818979773565704, + "grad_norm": 2.783789873123169, + "learning_rate": 9.389571852479864e-07, + "loss": 0.4264, + "mean_token_accuracy": 0.8567996025085449, + "num_tokens": 86279596.0, + "step": 2216 + }, + { + "epoch": 0.28202518763516093, + "grad_norm": 2.8835794925689697, + "learning_rate": 9.393810936837643e-07, + "loss": 0.383, + "mean_token_accuracy": 0.8667146563529968, + "num_tokens": 86318275.0, + "step": 2217 + }, + { + "epoch": 0.2821523979137514, + "grad_norm": 2.891439199447632, + "learning_rate": 9.398050021195422e-07, + "loss": 0.4254, + "mean_token_accuracy": 0.8502475023269653, + "num_tokens": 86358521.0, + "step": 2218 + }, + { + "epoch": 0.28227960819234194, + "grad_norm": 2.996589183807373, + "learning_rate": 9.402289105553201e-07, + "loss": 0.4377, + "mean_token_accuracy": 0.8513578176498413, + "num_tokens": 86395597.0, + "step": 2219 + }, + { + "epoch": 0.28240681847093246, + "grad_norm": 2.7710916996002197, + "learning_rate": 9.406528189910978e-07, + "loss": 0.3813, + "mean_token_accuracy": 0.8708346486091614, + "num_tokens": 86435770.0, + "step": 2220 + }, + { + "epoch": 0.28253402874952294, + "grad_norm": 2.6657679080963135, + "learning_rate": 9.410767274268757e-07, + "loss": 0.3731, + "mean_token_accuracy": 0.8710395693778992, + "num_tokens": 86475303.0, + "step": 2221 + }, + { + "epoch": 0.28266123902811346, + "grad_norm": 2.9542043209075928, + "learning_rate": 9.415006358626536e-07, + "loss": 0.3361, + "mean_token_accuracy": 0.8803592324256897, + "num_tokens": 86506421.0, + "step": 2222 + }, + { + "epoch": 0.282788449306704, + "grad_norm": 2.8291232585906982, + "learning_rate": 9.419245442984314e-07, + "loss": 0.3832, + "mean_token_accuracy": 0.8736134767532349, + "num_tokens": 86549220.0, + "step": 2223 + }, + { + "epoch": 0.28291565958529447, + "grad_norm": 2.9383273124694824, + "learning_rate": 9.423484527342094e-07, + "loss": 0.399, + "mean_token_accuracy": 0.8653085827827454, + "num_tokens": 86587147.0, + "step": 2224 + }, + { + "epoch": 0.283042869863885, + "grad_norm": 2.7065510749816895, + "learning_rate": 9.427723611699872e-07, + "loss": 0.3558, + "mean_token_accuracy": 0.8756369352340698, + "num_tokens": 86626248.0, + "step": 2225 + }, + { + "epoch": 0.2831700801424755, + "grad_norm": 2.691786527633667, + "learning_rate": 9.431962696057652e-07, + "loss": 0.3307, + "mean_token_accuracy": 0.8822000622749329, + "num_tokens": 86662430.0, + "step": 2226 + }, + { + "epoch": 0.283297290421066, + "grad_norm": 2.7850263118743896, + "learning_rate": 9.43620178041543e-07, + "loss": 0.3616, + "mean_token_accuracy": 0.8729929327964783, + "num_tokens": 86699252.0, + "step": 2227 + }, + { + "epoch": 0.2834245006996565, + "grad_norm": 2.7336885929107666, + "learning_rate": 9.440440864773208e-07, + "loss": 0.3828, + "mean_token_accuracy": 0.8650749325752258, + "num_tokens": 86741128.0, + "step": 2228 + }, + { + "epoch": 0.28355171097824705, + "grad_norm": 3.0601234436035156, + "learning_rate": 9.444679949130987e-07, + "loss": 0.4254, + "mean_token_accuracy": 0.856117844581604, + "num_tokens": 86779889.0, + "step": 2229 + }, + { + "epoch": 0.2836789212568376, + "grad_norm": 3.000398874282837, + "learning_rate": 9.448919033488766e-07, + "loss": 0.4031, + "mean_token_accuracy": 0.8618572354316711, + "num_tokens": 86814350.0, + "step": 2230 + }, + { + "epoch": 0.28380613153542805, + "grad_norm": 2.689890146255493, + "learning_rate": 9.453158117846544e-07, + "loss": 0.3934, + "mean_token_accuracy": 0.8619072437286377, + "num_tokens": 86854976.0, + "step": 2231 + }, + { + "epoch": 0.2839333418140186, + "grad_norm": 2.9433653354644775, + "learning_rate": 9.457397202204324e-07, + "loss": 0.4159, + "mean_token_accuracy": 0.861524224281311, + "num_tokens": 86894490.0, + "step": 2232 + }, + { + "epoch": 0.2840605520926091, + "grad_norm": 2.6285457611083984, + "learning_rate": 9.461636286562102e-07, + "loss": 0.3615, + "mean_token_accuracy": 0.8740797638893127, + "num_tokens": 86938862.0, + "step": 2233 + }, + { + "epoch": 0.2841877623711996, + "grad_norm": 2.744516611099243, + "learning_rate": 9.465875370919882e-07, + "loss": 0.3657, + "mean_token_accuracy": 0.8747351169586182, + "num_tokens": 86977741.0, + "step": 2234 + }, + { + "epoch": 0.2843149726497901, + "grad_norm": 2.9983954429626465, + "learning_rate": 9.470114455277659e-07, + "loss": 0.3762, + "mean_token_accuracy": 0.8706795573234558, + "num_tokens": 87017076.0, + "step": 2235 + }, + { + "epoch": 0.28444218292838064, + "grad_norm": 2.9948313236236572, + "learning_rate": 9.474353539635438e-07, + "loss": 0.3833, + "mean_token_accuracy": 0.8677651286125183, + "num_tokens": 87052046.0, + "step": 2236 + }, + { + "epoch": 0.2845693932069711, + "grad_norm": 2.7735724449157715, + "learning_rate": 9.478592623993217e-07, + "loss": 0.3541, + "mean_token_accuracy": 0.8767464756965637, + "num_tokens": 87090236.0, + "step": 2237 + }, + { + "epoch": 0.28469660348556164, + "grad_norm": 3.225250720977783, + "learning_rate": 9.482831708350996e-07, + "loss": 0.4165, + "mean_token_accuracy": 0.8564509153366089, + "num_tokens": 87127641.0, + "step": 2238 + }, + { + "epoch": 0.28482381376415217, + "grad_norm": 2.8530497550964355, + "learning_rate": 9.487070792708775e-07, + "loss": 0.3692, + "mean_token_accuracy": 0.8705811500549316, + "num_tokens": 87161316.0, + "step": 2239 + }, + { + "epoch": 0.28495102404274264, + "grad_norm": 2.8258440494537354, + "learning_rate": 9.491309877066554e-07, + "loss": 0.3902, + "mean_token_accuracy": 0.8650837540626526, + "num_tokens": 87198034.0, + "step": 2240 + }, + { + "epoch": 0.28507823432133317, + "grad_norm": 2.781031847000122, + "learning_rate": 9.495548961424332e-07, + "loss": 0.3963, + "mean_token_accuracy": 0.864202082157135, + "num_tokens": 87238558.0, + "step": 2241 + }, + { + "epoch": 0.2852054445999237, + "grad_norm": 2.7786920070648193, + "learning_rate": 9.499788045782111e-07, + "loss": 0.3636, + "mean_token_accuracy": 0.8743617534637451, + "num_tokens": 87278098.0, + "step": 2242 + }, + { + "epoch": 0.28533265487851417, + "grad_norm": 2.840749740600586, + "learning_rate": 9.504027130139889e-07, + "loss": 0.3618, + "mean_token_accuracy": 0.8749897480010986, + "num_tokens": 87315778.0, + "step": 2243 + }, + { + "epoch": 0.2854598651571047, + "grad_norm": 2.859877347946167, + "learning_rate": 9.508266214497667e-07, + "loss": 0.3928, + "mean_token_accuracy": 0.8638923168182373, + "num_tokens": 87354375.0, + "step": 2244 + }, + { + "epoch": 0.2855870754356952, + "grad_norm": 2.6770687103271484, + "learning_rate": 9.512505298855447e-07, + "loss": 0.366, + "mean_token_accuracy": 0.8735561370849609, + "num_tokens": 87393456.0, + "step": 2245 + }, + { + "epoch": 0.2857142857142857, + "grad_norm": 2.9870412349700928, + "learning_rate": 9.516744383213225e-07, + "loss": 0.3764, + "mean_token_accuracy": 0.8691232800483704, + "num_tokens": 87429714.0, + "step": 2246 + }, + { + "epoch": 0.2858414959928762, + "grad_norm": 2.843510389328003, + "learning_rate": 9.520983467571005e-07, + "loss": 0.4289, + "mean_token_accuracy": 0.8520975708961487, + "num_tokens": 87467755.0, + "step": 2247 + }, + { + "epoch": 0.28596870627146675, + "grad_norm": 2.9264652729034424, + "learning_rate": 9.525222551928783e-07, + "loss": 0.4412, + "mean_token_accuracy": 0.8518091440200806, + "num_tokens": 87506961.0, + "step": 2248 + }, + { + "epoch": 0.2860959165500572, + "grad_norm": 2.564242362976074, + "learning_rate": 9.529461636286562e-07, + "loss": 0.3264, + "mean_token_accuracy": 0.8872823715209961, + "num_tokens": 87544854.0, + "step": 2249 + }, + { + "epoch": 0.28622312682864776, + "grad_norm": 2.921067953109741, + "learning_rate": 9.533700720644341e-07, + "loss": 0.3794, + "mean_token_accuracy": 0.8687934875488281, + "num_tokens": 87578592.0, + "step": 2250 + }, + { + "epoch": 0.2863503371072383, + "grad_norm": 2.7284975051879883, + "learning_rate": 9.537939805002118e-07, + "loss": 0.3887, + "mean_token_accuracy": 0.8650225400924683, + "num_tokens": 87620180.0, + "step": 2251 + }, + { + "epoch": 0.28647754738582876, + "grad_norm": 2.77010178565979, + "learning_rate": 9.542178889359898e-07, + "loss": 0.4094, + "mean_token_accuracy": 0.8615896105766296, + "num_tokens": 87662515.0, + "step": 2252 + }, + { + "epoch": 0.2866047576644193, + "grad_norm": 2.8689310550689697, + "learning_rate": 9.546417973717677e-07, + "loss": 0.3821, + "mean_token_accuracy": 0.8691998720169067, + "num_tokens": 87701286.0, + "step": 2253 + }, + { + "epoch": 0.2867319679430098, + "grad_norm": 2.947037696838379, + "learning_rate": 9.550657058075455e-07, + "loss": 0.3743, + "mean_token_accuracy": 0.870458722114563, + "num_tokens": 87738291.0, + "step": 2254 + }, + { + "epoch": 0.2868591782216003, + "grad_norm": 2.865156888961792, + "learning_rate": 9.554896142433234e-07, + "loss": 0.3839, + "mean_token_accuracy": 0.8709405660629272, + "num_tokens": 87774049.0, + "step": 2255 + }, + { + "epoch": 0.2869863885001908, + "grad_norm": 2.8339858055114746, + "learning_rate": 9.559135226791012e-07, + "loss": 0.3275, + "mean_token_accuracy": 0.8802194595336914, + "num_tokens": 87806425.0, + "step": 2256 + }, + { + "epoch": 0.28711359877878134, + "grad_norm": 2.7519752979278564, + "learning_rate": 9.563374311148793e-07, + "loss": 0.3582, + "mean_token_accuracy": 0.8708231449127197, + "num_tokens": 87843570.0, + "step": 2257 + }, + { + "epoch": 0.2872408090573718, + "grad_norm": 2.8286116123199463, + "learning_rate": 9.56761339550657e-07, + "loss": 0.3773, + "mean_token_accuracy": 0.8698127269744873, + "num_tokens": 87883589.0, + "step": 2258 + }, + { + "epoch": 0.28736801933596234, + "grad_norm": 3.1616032123565674, + "learning_rate": 9.57185247986435e-07, + "loss": 0.4097, + "mean_token_accuracy": 0.8610409498214722, + "num_tokens": 87919066.0, + "step": 2259 + }, + { + "epoch": 0.28749522961455287, + "grad_norm": 2.644240140914917, + "learning_rate": 9.576091564222128e-07, + "loss": 0.38, + "mean_token_accuracy": 0.8688547611236572, + "num_tokens": 87963206.0, + "step": 2260 + }, + { + "epoch": 0.28762243989314334, + "grad_norm": 2.8518645763397217, + "learning_rate": 9.580330648579906e-07, + "loss": 0.3902, + "mean_token_accuracy": 0.8674885630607605, + "num_tokens": 88002005.0, + "step": 2261 + }, + { + "epoch": 0.2877496501717339, + "grad_norm": 3.125490188598633, + "learning_rate": 9.584569732937685e-07, + "loss": 0.3687, + "mean_token_accuracy": 0.8732179403305054, + "num_tokens": 88035297.0, + "step": 2262 + }, + { + "epoch": 0.2878768604503244, + "grad_norm": 2.6753435134887695, + "learning_rate": 9.588808817295463e-07, + "loss": 0.3618, + "mean_token_accuracy": 0.8754271268844604, + "num_tokens": 88076871.0, + "step": 2263 + }, + { + "epoch": 0.2880040707289149, + "grad_norm": 3.045807123184204, + "learning_rate": 9.593047901653242e-07, + "loss": 0.3773, + "mean_token_accuracy": 0.8699828386306763, + "num_tokens": 88113441.0, + "step": 2264 + }, + { + "epoch": 0.2881312810075054, + "grad_norm": 3.1963717937469482, + "learning_rate": 9.597286986011022e-07, + "loss": 0.3904, + "mean_token_accuracy": 0.8668310642242432, + "num_tokens": 88151254.0, + "step": 2265 + }, + { + "epoch": 0.28825849128609593, + "grad_norm": 3.0200703144073486, + "learning_rate": 9.601526070368799e-07, + "loss": 0.422, + "mean_token_accuracy": 0.8573436737060547, + "num_tokens": 88188122.0, + "step": 2266 + }, + { + "epoch": 0.2883857015646864, + "grad_norm": 2.6291067600250244, + "learning_rate": 9.60576515472658e-07, + "loss": 0.3589, + "mean_token_accuracy": 0.8753935694694519, + "num_tokens": 88229420.0, + "step": 2267 + }, + { + "epoch": 0.28851291184327693, + "grad_norm": 2.8677055835723877, + "learning_rate": 9.610004239084358e-07, + "loss": 0.3654, + "mean_token_accuracy": 0.8736908435821533, + "num_tokens": 88265753.0, + "step": 2268 + }, + { + "epoch": 0.28864012212186746, + "grad_norm": 2.8853342533111572, + "learning_rate": 9.614243323442136e-07, + "loss": 0.3698, + "mean_token_accuracy": 0.8709169030189514, + "num_tokens": 88300245.0, + "step": 2269 + }, + { + "epoch": 0.28876733240045793, + "grad_norm": 2.96233868598938, + "learning_rate": 9.618482407799915e-07, + "loss": 0.3538, + "mean_token_accuracy": 0.8783504962921143, + "num_tokens": 88335535.0, + "step": 2270 + }, + { + "epoch": 0.28889454267904846, + "grad_norm": 3.137964963912964, + "learning_rate": 9.622721492157693e-07, + "loss": 0.3498, + "mean_token_accuracy": 0.8791447877883911, + "num_tokens": 88373696.0, + "step": 2271 + }, + { + "epoch": 0.289021752957639, + "grad_norm": 3.131693124771118, + "learning_rate": 9.626960576515472e-07, + "loss": 0.3816, + "mean_token_accuracy": 0.8677778244018555, + "num_tokens": 88406047.0, + "step": 2272 + }, + { + "epoch": 0.28914896323622946, + "grad_norm": 2.884108781814575, + "learning_rate": 9.63119966087325e-07, + "loss": 0.3685, + "mean_token_accuracy": 0.8721436262130737, + "num_tokens": 88436463.0, + "step": 2273 + }, + { + "epoch": 0.28927617351482, + "grad_norm": 2.659634590148926, + "learning_rate": 9.635438745231029e-07, + "loss": 0.3233, + "mean_token_accuracy": 0.8860698342323303, + "num_tokens": 88472662.0, + "step": 2274 + }, + { + "epoch": 0.2894033837934105, + "grad_norm": 2.8044793605804443, + "learning_rate": 9.63967782958881e-07, + "loss": 0.3755, + "mean_token_accuracy": 0.8692370653152466, + "num_tokens": 88512037.0, + "step": 2275 + }, + { + "epoch": 0.289530594072001, + "grad_norm": 2.94474720954895, + "learning_rate": 9.643916913946588e-07, + "loss": 0.4061, + "mean_token_accuracy": 0.8604599833488464, + "num_tokens": 88548936.0, + "step": 2276 + }, + { + "epoch": 0.2896578043505915, + "grad_norm": 2.9403395652770996, + "learning_rate": 9.648155998304366e-07, + "loss": 0.3955, + "mean_token_accuracy": 0.8608959913253784, + "num_tokens": 88582395.0, + "step": 2277 + }, + { + "epoch": 0.28978501462918205, + "grad_norm": 2.667269229888916, + "learning_rate": 9.652395082662145e-07, + "loss": 0.382, + "mean_token_accuracy": 0.8729516267776489, + "num_tokens": 88621123.0, + "step": 2278 + }, + { + "epoch": 0.2899122249077726, + "grad_norm": 2.8788299560546875, + "learning_rate": 9.656634167019923e-07, + "loss": 0.3665, + "mean_token_accuracy": 0.8710350394248962, + "num_tokens": 88655529.0, + "step": 2279 + }, + { + "epoch": 0.29003943518636305, + "grad_norm": 2.7039928436279297, + "learning_rate": 9.660873251377701e-07, + "loss": 0.3974, + "mean_token_accuracy": 0.8613495826721191, + "num_tokens": 88697175.0, + "step": 2280 + }, + { + "epoch": 0.2901666454649536, + "grad_norm": 2.775266647338867, + "learning_rate": 9.66511233573548e-07, + "loss": 0.393, + "mean_token_accuracy": 0.8641763925552368, + "num_tokens": 88739427.0, + "step": 2281 + }, + { + "epoch": 0.2902938557435441, + "grad_norm": 2.726736545562744, + "learning_rate": 9.669351420093258e-07, + "loss": 0.3452, + "mean_token_accuracy": 0.8774977326393127, + "num_tokens": 88777383.0, + "step": 2282 + }, + { + "epoch": 0.2904210660221346, + "grad_norm": 2.8890626430511475, + "learning_rate": 9.67359050445104e-07, + "loss": 0.3628, + "mean_token_accuracy": 0.8747849464416504, + "num_tokens": 88813046.0, + "step": 2283 + }, + { + "epoch": 0.2905482763007251, + "grad_norm": 2.940753936767578, + "learning_rate": 9.677829588808817e-07, + "loss": 0.4269, + "mean_token_accuracy": 0.8530837297439575, + "num_tokens": 88850956.0, + "step": 2284 + }, + { + "epoch": 0.29067548657931563, + "grad_norm": 2.7933743000030518, + "learning_rate": 9.682068673166596e-07, + "loss": 0.4151, + "mean_token_accuracy": 0.8586203455924988, + "num_tokens": 88895110.0, + "step": 2285 + }, + { + "epoch": 0.2908026968579061, + "grad_norm": 2.7546536922454834, + "learning_rate": 9.686307757524374e-07, + "loss": 0.3879, + "mean_token_accuracy": 0.8661273717880249, + "num_tokens": 88935442.0, + "step": 2286 + }, + { + "epoch": 0.29092990713649663, + "grad_norm": 2.5243330001831055, + "learning_rate": 9.690546841882153e-07, + "loss": 0.3696, + "mean_token_accuracy": 0.8737214803695679, + "num_tokens": 88979499.0, + "step": 2287 + }, + { + "epoch": 0.29105711741508716, + "grad_norm": 3.0177948474884033, + "learning_rate": 9.694785926239931e-07, + "loss": 0.4031, + "mean_token_accuracy": 0.8614988327026367, + "num_tokens": 89018010.0, + "step": 2288 + }, + { + "epoch": 0.29118432769367764, + "grad_norm": 2.5237138271331787, + "learning_rate": 9.69902501059771e-07, + "loss": 0.3569, + "mean_token_accuracy": 0.8761220574378967, + "num_tokens": 89061941.0, + "step": 2289 + }, + { + "epoch": 0.29131153797226816, + "grad_norm": 2.7622556686401367, + "learning_rate": 9.703264094955488e-07, + "loss": 0.3878, + "mean_token_accuracy": 0.8673961758613586, + "num_tokens": 89101434.0, + "step": 2290 + }, + { + "epoch": 0.2914387482508587, + "grad_norm": 2.5812625885009766, + "learning_rate": 9.707503179313269e-07, + "loss": 0.3674, + "mean_token_accuracy": 0.8731054663658142, + "num_tokens": 89143169.0, + "step": 2291 + }, + { + "epoch": 0.29156595852944917, + "grad_norm": 2.6583385467529297, + "learning_rate": 9.711742263671047e-07, + "loss": 0.3941, + "mean_token_accuracy": 0.8651498556137085, + "num_tokens": 89188679.0, + "step": 2292 + }, + { + "epoch": 0.2916931688080397, + "grad_norm": 2.7237296104431152, + "learning_rate": 9.715981348028826e-07, + "loss": 0.4078, + "mean_token_accuracy": 0.8580435514450073, + "num_tokens": 89228330.0, + "step": 2293 + }, + { + "epoch": 0.2918203790866302, + "grad_norm": 2.898108720779419, + "learning_rate": 9.720220432386604e-07, + "loss": 0.366, + "mean_token_accuracy": 0.8723127245903015, + "num_tokens": 89261628.0, + "step": 2294 + }, + { + "epoch": 0.2919475893652207, + "grad_norm": 2.7612192630767822, + "learning_rate": 9.724459516744383e-07, + "loss": 0.3656, + "mean_token_accuracy": 0.8690080642700195, + "num_tokens": 89299587.0, + "step": 2295 + }, + { + "epoch": 0.2920747996438112, + "grad_norm": 2.7158641815185547, + "learning_rate": 9.728698601102161e-07, + "loss": 0.3384, + "mean_token_accuracy": 0.880020260810852, + "num_tokens": 89337094.0, + "step": 2296 + }, + { + "epoch": 0.29220200992240175, + "grad_norm": 2.974348306655884, + "learning_rate": 9.73293768545994e-07, + "loss": 0.3835, + "mean_token_accuracy": 0.8701677918434143, + "num_tokens": 89371053.0, + "step": 2297 + }, + { + "epoch": 0.2923292202009922, + "grad_norm": 2.778455972671509, + "learning_rate": 9.737176769817718e-07, + "loss": 0.406, + "mean_token_accuracy": 0.8603003025054932, + "num_tokens": 89411045.0, + "step": 2298 + }, + { + "epoch": 0.29245643047958275, + "grad_norm": 2.6931204795837402, + "learning_rate": 9.741415854175499e-07, + "loss": 0.3735, + "mean_token_accuracy": 0.8687639832496643, + "num_tokens": 89450929.0, + "step": 2299 + }, + { + "epoch": 0.2925836407581733, + "grad_norm": 2.996934413909912, + "learning_rate": 9.745654938533277e-07, + "loss": 0.406, + "mean_token_accuracy": 0.8577395677566528, + "num_tokens": 89486164.0, + "step": 2300 + }, + { + "epoch": 0.29271085103676375, + "grad_norm": 3.1178925037384033, + "learning_rate": 9.749894022891056e-07, + "loss": 0.4449, + "mean_token_accuracy": 0.8481311202049255, + "num_tokens": 89519102.0, + "step": 2301 + }, + { + "epoch": 0.2928380613153543, + "grad_norm": 2.762969493865967, + "learning_rate": 9.754133107248834e-07, + "loss": 0.3737, + "mean_token_accuracy": 0.87203049659729, + "num_tokens": 89560253.0, + "step": 2302 + }, + { + "epoch": 0.2929652715939448, + "grad_norm": 2.5904035568237305, + "learning_rate": 9.758372191606612e-07, + "loss": 0.4145, + "mean_token_accuracy": 0.8607776165008545, + "num_tokens": 89607039.0, + "step": 2303 + }, + { + "epoch": 0.2930924818725353, + "grad_norm": 2.785189628601074, + "learning_rate": 9.76261127596439e-07, + "loss": 0.3852, + "mean_token_accuracy": 0.8640811443328857, + "num_tokens": 89650024.0, + "step": 2304 + }, + { + "epoch": 0.2932196921511258, + "grad_norm": 2.9949018955230713, + "learning_rate": 9.76685036032217e-07, + "loss": 0.4204, + "mean_token_accuracy": 0.8545665740966797, + "num_tokens": 89688920.0, + "step": 2305 + }, + { + "epoch": 0.29334690242971634, + "grad_norm": 2.773519992828369, + "learning_rate": 9.771089444679948e-07, + "loss": 0.35, + "mean_token_accuracy": 0.8762860894203186, + "num_tokens": 89727941.0, + "step": 2306 + }, + { + "epoch": 0.2934741127083068, + "grad_norm": 2.908632278442383, + "learning_rate": 9.775328529037728e-07, + "loss": 0.4172, + "mean_token_accuracy": 0.8558127880096436, + "num_tokens": 89764938.0, + "step": 2307 + }, + { + "epoch": 0.29360132298689734, + "grad_norm": 2.709054470062256, + "learning_rate": 9.779567613395507e-07, + "loss": 0.3587, + "mean_token_accuracy": 0.8756242990493774, + "num_tokens": 89804820.0, + "step": 2308 + }, + { + "epoch": 0.29372853326548787, + "grad_norm": 2.814892053604126, + "learning_rate": 9.783806697753285e-07, + "loss": 0.3877, + "mean_token_accuracy": 0.867827296257019, + "num_tokens": 89843992.0, + "step": 2309 + }, + { + "epoch": 0.29385574354407834, + "grad_norm": 2.7891576290130615, + "learning_rate": 9.788045782111064e-07, + "loss": 0.3937, + "mean_token_accuracy": 0.8644037842750549, + "num_tokens": 89883165.0, + "step": 2310 + }, + { + "epoch": 0.29398295382266887, + "grad_norm": 2.7745978832244873, + "learning_rate": 9.792284866468842e-07, + "loss": 0.3592, + "mean_token_accuracy": 0.8750046491622925, + "num_tokens": 89923018.0, + "step": 2311 + }, + { + "epoch": 0.2941101641012594, + "grad_norm": 2.664064645767212, + "learning_rate": 9.79652395082662e-07, + "loss": 0.3848, + "mean_token_accuracy": 0.8694027066230774, + "num_tokens": 89963596.0, + "step": 2312 + }, + { + "epoch": 0.29423737437984987, + "grad_norm": 2.7186734676361084, + "learning_rate": 9.8007630351844e-07, + "loss": 0.3839, + "mean_token_accuracy": 0.8695852160453796, + "num_tokens": 90004453.0, + "step": 2313 + }, + { + "epoch": 0.2943645846584404, + "grad_norm": 2.527956008911133, + "learning_rate": 9.805002119542178e-07, + "loss": 0.3622, + "mean_token_accuracy": 0.8749796152114868, + "num_tokens": 90045067.0, + "step": 2314 + }, + { + "epoch": 0.2944917949370309, + "grad_norm": 2.790546178817749, + "learning_rate": 9.809241203899958e-07, + "loss": 0.3839, + "mean_token_accuracy": 0.8692500591278076, + "num_tokens": 90083879.0, + "step": 2315 + }, + { + "epoch": 0.2946190052156214, + "grad_norm": 2.836688756942749, + "learning_rate": 9.813480288257737e-07, + "loss": 0.4511, + "mean_token_accuracy": 0.8473811149597168, + "num_tokens": 90125470.0, + "step": 2316 + }, + { + "epoch": 0.2947462154942119, + "grad_norm": 2.829617977142334, + "learning_rate": 9.817719372615515e-07, + "loss": 0.3722, + "mean_token_accuracy": 0.872199535369873, + "num_tokens": 90161893.0, + "step": 2317 + }, + { + "epoch": 0.29487342577280246, + "grad_norm": 2.7933108806610107, + "learning_rate": 9.821958456973294e-07, + "loss": 0.3651, + "mean_token_accuracy": 0.8734002709388733, + "num_tokens": 90198722.0, + "step": 2318 + }, + { + "epoch": 0.29500063605139293, + "grad_norm": 2.6929757595062256, + "learning_rate": 9.826197541331072e-07, + "loss": 0.3804, + "mean_token_accuracy": 0.8698097467422485, + "num_tokens": 90239901.0, + "step": 2319 + }, + { + "epoch": 0.29512784632998346, + "grad_norm": 3.173417329788208, + "learning_rate": 9.83043662568885e-07, + "loss": 0.4863, + "mean_token_accuracy": 0.8397109508514404, + "num_tokens": 90276575.0, + "step": 2320 + }, + { + "epoch": 0.295255056608574, + "grad_norm": 2.694688558578491, + "learning_rate": 9.83467571004663e-07, + "loss": 0.3738, + "mean_token_accuracy": 0.8691594004631042, + "num_tokens": 90317930.0, + "step": 2321 + }, + { + "epoch": 0.29538226688716446, + "grad_norm": 2.8908746242523193, + "learning_rate": 9.838914794404407e-07, + "loss": 0.4484, + "mean_token_accuracy": 0.8528677821159363, + "num_tokens": 90359705.0, + "step": 2322 + }, + { + "epoch": 0.295509477165755, + "grad_norm": 2.979226589202881, + "learning_rate": 9.843153878762188e-07, + "loss": 0.4006, + "mean_token_accuracy": 0.8636234998703003, + "num_tokens": 90399685.0, + "step": 2323 + }, + { + "epoch": 0.2956366874443455, + "grad_norm": 2.646864652633667, + "learning_rate": 9.847392963119966e-07, + "loss": 0.4051, + "mean_token_accuracy": 0.8610400557518005, + "num_tokens": 90445215.0, + "step": 2324 + }, + { + "epoch": 0.295763897722936, + "grad_norm": 2.705331325531006, + "learning_rate": 9.851632047477745e-07, + "loss": 0.3532, + "mean_token_accuracy": 0.8789695501327515, + "num_tokens": 90482637.0, + "step": 2325 + }, + { + "epoch": 0.2958911080015265, + "grad_norm": 2.5720651149749756, + "learning_rate": 9.855871131835523e-07, + "loss": 0.3605, + "mean_token_accuracy": 0.8714461326599121, + "num_tokens": 90522366.0, + "step": 2326 + }, + { + "epoch": 0.29601831828011704, + "grad_norm": 2.7731292247772217, + "learning_rate": 9.860110216193302e-07, + "loss": 0.3672, + "mean_token_accuracy": 0.8724501132965088, + "num_tokens": 90562100.0, + "step": 2327 + }, + { + "epoch": 0.2961455285587075, + "grad_norm": 2.5897600650787354, + "learning_rate": 9.86434930055108e-07, + "loss": 0.3846, + "mean_token_accuracy": 0.8667479157447815, + "num_tokens": 90605012.0, + "step": 2328 + }, + { + "epoch": 0.29627273883729804, + "grad_norm": 2.833571672439575, + "learning_rate": 9.868588384908859e-07, + "loss": 0.378, + "mean_token_accuracy": 0.8720180988311768, + "num_tokens": 90643734.0, + "step": 2329 + }, + { + "epoch": 0.2963999491158886, + "grad_norm": 2.852104663848877, + "learning_rate": 9.872827469266637e-07, + "loss": 0.3622, + "mean_token_accuracy": 0.8747272491455078, + "num_tokens": 90677364.0, + "step": 2330 + }, + { + "epoch": 0.2965271593944791, + "grad_norm": 2.942983865737915, + "learning_rate": 9.877066553624418e-07, + "loss": 0.3999, + "mean_token_accuracy": 0.8620853424072266, + "num_tokens": 90714445.0, + "step": 2331 + }, + { + "epoch": 0.2966543696730696, + "grad_norm": 2.7599992752075195, + "learning_rate": 9.881305637982196e-07, + "loss": 0.3872, + "mean_token_accuracy": 0.8675711154937744, + "num_tokens": 90753895.0, + "step": 2332 + }, + { + "epoch": 0.2967815799516601, + "grad_norm": 2.8682262897491455, + "learning_rate": 9.885544722339975e-07, + "loss": 0.3792, + "mean_token_accuracy": 0.8752695918083191, + "num_tokens": 90789121.0, + "step": 2333 + }, + { + "epoch": 0.29690879023025063, + "grad_norm": 3.059065103530884, + "learning_rate": 9.889783806697753e-07, + "loss": 0.3862, + "mean_token_accuracy": 0.8662296533584595, + "num_tokens": 90822878.0, + "step": 2334 + }, + { + "epoch": 0.2970360005088411, + "grad_norm": 3.016775369644165, + "learning_rate": 9.894022891055532e-07, + "loss": 0.4004, + "mean_token_accuracy": 0.8606677651405334, + "num_tokens": 90857479.0, + "step": 2335 + }, + { + "epoch": 0.29716321078743163, + "grad_norm": 2.928112745285034, + "learning_rate": 9.89826197541331e-07, + "loss": 0.4427, + "mean_token_accuracy": 0.8531336784362793, + "num_tokens": 90897602.0, + "step": 2336 + }, + { + "epoch": 0.29729042106602216, + "grad_norm": 2.940838098526001, + "learning_rate": 9.902501059771089e-07, + "loss": 0.4392, + "mean_token_accuracy": 0.8516507148742676, + "num_tokens": 90936200.0, + "step": 2337 + }, + { + "epoch": 0.29741763134461263, + "grad_norm": 2.6256401538848877, + "learning_rate": 9.906740144128867e-07, + "loss": 0.3626, + "mean_token_accuracy": 0.8756101131439209, + "num_tokens": 90975811.0, + "step": 2338 + }, + { + "epoch": 0.29754484162320316, + "grad_norm": 3.112347364425659, + "learning_rate": 9.910979228486648e-07, + "loss": 0.4368, + "mean_token_accuracy": 0.849376380443573, + "num_tokens": 91007952.0, + "step": 2339 + }, + { + "epoch": 0.2976720519017937, + "grad_norm": 2.6751792430877686, + "learning_rate": 9.915218312844426e-07, + "loss": 0.3471, + "mean_token_accuracy": 0.8795270919799805, + "num_tokens": 91047878.0, + "step": 2340 + }, + { + "epoch": 0.29779926218038416, + "grad_norm": 2.9713966846466064, + "learning_rate": 9.919457397202205e-07, + "loss": 0.4138, + "mean_token_accuracy": 0.8625006675720215, + "num_tokens": 91085613.0, + "step": 2341 + }, + { + "epoch": 0.2979264724589747, + "grad_norm": 2.7131528854370117, + "learning_rate": 9.923696481559983e-07, + "loss": 0.3615, + "mean_token_accuracy": 0.8768137693405151, + "num_tokens": 91124266.0, + "step": 2342 + }, + { + "epoch": 0.2980536827375652, + "grad_norm": 2.766042470932007, + "learning_rate": 9.927935565917761e-07, + "loss": 0.3326, + "mean_token_accuracy": 0.8851811289787292, + "num_tokens": 91156619.0, + "step": 2343 + }, + { + "epoch": 0.2981808930161557, + "grad_norm": 2.8201866149902344, + "learning_rate": 9.93217465027554e-07, + "loss": 0.3565, + "mean_token_accuracy": 0.877985954284668, + "num_tokens": 91191317.0, + "step": 2344 + }, + { + "epoch": 0.2983081032947462, + "grad_norm": 2.7411141395568848, + "learning_rate": 9.936413734633318e-07, + "loss": 0.4403, + "mean_token_accuracy": 0.856843113899231, + "num_tokens": 91232895.0, + "step": 2345 + }, + { + "epoch": 0.29843531357333675, + "grad_norm": 2.8927695751190186, + "learning_rate": 9.940652818991097e-07, + "loss": 0.4257, + "mean_token_accuracy": 0.857388973236084, + "num_tokens": 91269130.0, + "step": 2346 + }, + { + "epoch": 0.2985625238519272, + "grad_norm": 2.9010677337646484, + "learning_rate": 9.944891903348877e-07, + "loss": 0.3777, + "mean_token_accuracy": 0.8699735403060913, + "num_tokens": 91304442.0, + "step": 2347 + }, + { + "epoch": 0.29868973413051775, + "grad_norm": 2.6836743354797363, + "learning_rate": 9.949130987706656e-07, + "loss": 0.3952, + "mean_token_accuracy": 0.8667117357254028, + "num_tokens": 91346507.0, + "step": 2348 + }, + { + "epoch": 0.2988169444091083, + "grad_norm": 3.026071786880493, + "learning_rate": 9.953370072064432e-07, + "loss": 0.4548, + "mean_token_accuracy": 0.8447823524475098, + "num_tokens": 91384545.0, + "step": 2349 + }, + { + "epoch": 0.29894415468769875, + "grad_norm": 2.7025344371795654, + "learning_rate": 9.957609156422213e-07, + "loss": 0.3882, + "mean_token_accuracy": 0.869705855846405, + "num_tokens": 91423465.0, + "step": 2350 + }, + { + "epoch": 0.2990713649662893, + "grad_norm": 2.840877056121826, + "learning_rate": 9.961848240779991e-07, + "loss": 0.3362, + "mean_token_accuracy": 0.8799127340316772, + "num_tokens": 91458759.0, + "step": 2351 + }, + { + "epoch": 0.2991985752448798, + "grad_norm": 2.5472562313079834, + "learning_rate": 9.96608732513777e-07, + "loss": 0.3444, + "mean_token_accuracy": 0.8821926712989807, + "num_tokens": 91499401.0, + "step": 2352 + }, + { + "epoch": 0.2993257855234703, + "grad_norm": 3.152791976928711, + "learning_rate": 9.970326409495548e-07, + "loss": 0.4189, + "mean_token_accuracy": 0.8583439588546753, + "num_tokens": 91547053.0, + "step": 2353 + }, + { + "epoch": 0.2994529958020608, + "grad_norm": 2.7390260696411133, + "learning_rate": 9.974565493853327e-07, + "loss": 0.4064, + "mean_token_accuracy": 0.8644988536834717, + "num_tokens": 91589500.0, + "step": 2354 + }, + { + "epoch": 0.29958020608065133, + "grad_norm": 2.5785434246063232, + "learning_rate": 9.978804578211107e-07, + "loss": 0.3513, + "mean_token_accuracy": 0.8769204020500183, + "num_tokens": 91630963.0, + "step": 2355 + }, + { + "epoch": 0.2997074163592418, + "grad_norm": 2.8077712059020996, + "learning_rate": 9.983043662568886e-07, + "loss": 0.3541, + "mean_token_accuracy": 0.8781718015670776, + "num_tokens": 91666101.0, + "step": 2356 + }, + { + "epoch": 0.29983462663783234, + "grad_norm": 3.015226364135742, + "learning_rate": 9.987282746926662e-07, + "loss": 0.3875, + "mean_token_accuracy": 0.8668273091316223, + "num_tokens": 91699278.0, + "step": 2357 + }, + { + "epoch": 0.29996183691642286, + "grad_norm": 2.659271240234375, + "learning_rate": 9.991521831284443e-07, + "loss": 0.3704, + "mean_token_accuracy": 0.8742287755012512, + "num_tokens": 91739761.0, + "step": 2358 + }, + { + "epoch": 0.30008904719501334, + "grad_norm": 2.9246561527252197, + "learning_rate": 9.995760915642221e-07, + "loss": 0.3877, + "mean_token_accuracy": 0.8647199273109436, + "num_tokens": 91778530.0, + "step": 2359 + }, + { + "epoch": 0.30021625747360386, + "grad_norm": 2.6710140705108643, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8744999766349792, + "num_tokens": 91818196.0, + "step": 2360 + }, + { + "epoch": 0.3003434677521944, + "grad_norm": 2.9034461975097656, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.878049910068512, + "num_tokens": 91852264.0, + "step": 2361 + }, + { + "epoch": 0.30047067803078487, + "grad_norm": 2.554995059967041, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8768774271011353, + "num_tokens": 91895637.0, + "step": 2362 + }, + { + "epoch": 0.3005978883093754, + "grad_norm": 2.690037250518799, + "learning_rate": 1e-06, + "loss": 0.4046, + "mean_token_accuracy": 0.8625532388687134, + "num_tokens": 91940634.0, + "step": 2363 + }, + { + "epoch": 0.3007250985879659, + "grad_norm": 2.7225985527038574, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8794081807136536, + "num_tokens": 91974915.0, + "step": 2364 + }, + { + "epoch": 0.3008523088665564, + "grad_norm": 2.782878875732422, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8699139952659607, + "num_tokens": 92018595.0, + "step": 2365 + }, + { + "epoch": 0.3009795191451469, + "grad_norm": 2.603720188140869, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8680302500724792, + "num_tokens": 92063934.0, + "step": 2366 + }, + { + "epoch": 0.30110672942373745, + "grad_norm": 2.851264238357544, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8727489709854126, + "num_tokens": 92099386.0, + "step": 2367 + }, + { + "epoch": 0.3012339397023279, + "grad_norm": 2.5949058532714844, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8700655102729797, + "num_tokens": 92145989.0, + "step": 2368 + }, + { + "epoch": 0.30136114998091845, + "grad_norm": 2.8713767528533936, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8668208122253418, + "num_tokens": 92181313.0, + "step": 2369 + }, + { + "epoch": 0.301488360259509, + "grad_norm": 2.634929656982422, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8726486563682556, + "num_tokens": 92224219.0, + "step": 2370 + }, + { + "epoch": 0.30161557053809945, + "grad_norm": 2.5686373710632324, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8718521595001221, + "num_tokens": 92269932.0, + "step": 2371 + }, + { + "epoch": 0.30174278081669, + "grad_norm": 2.819525718688965, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.866909921169281, + "num_tokens": 92310760.0, + "step": 2372 + }, + { + "epoch": 0.3018699910952805, + "grad_norm": 2.8618369102478027, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.866919755935669, + "num_tokens": 92347382.0, + "step": 2373 + }, + { + "epoch": 0.301997201373871, + "grad_norm": 2.7064688205718994, + "learning_rate": 1e-06, + "loss": 0.4018, + "mean_token_accuracy": 0.8652317523956299, + "num_tokens": 92391246.0, + "step": 2374 + }, + { + "epoch": 0.3021244116524615, + "grad_norm": 2.7533035278320312, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8784241676330566, + "num_tokens": 92428886.0, + "step": 2375 + }, + { + "epoch": 0.30225162193105204, + "grad_norm": 2.9650423526763916, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8726522326469421, + "num_tokens": 92463450.0, + "step": 2376 + }, + { + "epoch": 0.3023788322096425, + "grad_norm": 2.5325851440429688, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8783698081970215, + "num_tokens": 92505374.0, + "step": 2377 + }, + { + "epoch": 0.30250604248823304, + "grad_norm": 2.7904906272888184, + "learning_rate": 1e-06, + "loss": 0.387, + "mean_token_accuracy": 0.8677110075950623, + "num_tokens": 92544396.0, + "step": 2378 + }, + { + "epoch": 0.30263325276682357, + "grad_norm": 2.987346649169922, + "learning_rate": 1e-06, + "loss": 0.4436, + "mean_token_accuracy": 0.8458865880966187, + "num_tokens": 92580324.0, + "step": 2379 + }, + { + "epoch": 0.3027604630454141, + "grad_norm": 2.70670747756958, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8743817806243896, + "num_tokens": 92621174.0, + "step": 2380 + }, + { + "epoch": 0.30288767332400457, + "grad_norm": 2.7577834129333496, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8704323768615723, + "num_tokens": 92663548.0, + "step": 2381 + }, + { + "epoch": 0.3030148836025951, + "grad_norm": 3.288037061691284, + "learning_rate": 1e-06, + "loss": 0.3907, + "mean_token_accuracy": 0.862136721611023, + "num_tokens": 92696389.0, + "step": 2382 + }, + { + "epoch": 0.3031420938811856, + "grad_norm": 2.856981039047241, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.8625707626342773, + "num_tokens": 92734943.0, + "step": 2383 + }, + { + "epoch": 0.3032693041597761, + "grad_norm": 2.79591965675354, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8798015713691711, + "num_tokens": 92768888.0, + "step": 2384 + }, + { + "epoch": 0.3033965144383666, + "grad_norm": 2.7759788036346436, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8648368716239929, + "num_tokens": 92807236.0, + "step": 2385 + }, + { + "epoch": 0.30352372471695716, + "grad_norm": 2.843479633331299, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8711614012718201, + "num_tokens": 92845738.0, + "step": 2386 + }, + { + "epoch": 0.30365093499554763, + "grad_norm": 2.808528423309326, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8737161755561829, + "num_tokens": 92887972.0, + "step": 2387 + }, + { + "epoch": 0.30377814527413816, + "grad_norm": 2.883279323577881, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.8717001676559448, + "num_tokens": 92923200.0, + "step": 2388 + }, + { + "epoch": 0.3039053555527287, + "grad_norm": 2.6251883506774902, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8850962519645691, + "num_tokens": 92960581.0, + "step": 2389 + }, + { + "epoch": 0.30403256583131916, + "grad_norm": 2.9846413135528564, + "learning_rate": 1e-06, + "loss": 0.4195, + "mean_token_accuracy": 0.8545437455177307, + "num_tokens": 92998979.0, + "step": 2390 + }, + { + "epoch": 0.3041597761099097, + "grad_norm": 3.5255374908447266, + "learning_rate": 1e-06, + "loss": 0.4307, + "mean_token_accuracy": 0.8500379323959351, + "num_tokens": 93028559.0, + "step": 2391 + }, + { + "epoch": 0.3042869863885002, + "grad_norm": 3.0664310455322266, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.8736716508865356, + "num_tokens": 93060456.0, + "step": 2392 + }, + { + "epoch": 0.3044141966670907, + "grad_norm": 2.837190628051758, + "learning_rate": 1e-06, + "loss": 0.4053, + "mean_token_accuracy": 0.8644588589668274, + "num_tokens": 93099450.0, + "step": 2393 + }, + { + "epoch": 0.3045414069456812, + "grad_norm": 2.8591582775115967, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8678141236305237, + "num_tokens": 93138845.0, + "step": 2394 + }, + { + "epoch": 0.30466861722427174, + "grad_norm": 2.6715481281280518, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8704079389572144, + "num_tokens": 93182249.0, + "step": 2395 + }, + { + "epoch": 0.3047958275028622, + "grad_norm": 2.6969475746154785, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8709934949874878, + "num_tokens": 93224203.0, + "step": 2396 + }, + { + "epoch": 0.30492303778145274, + "grad_norm": 2.6304919719696045, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.8744851350784302, + "num_tokens": 93266575.0, + "step": 2397 + }, + { + "epoch": 0.30505024806004327, + "grad_norm": 2.806765556335449, + "learning_rate": 1e-06, + "loss": 0.4044, + "mean_token_accuracy": 0.8598151206970215, + "num_tokens": 93306755.0, + "step": 2398 + }, + { + "epoch": 0.30517745833863374, + "grad_norm": 2.889547109603882, + "learning_rate": 1e-06, + "loss": 0.422, + "mean_token_accuracy": 0.857994794845581, + "num_tokens": 93345345.0, + "step": 2399 + }, + { + "epoch": 0.3053046686172243, + "grad_norm": 2.6103272438049316, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8690163493156433, + "num_tokens": 93386520.0, + "step": 2400 + }, + { + "epoch": 0.3054318788958148, + "grad_norm": 2.9278476238250732, + "learning_rate": 1e-06, + "loss": 0.4029, + "mean_token_accuracy": 0.8643908500671387, + "num_tokens": 93422699.0, + "step": 2401 + }, + { + "epoch": 0.3055590891744053, + "grad_norm": 2.913856029510498, + "learning_rate": 1e-06, + "loss": 0.393, + "mean_token_accuracy": 0.8664682507514954, + "num_tokens": 93462326.0, + "step": 2402 + }, + { + "epoch": 0.3056862994529958, + "grad_norm": 2.823436975479126, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.866790771484375, + "num_tokens": 93499289.0, + "step": 2403 + }, + { + "epoch": 0.30581350973158633, + "grad_norm": 2.8538691997528076, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8715939521789551, + "num_tokens": 93535592.0, + "step": 2404 + }, + { + "epoch": 0.3059407200101768, + "grad_norm": 2.65781569480896, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8757572770118713, + "num_tokens": 93575017.0, + "step": 2405 + }, + { + "epoch": 0.30606793028876733, + "grad_norm": 2.5274269580841064, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8676412105560303, + "num_tokens": 93617554.0, + "step": 2406 + }, + { + "epoch": 0.30619514056735786, + "grad_norm": 2.6627273559570312, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8811402320861816, + "num_tokens": 93657225.0, + "step": 2407 + }, + { + "epoch": 0.30632235084594833, + "grad_norm": 2.899547815322876, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8670939207077026, + "num_tokens": 93692119.0, + "step": 2408 + }, + { + "epoch": 0.30644956112453886, + "grad_norm": 2.8594651222229004, + "learning_rate": 1e-06, + "loss": 0.4184, + "mean_token_accuracy": 0.8566087484359741, + "num_tokens": 93728212.0, + "step": 2409 + }, + { + "epoch": 0.3065767714031294, + "grad_norm": 2.833116292953491, + "learning_rate": 1e-06, + "loss": 0.422, + "mean_token_accuracy": 0.8543214797973633, + "num_tokens": 93766420.0, + "step": 2410 + }, + { + "epoch": 0.30670398168171986, + "grad_norm": 2.9561989307403564, + "learning_rate": 1e-06, + "loss": 0.3886, + "mean_token_accuracy": 0.8651460409164429, + "num_tokens": 93801439.0, + "step": 2411 + }, + { + "epoch": 0.3068311919603104, + "grad_norm": 2.8476736545562744, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8646553158760071, + "num_tokens": 93837326.0, + "step": 2412 + }, + { + "epoch": 0.3069584022389009, + "grad_norm": 2.572298765182495, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8709917664527893, + "num_tokens": 93883180.0, + "step": 2413 + }, + { + "epoch": 0.3070856125174914, + "grad_norm": 2.9227089881896973, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8738962411880493, + "num_tokens": 93917317.0, + "step": 2414 + }, + { + "epoch": 0.3072128227960819, + "grad_norm": 2.8047938346862793, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8700823783874512, + "num_tokens": 93955376.0, + "step": 2415 + }, + { + "epoch": 0.30734003307467245, + "grad_norm": 2.807767152786255, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8766098022460938, + "num_tokens": 93990625.0, + "step": 2416 + }, + { + "epoch": 0.3074672433532629, + "grad_norm": 2.678124189376831, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8762457370758057, + "num_tokens": 94032694.0, + "step": 2417 + }, + { + "epoch": 0.30759445363185345, + "grad_norm": 2.7875828742980957, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.8593795299530029, + "num_tokens": 94072481.0, + "step": 2418 + }, + { + "epoch": 0.307721663910444, + "grad_norm": 2.508636951446533, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8879914283752441, + "num_tokens": 94117218.0, + "step": 2419 + }, + { + "epoch": 0.30784887418903445, + "grad_norm": 2.9714343547821045, + "learning_rate": 1e-06, + "loss": 0.4078, + "mean_token_accuracy": 0.8621170520782471, + "num_tokens": 94153966.0, + "step": 2420 + }, + { + "epoch": 0.307976084467625, + "grad_norm": 2.875844955444336, + "learning_rate": 1e-06, + "loss": 0.4396, + "mean_token_accuracy": 0.8526939153671265, + "num_tokens": 94195430.0, + "step": 2421 + }, + { + "epoch": 0.3081032947462155, + "grad_norm": 2.772979974746704, + "learning_rate": 1e-06, + "loss": 0.4519, + "mean_token_accuracy": 0.8456584215164185, + "num_tokens": 94235840.0, + "step": 2422 + }, + { + "epoch": 0.308230505024806, + "grad_norm": 3.2653017044067383, + "learning_rate": 1e-06, + "loss": 0.4357, + "mean_token_accuracy": 0.851370096206665, + "num_tokens": 94267280.0, + "step": 2423 + }, + { + "epoch": 0.3083577153033965, + "grad_norm": 2.9722602367401123, + "learning_rate": 1e-06, + "loss": 0.3907, + "mean_token_accuracy": 0.867517352104187, + "num_tokens": 94299604.0, + "step": 2424 + }, + { + "epoch": 0.30848492558198704, + "grad_norm": 2.852935552597046, + "learning_rate": 1e-06, + "loss": 0.4355, + "mean_token_accuracy": 0.8513518571853638, + "num_tokens": 94340501.0, + "step": 2425 + }, + { + "epoch": 0.3086121358605775, + "grad_norm": 2.6301302909851074, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.869072675704956, + "num_tokens": 94382917.0, + "step": 2426 + }, + { + "epoch": 0.30873934613916804, + "grad_norm": 2.896770715713501, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8670793175697327, + "num_tokens": 94417437.0, + "step": 2427 + }, + { + "epoch": 0.30886655641775856, + "grad_norm": 2.9172329902648926, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.8651623725891113, + "num_tokens": 94452246.0, + "step": 2428 + }, + { + "epoch": 0.3089937666963491, + "grad_norm": 2.706143617630005, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.863379180431366, + "num_tokens": 94500071.0, + "step": 2429 + }, + { + "epoch": 0.30912097697493957, + "grad_norm": 2.8770768642425537, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8793853521347046, + "num_tokens": 94532282.0, + "step": 2430 + }, + { + "epoch": 0.3092481872535301, + "grad_norm": 2.800621509552002, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.8672631978988647, + "num_tokens": 94571700.0, + "step": 2431 + }, + { + "epoch": 0.3093753975321206, + "grad_norm": 2.993791103363037, + "learning_rate": 1e-06, + "loss": 0.4286, + "mean_token_accuracy": 0.8573663234710693, + "num_tokens": 94606011.0, + "step": 2432 + }, + { + "epoch": 0.3095026078107111, + "grad_norm": 2.7006068229675293, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8705911636352539, + "num_tokens": 94644451.0, + "step": 2433 + }, + { + "epoch": 0.3096298180893016, + "grad_norm": 2.6575427055358887, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8746918439865112, + "num_tokens": 94682117.0, + "step": 2434 + }, + { + "epoch": 0.30975702836789215, + "grad_norm": 2.850050687789917, + "learning_rate": 1e-06, + "loss": 0.3913, + "mean_token_accuracy": 0.8656077980995178, + "num_tokens": 94722150.0, + "step": 2435 + }, + { + "epoch": 0.3098842386464826, + "grad_norm": 2.991039276123047, + "learning_rate": 1e-06, + "loss": 0.4316, + "mean_token_accuracy": 0.8511991500854492, + "num_tokens": 94758126.0, + "step": 2436 + }, + { + "epoch": 0.31001144892507315, + "grad_norm": 2.757878303527832, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8683370351791382, + "num_tokens": 94796704.0, + "step": 2437 + }, + { + "epoch": 0.3101386592036637, + "grad_norm": 2.8028695583343506, + "learning_rate": 1e-06, + "loss": 0.4, + "mean_token_accuracy": 0.8660423755645752, + "num_tokens": 94835285.0, + "step": 2438 + }, + { + "epoch": 0.31026586948225415, + "grad_norm": 2.8874640464782715, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8745115399360657, + "num_tokens": 94870648.0, + "step": 2439 + }, + { + "epoch": 0.3103930797608447, + "grad_norm": 2.7694616317749023, + "learning_rate": 1e-06, + "loss": 0.4177, + "mean_token_accuracy": 0.8613091707229614, + "num_tokens": 94914244.0, + "step": 2440 + }, + { + "epoch": 0.3105202900394352, + "grad_norm": 2.7408740520477295, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8823232650756836, + "num_tokens": 94950960.0, + "step": 2441 + }, + { + "epoch": 0.3106475003180257, + "grad_norm": 2.629188060760498, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8724942803382874, + "num_tokens": 94993266.0, + "step": 2442 + }, + { + "epoch": 0.3107747105966162, + "grad_norm": 2.6080777645111084, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8808223605155945, + "num_tokens": 95029539.0, + "step": 2443 + }, + { + "epoch": 0.31090192087520674, + "grad_norm": 2.5953867435455322, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8694105744361877, + "num_tokens": 95071836.0, + "step": 2444 + }, + { + "epoch": 0.3110291311537972, + "grad_norm": 2.870555877685547, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.870356559753418, + "num_tokens": 95106882.0, + "step": 2445 + }, + { + "epoch": 0.31115634143238774, + "grad_norm": 2.842164993286133, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8671977519989014, + "num_tokens": 95145079.0, + "step": 2446 + }, + { + "epoch": 0.31128355171097827, + "grad_norm": 2.6479766368865967, + "learning_rate": 1e-06, + "loss": 0.4139, + "mean_token_accuracy": 0.8614822626113892, + "num_tokens": 95187942.0, + "step": 2447 + }, + { + "epoch": 0.31141076198956874, + "grad_norm": 2.5364632606506348, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8747096657752991, + "num_tokens": 95233356.0, + "step": 2448 + }, + { + "epoch": 0.31153797226815927, + "grad_norm": 2.4849750995635986, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8754032850265503, + "num_tokens": 95278973.0, + "step": 2449 + }, + { + "epoch": 0.3116651825467498, + "grad_norm": 2.8713669776916504, + "learning_rate": 1e-06, + "loss": 0.4173, + "mean_token_accuracy": 0.8643466234207153, + "num_tokens": 95318461.0, + "step": 2450 + }, + { + "epoch": 0.31179239282534027, + "grad_norm": 2.812445878982544, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8659434914588928, + "num_tokens": 95355541.0, + "step": 2451 + }, + { + "epoch": 0.3119196031039308, + "grad_norm": 2.768425941467285, + "learning_rate": 1e-06, + "loss": 0.4119, + "mean_token_accuracy": 0.8643842935562134, + "num_tokens": 95397028.0, + "step": 2452 + }, + { + "epoch": 0.3120468133825213, + "grad_norm": 2.75663685798645, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8711445927619934, + "num_tokens": 95433272.0, + "step": 2453 + }, + { + "epoch": 0.3121740236611118, + "grad_norm": 2.7714755535125732, + "learning_rate": 1e-06, + "loss": 0.4112, + "mean_token_accuracy": 0.863620400428772, + "num_tokens": 95474877.0, + "step": 2454 + }, + { + "epoch": 0.3123012339397023, + "grad_norm": 3.1222078800201416, + "learning_rate": 1e-06, + "loss": 0.4017, + "mean_token_accuracy": 0.8633798956871033, + "num_tokens": 95510143.0, + "step": 2455 + }, + { + "epoch": 0.31242844421829286, + "grad_norm": 2.513326406478882, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8936888575553894, + "num_tokens": 95549540.0, + "step": 2456 + }, + { + "epoch": 0.31255565449688333, + "grad_norm": 2.5528817176818848, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.8716424107551575, + "num_tokens": 95592979.0, + "step": 2457 + }, + { + "epoch": 0.31268286477547386, + "grad_norm": 2.729308605194092, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8745574951171875, + "num_tokens": 95633096.0, + "step": 2458 + }, + { + "epoch": 0.3128100750540644, + "grad_norm": 2.557861328125, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8707829713821411, + "num_tokens": 95677952.0, + "step": 2459 + }, + { + "epoch": 0.31293728533265486, + "grad_norm": 2.5073955059051514, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8685060143470764, + "num_tokens": 95721308.0, + "step": 2460 + }, + { + "epoch": 0.3130644956112454, + "grad_norm": 3.001436948776245, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8692230582237244, + "num_tokens": 95756738.0, + "step": 2461 + }, + { + "epoch": 0.3131917058898359, + "grad_norm": 2.793936014175415, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.8657182455062866, + "num_tokens": 95795475.0, + "step": 2462 + }, + { + "epoch": 0.3133189161684264, + "grad_norm": 2.9467036724090576, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8698466420173645, + "num_tokens": 95829215.0, + "step": 2463 + }, + { + "epoch": 0.3134461264470169, + "grad_norm": 3.118074893951416, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8652028441429138, + "num_tokens": 95861939.0, + "step": 2464 + }, + { + "epoch": 0.31357333672560744, + "grad_norm": 2.669464111328125, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8674068450927734, + "num_tokens": 95907147.0, + "step": 2465 + }, + { + "epoch": 0.3137005470041979, + "grad_norm": 2.9490323066711426, + "learning_rate": 1e-06, + "loss": 0.4367, + "mean_token_accuracy": 0.8549654483795166, + "num_tokens": 95945530.0, + "step": 2466 + }, + { + "epoch": 0.31382775728278844, + "grad_norm": 3.0888333320617676, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8775269985198975, + "num_tokens": 95980862.0, + "step": 2467 + }, + { + "epoch": 0.313954967561379, + "grad_norm": 2.756612539291382, + "learning_rate": 1e-06, + "loss": 0.4057, + "mean_token_accuracy": 0.8618004322052002, + "num_tokens": 96022698.0, + "step": 2468 + }, + { + "epoch": 0.31408217783996945, + "grad_norm": 2.489513397216797, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8742263317108154, + "num_tokens": 96066034.0, + "step": 2469 + }, + { + "epoch": 0.31420938811856, + "grad_norm": 2.720383882522583, + "learning_rate": 1e-06, + "loss": 0.4352, + "mean_token_accuracy": 0.8527101278305054, + "num_tokens": 96111794.0, + "step": 2470 + }, + { + "epoch": 0.3143365983971505, + "grad_norm": 2.5793535709381104, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8810006380081177, + "num_tokens": 96151504.0, + "step": 2471 + }, + { + "epoch": 0.314463808675741, + "grad_norm": 2.543321132659912, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.871074378490448, + "num_tokens": 96195878.0, + "step": 2472 + }, + { + "epoch": 0.3145910189543315, + "grad_norm": 2.691667079925537, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8743379712104797, + "num_tokens": 96235030.0, + "step": 2473 + }, + { + "epoch": 0.31471822923292203, + "grad_norm": 2.778557777404785, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8704768419265747, + "num_tokens": 96272723.0, + "step": 2474 + }, + { + "epoch": 0.3148454395115125, + "grad_norm": 2.8047351837158203, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8669875264167786, + "num_tokens": 96308759.0, + "step": 2475 + }, + { + "epoch": 0.31497264979010303, + "grad_norm": 2.430840253829956, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.8672472834587097, + "num_tokens": 96358180.0, + "step": 2476 + }, + { + "epoch": 0.31509986006869356, + "grad_norm": 2.504155158996582, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8701719641685486, + "num_tokens": 96404635.0, + "step": 2477 + }, + { + "epoch": 0.31522707034728403, + "grad_norm": 2.6489670276641846, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8729525804519653, + "num_tokens": 96447540.0, + "step": 2478 + }, + { + "epoch": 0.31535428062587456, + "grad_norm": 2.6552066802978516, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8686469197273254, + "num_tokens": 96488193.0, + "step": 2479 + }, + { + "epoch": 0.3154814909044651, + "grad_norm": 2.7128684520721436, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8714674711227417, + "num_tokens": 96526268.0, + "step": 2480 + }, + { + "epoch": 0.3156087011830556, + "grad_norm": 2.7359867095947266, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8661690354347229, + "num_tokens": 96565757.0, + "step": 2481 + }, + { + "epoch": 0.3157359114616461, + "grad_norm": 2.7265734672546387, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.874597430229187, + "num_tokens": 96607721.0, + "step": 2482 + }, + { + "epoch": 0.3158631217402366, + "grad_norm": 2.899097204208374, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8766897320747375, + "num_tokens": 96644376.0, + "step": 2483 + }, + { + "epoch": 0.31599033201882715, + "grad_norm": 2.6527020931243896, + "learning_rate": 1e-06, + "loss": 0.3935, + "mean_token_accuracy": 0.8652709722518921, + "num_tokens": 96685194.0, + "step": 2484 + }, + { + "epoch": 0.3161175422974176, + "grad_norm": 2.843153238296509, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8790479302406311, + "num_tokens": 96716790.0, + "step": 2485 + }, + { + "epoch": 0.31624475257600815, + "grad_norm": 2.6154892444610596, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8730746507644653, + "num_tokens": 96759915.0, + "step": 2486 + }, + { + "epoch": 0.3163719628545987, + "grad_norm": 2.6400864124298096, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.8676813244819641, + "num_tokens": 96801294.0, + "step": 2487 + }, + { + "epoch": 0.31649917313318915, + "grad_norm": 2.8384761810302734, + "learning_rate": 1e-06, + "loss": 0.4025, + "mean_token_accuracy": 0.8571317195892334, + "num_tokens": 96838453.0, + "step": 2488 + }, + { + "epoch": 0.3166263834117797, + "grad_norm": 2.949064016342163, + "learning_rate": 1e-06, + "loss": 0.4021, + "mean_token_accuracy": 0.8627256155014038, + "num_tokens": 96871728.0, + "step": 2489 + }, + { + "epoch": 0.3167535936903702, + "grad_norm": 2.6719775199890137, + "learning_rate": 1e-06, + "loss": 0.3942, + "mean_token_accuracy": 0.8668162822723389, + "num_tokens": 96913832.0, + "step": 2490 + }, + { + "epoch": 0.3168808039689607, + "grad_norm": 2.8214311599731445, + "learning_rate": 1e-06, + "loss": 0.4002, + "mean_token_accuracy": 0.8629467487335205, + "num_tokens": 96950054.0, + "step": 2491 + }, + { + "epoch": 0.3170080142475512, + "grad_norm": 2.87636661529541, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8629648685455322, + "num_tokens": 96988063.0, + "step": 2492 + }, + { + "epoch": 0.31713522452614173, + "grad_norm": 3.0017004013061523, + "learning_rate": 1e-06, + "loss": 0.4341, + "mean_token_accuracy": 0.8562490940093994, + "num_tokens": 97025163.0, + "step": 2493 + }, + { + "epoch": 0.3172624348047322, + "grad_norm": 2.7912392616271973, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8702528476715088, + "num_tokens": 97064890.0, + "step": 2494 + }, + { + "epoch": 0.31738964508332274, + "grad_norm": 2.7987916469573975, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8642077445983887, + "num_tokens": 97102455.0, + "step": 2495 + }, + { + "epoch": 0.31751685536191326, + "grad_norm": 2.5934908390045166, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8885826468467712, + "num_tokens": 97139665.0, + "step": 2496 + }, + { + "epoch": 0.31764406564050374, + "grad_norm": 2.7668163776397705, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8714463710784912, + "num_tokens": 97180303.0, + "step": 2497 + }, + { + "epoch": 0.31777127591909426, + "grad_norm": 2.7937517166137695, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8700060248374939, + "num_tokens": 97216141.0, + "step": 2498 + }, + { + "epoch": 0.3178984861976848, + "grad_norm": 2.875917673110962, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8811472654342651, + "num_tokens": 97251293.0, + "step": 2499 + }, + { + "epoch": 0.31802569647627527, + "grad_norm": 3.0418922901153564, + "learning_rate": 1e-06, + "loss": 0.4158, + "mean_token_accuracy": 0.8608131408691406, + "num_tokens": 97288758.0, + "step": 2500 + }, + { + "epoch": 0.3181529067548658, + "grad_norm": 2.799614429473877, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.867209255695343, + "num_tokens": 97327221.0, + "step": 2501 + }, + { + "epoch": 0.3182801170334563, + "grad_norm": 2.742499351501465, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.866546630859375, + "num_tokens": 97366618.0, + "step": 2502 + }, + { + "epoch": 0.3184073273120468, + "grad_norm": 2.837960958480835, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8707809448242188, + "num_tokens": 97401776.0, + "step": 2503 + }, + { + "epoch": 0.3185345375906373, + "grad_norm": 3.135411500930786, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8617076873779297, + "num_tokens": 97434060.0, + "step": 2504 + }, + { + "epoch": 0.31866174786922785, + "grad_norm": 2.7625558376312256, + "learning_rate": 1e-06, + "loss": 0.4307, + "mean_token_accuracy": 0.8510690927505493, + "num_tokens": 97478019.0, + "step": 2505 + }, + { + "epoch": 0.3187889581478183, + "grad_norm": 2.7018587589263916, + "learning_rate": 1e-06, + "loss": 0.4148, + "mean_token_accuracy": 0.860713005065918, + "num_tokens": 97522503.0, + "step": 2506 + }, + { + "epoch": 0.31891616842640885, + "grad_norm": 2.8830690383911133, + "learning_rate": 1e-06, + "loss": 0.4112, + "mean_token_accuracy": 0.859481692314148, + "num_tokens": 97559956.0, + "step": 2507 + }, + { + "epoch": 0.3190433787049994, + "grad_norm": 2.747351884841919, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8700471520423889, + "num_tokens": 97599478.0, + "step": 2508 + }, + { + "epoch": 0.31917058898358985, + "grad_norm": 2.8039982318878174, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8778149485588074, + "num_tokens": 97640385.0, + "step": 2509 + }, + { + "epoch": 0.3192977992621804, + "grad_norm": 2.764360189437866, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.8684303164482117, + "num_tokens": 97678926.0, + "step": 2510 + }, + { + "epoch": 0.3194250095407709, + "grad_norm": 2.764056444168091, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8828518390655518, + "num_tokens": 97716011.0, + "step": 2511 + }, + { + "epoch": 0.3195522198193614, + "grad_norm": 2.610717535018921, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8697122931480408, + "num_tokens": 97761210.0, + "step": 2512 + }, + { + "epoch": 0.3196794300979519, + "grad_norm": 2.6814022064208984, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.887639045715332, + "num_tokens": 97797382.0, + "step": 2513 + }, + { + "epoch": 0.31980664037654244, + "grad_norm": 2.6209194660186768, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8786876797676086, + "num_tokens": 97837424.0, + "step": 2514 + }, + { + "epoch": 0.3199338506551329, + "grad_norm": 2.827014923095703, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8741649389266968, + "num_tokens": 97879029.0, + "step": 2515 + }, + { + "epoch": 0.32006106093372344, + "grad_norm": 2.8321032524108887, + "learning_rate": 1e-06, + "loss": 0.4144, + "mean_token_accuracy": 0.8566957712173462, + "num_tokens": 97917655.0, + "step": 2516 + }, + { + "epoch": 0.32018827121231397, + "grad_norm": 2.813511610031128, + "learning_rate": 1e-06, + "loss": 0.4048, + "mean_token_accuracy": 0.8631609678268433, + "num_tokens": 97956627.0, + "step": 2517 + }, + { + "epoch": 0.32031548149090444, + "grad_norm": 3.092038631439209, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8743257522583008, + "num_tokens": 97986888.0, + "step": 2518 + }, + { + "epoch": 0.32044269176949497, + "grad_norm": 2.9180474281311035, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.865860104560852, + "num_tokens": 98024628.0, + "step": 2519 + }, + { + "epoch": 0.3205699020480855, + "grad_norm": 2.625171184539795, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8680199980735779, + "num_tokens": 98069050.0, + "step": 2520 + }, + { + "epoch": 0.32069711232667597, + "grad_norm": 2.888751983642578, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8645603656768799, + "num_tokens": 98105086.0, + "step": 2521 + }, + { + "epoch": 0.3208243226052665, + "grad_norm": 2.5634403228759766, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8856780529022217, + "num_tokens": 98144614.0, + "step": 2522 + }, + { + "epoch": 0.320951532883857, + "grad_norm": 2.7234103679656982, + "learning_rate": 1e-06, + "loss": 0.4159, + "mean_token_accuracy": 0.8601298332214355, + "num_tokens": 98187345.0, + "step": 2523 + }, + { + "epoch": 0.3210787431624475, + "grad_norm": 2.7395949363708496, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8696278929710388, + "num_tokens": 98224656.0, + "step": 2524 + }, + { + "epoch": 0.32120595344103803, + "grad_norm": 2.814284563064575, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8707417845726013, + "num_tokens": 98260913.0, + "step": 2525 + }, + { + "epoch": 0.32133316371962856, + "grad_norm": 2.8102095127105713, + "learning_rate": 1e-06, + "loss": 0.4242, + "mean_token_accuracy": 0.8575136661529541, + "num_tokens": 98301161.0, + "step": 2526 + }, + { + "epoch": 0.32146037399821903, + "grad_norm": 2.7540054321289062, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8877068758010864, + "num_tokens": 98336955.0, + "step": 2527 + }, + { + "epoch": 0.32158758427680956, + "grad_norm": 2.9121031761169434, + "learning_rate": 1e-06, + "loss": 0.4158, + "mean_token_accuracy": 0.8575220704078674, + "num_tokens": 98374524.0, + "step": 2528 + }, + { + "epoch": 0.3217147945554001, + "grad_norm": 3.011953830718994, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8705959320068359, + "num_tokens": 98412241.0, + "step": 2529 + }, + { + "epoch": 0.3218420048339906, + "grad_norm": 2.9939725399017334, + "learning_rate": 1e-06, + "loss": 0.4379, + "mean_token_accuracy": 0.8484634757041931, + "num_tokens": 98447734.0, + "step": 2530 + }, + { + "epoch": 0.3219692151125811, + "grad_norm": 2.901174783706665, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8654455542564392, + "num_tokens": 98485801.0, + "step": 2531 + }, + { + "epoch": 0.3220964253911716, + "grad_norm": 2.497652053833008, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8809295892715454, + "num_tokens": 98529915.0, + "step": 2532 + }, + { + "epoch": 0.32222363566976214, + "grad_norm": 2.582228899002075, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8860191106796265, + "num_tokens": 98573652.0, + "step": 2533 + }, + { + "epoch": 0.3223508459483526, + "grad_norm": 3.1081247329711914, + "learning_rate": 1e-06, + "loss": 0.416, + "mean_token_accuracy": 0.8560875654220581, + "num_tokens": 98608078.0, + "step": 2534 + }, + { + "epoch": 0.32247805622694314, + "grad_norm": 2.652085304260254, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8743720054626465, + "num_tokens": 98650818.0, + "step": 2535 + }, + { + "epoch": 0.32260526650553367, + "grad_norm": 2.7051773071289062, + "learning_rate": 1e-06, + "loss": 0.4011, + "mean_token_accuracy": 0.8634435534477234, + "num_tokens": 98691946.0, + "step": 2536 + }, + { + "epoch": 0.32273247678412414, + "grad_norm": 2.7664852142333984, + "learning_rate": 1e-06, + "loss": 0.4173, + "mean_token_accuracy": 0.8599851727485657, + "num_tokens": 98732758.0, + "step": 2537 + }, + { + "epoch": 0.3228596870627147, + "grad_norm": 2.537302017211914, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8804741501808167, + "num_tokens": 98771685.0, + "step": 2538 + }, + { + "epoch": 0.3229868973413052, + "grad_norm": 2.860081434249878, + "learning_rate": 1e-06, + "loss": 0.397, + "mean_token_accuracy": 0.8661824464797974, + "num_tokens": 98810487.0, + "step": 2539 + }, + { + "epoch": 0.3231141076198957, + "grad_norm": 2.5487759113311768, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8742889761924744, + "num_tokens": 98854162.0, + "step": 2540 + }, + { + "epoch": 0.3232413178984862, + "grad_norm": 2.8031668663024902, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.8572912812232971, + "num_tokens": 98891527.0, + "step": 2541 + }, + { + "epoch": 0.32336852817707673, + "grad_norm": 2.719862937927246, + "learning_rate": 1e-06, + "loss": 0.4163, + "mean_token_accuracy": 0.855243444442749, + "num_tokens": 98934154.0, + "step": 2542 + }, + { + "epoch": 0.3234957384556672, + "grad_norm": 2.726579189300537, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8756121397018433, + "num_tokens": 98970173.0, + "step": 2543 + }, + { + "epoch": 0.32362294873425773, + "grad_norm": 2.6595685482025146, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8845392465591431, + "num_tokens": 99011066.0, + "step": 2544 + }, + { + "epoch": 0.32375015901284826, + "grad_norm": 2.748197078704834, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8669312000274658, + "num_tokens": 99050731.0, + "step": 2545 + }, + { + "epoch": 0.32387736929143873, + "grad_norm": 2.6235134601593018, + "learning_rate": 1e-06, + "loss": 0.3934, + "mean_token_accuracy": 0.865356981754303, + "num_tokens": 99095216.0, + "step": 2546 + }, + { + "epoch": 0.32400457957002926, + "grad_norm": 2.9971635341644287, + "learning_rate": 1e-06, + "loss": 0.4665, + "mean_token_accuracy": 0.8435901403427124, + "num_tokens": 99134701.0, + "step": 2547 + }, + { + "epoch": 0.3241317898486198, + "grad_norm": 2.650416612625122, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8765721321105957, + "num_tokens": 99173362.0, + "step": 2548 + }, + { + "epoch": 0.32425900012721026, + "grad_norm": 2.7473950386047363, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8757829666137695, + "num_tokens": 99214917.0, + "step": 2549 + }, + { + "epoch": 0.3243862104058008, + "grad_norm": 2.7405943870544434, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8796244859695435, + "num_tokens": 99254472.0, + "step": 2550 + }, + { + "epoch": 0.3245134206843913, + "grad_norm": 2.635627508163452, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.891350507736206, + "num_tokens": 99289504.0, + "step": 2551 + }, + { + "epoch": 0.3246406309629818, + "grad_norm": 2.591562032699585, + "learning_rate": 1e-06, + "loss": 0.4121, + "mean_token_accuracy": 0.8586758971214294, + "num_tokens": 99336547.0, + "step": 2552 + }, + { + "epoch": 0.3247678412415723, + "grad_norm": 2.5419373512268066, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8705364465713501, + "num_tokens": 99381621.0, + "step": 2553 + }, + { + "epoch": 0.32489505152016285, + "grad_norm": 2.8011107444763184, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8692716360092163, + "num_tokens": 99421460.0, + "step": 2554 + }, + { + "epoch": 0.3250222617987533, + "grad_norm": 2.882941961288452, + "learning_rate": 1e-06, + "loss": 0.4261, + "mean_token_accuracy": 0.8549572229385376, + "num_tokens": 99459148.0, + "step": 2555 + }, + { + "epoch": 0.32514947207734385, + "grad_norm": 2.795734167098999, + "learning_rate": 1e-06, + "loss": 0.4044, + "mean_token_accuracy": 0.8608903884887695, + "num_tokens": 99497231.0, + "step": 2556 + }, + { + "epoch": 0.3252766823559344, + "grad_norm": 2.8402316570281982, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.874355673789978, + "num_tokens": 99531929.0, + "step": 2557 + }, + { + "epoch": 0.32540389263452485, + "grad_norm": 2.8785006999969482, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8751236200332642, + "num_tokens": 99566237.0, + "step": 2558 + }, + { + "epoch": 0.3255311029131154, + "grad_norm": 2.7278478145599365, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8823996782302856, + "num_tokens": 99606253.0, + "step": 2559 + }, + { + "epoch": 0.3256583131917059, + "grad_norm": 3.0605270862579346, + "learning_rate": 1e-06, + "loss": 0.4149, + "mean_token_accuracy": 0.8588341474533081, + "num_tokens": 99641246.0, + "step": 2560 + }, + { + "epoch": 0.3257855234702964, + "grad_norm": 2.793302059173584, + "learning_rate": 1e-06, + "loss": 0.4154, + "mean_token_accuracy": 0.8571199774742126, + "num_tokens": 99684105.0, + "step": 2561 + }, + { + "epoch": 0.3259127337488869, + "grad_norm": 2.6220343112945557, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.881823718547821, + "num_tokens": 99724999.0, + "step": 2562 + }, + { + "epoch": 0.32603994402747744, + "grad_norm": 3.082887887954712, + "learning_rate": 1e-06, + "loss": 0.4116, + "mean_token_accuracy": 0.860184371471405, + "num_tokens": 99760120.0, + "step": 2563 + }, + { + "epoch": 0.3261671543060679, + "grad_norm": 2.917253017425537, + "learning_rate": 1e-06, + "loss": 0.4214, + "mean_token_accuracy": 0.8566180467605591, + "num_tokens": 99799884.0, + "step": 2564 + }, + { + "epoch": 0.32629436458465844, + "grad_norm": 2.759697198867798, + "learning_rate": 1e-06, + "loss": 0.4184, + "mean_token_accuracy": 0.8645093441009521, + "num_tokens": 99841567.0, + "step": 2565 + }, + { + "epoch": 0.32642157486324896, + "grad_norm": 2.566455125808716, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8787239789962769, + "num_tokens": 99887086.0, + "step": 2566 + }, + { + "epoch": 0.32654878514183944, + "grad_norm": 2.659766435623169, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8731527328491211, + "num_tokens": 99925425.0, + "step": 2567 + }, + { + "epoch": 0.32667599542042997, + "grad_norm": 2.6757054328918457, + "learning_rate": 1e-06, + "loss": 0.4619, + "mean_token_accuracy": 0.8481234312057495, + "num_tokens": 99972243.0, + "step": 2568 + }, + { + "epoch": 0.3268032056990205, + "grad_norm": 2.8276007175445557, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8812775611877441, + "num_tokens": 100006864.0, + "step": 2569 + }, + { + "epoch": 0.32693041597761097, + "grad_norm": 2.6999216079711914, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8800367712974548, + "num_tokens": 100043680.0, + "step": 2570 + }, + { + "epoch": 0.3270576262562015, + "grad_norm": 2.8616552352905273, + "learning_rate": 1e-06, + "loss": 0.3982, + "mean_token_accuracy": 0.86316978931427, + "num_tokens": 100081475.0, + "step": 2571 + }, + { + "epoch": 0.327184836534792, + "grad_norm": 2.8869171142578125, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8683345913887024, + "num_tokens": 100114001.0, + "step": 2572 + }, + { + "epoch": 0.3273120468133825, + "grad_norm": 2.824990749359131, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8772625923156738, + "num_tokens": 100149426.0, + "step": 2573 + }, + { + "epoch": 0.327439257091973, + "grad_norm": 2.8596253395080566, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.872349202632904, + "num_tokens": 100185813.0, + "step": 2574 + }, + { + "epoch": 0.32756646737056355, + "grad_norm": 2.719008684158325, + "learning_rate": 1e-06, + "loss": 0.4111, + "mean_token_accuracy": 0.8630170822143555, + "num_tokens": 100223159.0, + "step": 2575 + }, + { + "epoch": 0.327693677649154, + "grad_norm": 2.7021501064300537, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.86391282081604, + "num_tokens": 100264323.0, + "step": 2576 + }, + { + "epoch": 0.32782088792774455, + "grad_norm": 2.7734694480895996, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.8655473589897156, + "num_tokens": 100301230.0, + "step": 2577 + }, + { + "epoch": 0.3279480982063351, + "grad_norm": 2.7440240383148193, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8659494519233704, + "num_tokens": 100344058.0, + "step": 2578 + }, + { + "epoch": 0.3280753084849256, + "grad_norm": 2.691863536834717, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8815696239471436, + "num_tokens": 100382646.0, + "step": 2579 + }, + { + "epoch": 0.3282025187635161, + "grad_norm": 3.2121329307556152, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.8676883578300476, + "num_tokens": 100412234.0, + "step": 2580 + }, + { + "epoch": 0.3283297290421066, + "grad_norm": 2.6698966026306152, + "learning_rate": 1e-06, + "loss": 0.3977, + "mean_token_accuracy": 0.8649232387542725, + "num_tokens": 100458055.0, + "step": 2581 + }, + { + "epoch": 0.32845693932069714, + "grad_norm": 2.7439119815826416, + "learning_rate": 1e-06, + "loss": 0.4184, + "mean_token_accuracy": 0.8583295345306396, + "num_tokens": 100499659.0, + "step": 2582 + }, + { + "epoch": 0.3285841495992876, + "grad_norm": 2.6888575553894043, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8695860505104065, + "num_tokens": 100539191.0, + "step": 2583 + }, + { + "epoch": 0.32871135987787814, + "grad_norm": 2.5896878242492676, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8847620487213135, + "num_tokens": 100576900.0, + "step": 2584 + }, + { + "epoch": 0.32883857015646867, + "grad_norm": 2.900982141494751, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8654152154922485, + "num_tokens": 100613608.0, + "step": 2585 + }, + { + "epoch": 0.32896578043505914, + "grad_norm": 3.0471510887145996, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8670498132705688, + "num_tokens": 100647163.0, + "step": 2586 + }, + { + "epoch": 0.32909299071364967, + "grad_norm": 2.519519090652466, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8758522272109985, + "num_tokens": 100689336.0, + "step": 2587 + }, + { + "epoch": 0.3292202009922402, + "grad_norm": 2.8084425926208496, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8691542744636536, + "num_tokens": 100725652.0, + "step": 2588 + }, + { + "epoch": 0.32934741127083067, + "grad_norm": 2.654343843460083, + "learning_rate": 1e-06, + "loss": 0.4286, + "mean_token_accuracy": 0.854050874710083, + "num_tokens": 100769779.0, + "step": 2589 + }, + { + "epoch": 0.3294746215494212, + "grad_norm": 3.0214033126831055, + "learning_rate": 1e-06, + "loss": 0.4399, + "mean_token_accuracy": 0.8516941070556641, + "num_tokens": 100807856.0, + "step": 2590 + }, + { + "epoch": 0.3296018318280117, + "grad_norm": 3.0156219005584717, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8745006918907166, + "num_tokens": 100838775.0, + "step": 2591 + }, + { + "epoch": 0.3297290421066022, + "grad_norm": 2.642345428466797, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8769673705101013, + "num_tokens": 100880365.0, + "step": 2592 + }, + { + "epoch": 0.3298562523851927, + "grad_norm": 2.7129995822906494, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.872812807559967, + "num_tokens": 100920059.0, + "step": 2593 + }, + { + "epoch": 0.32998346266378326, + "grad_norm": 2.8567259311676025, + "learning_rate": 1e-06, + "loss": 0.3983, + "mean_token_accuracy": 0.8670823574066162, + "num_tokens": 100957426.0, + "step": 2594 + }, + { + "epoch": 0.33011067294237373, + "grad_norm": 2.7823402881622314, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.864500880241394, + "num_tokens": 100995052.0, + "step": 2595 + }, + { + "epoch": 0.33023788322096426, + "grad_norm": 2.781038522720337, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8731710910797119, + "num_tokens": 101029836.0, + "step": 2596 + }, + { + "epoch": 0.3303650934995548, + "grad_norm": 2.9274585247039795, + "learning_rate": 1e-06, + "loss": 0.4374, + "mean_token_accuracy": 0.8512638211250305, + "num_tokens": 101068596.0, + "step": 2597 + }, + { + "epoch": 0.33049230377814526, + "grad_norm": 2.686225652694702, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8732815980911255, + "num_tokens": 101108349.0, + "step": 2598 + }, + { + "epoch": 0.3306195140567358, + "grad_norm": 3.2854926586151123, + "learning_rate": 1e-06, + "loss": 0.412, + "mean_token_accuracy": 0.8567607402801514, + "num_tokens": 101141357.0, + "step": 2599 + }, + { + "epoch": 0.3307467243353263, + "grad_norm": 2.8099148273468018, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.8608479499816895, + "num_tokens": 101178417.0, + "step": 2600 + }, + { + "epoch": 0.3308739346139168, + "grad_norm": 3.1539342403411865, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8845321536064148, + "num_tokens": 101207925.0, + "step": 2601 + }, + { + "epoch": 0.3310011448925073, + "grad_norm": 2.6850295066833496, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8851315379142761, + "num_tokens": 101245596.0, + "step": 2602 + }, + { + "epoch": 0.33112835517109784, + "grad_norm": 2.870856523513794, + "learning_rate": 1e-06, + "loss": 0.4155, + "mean_token_accuracy": 0.8590439558029175, + "num_tokens": 101284401.0, + "step": 2603 + }, + { + "epoch": 0.3312555654496883, + "grad_norm": 2.5805695056915283, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8793313503265381, + "num_tokens": 101324621.0, + "step": 2604 + }, + { + "epoch": 0.33138277572827884, + "grad_norm": 2.788992404937744, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8784677386283875, + "num_tokens": 101359178.0, + "step": 2605 + }, + { + "epoch": 0.3315099860068694, + "grad_norm": 2.5314509868621826, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8734973669052124, + "num_tokens": 101399498.0, + "step": 2606 + }, + { + "epoch": 0.33163719628545985, + "grad_norm": 2.9195868968963623, + "learning_rate": 1e-06, + "loss": 0.41, + "mean_token_accuracy": 0.8581048250198364, + "num_tokens": 101437458.0, + "step": 2607 + }, + { + "epoch": 0.3317644065640504, + "grad_norm": 2.646688222885132, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8706253170967102, + "num_tokens": 101477692.0, + "step": 2608 + }, + { + "epoch": 0.3318916168426409, + "grad_norm": 2.9197070598602295, + "learning_rate": 1e-06, + "loss": 0.4069, + "mean_token_accuracy": 0.8606222867965698, + "num_tokens": 101513662.0, + "step": 2609 + }, + { + "epoch": 0.3320188271212314, + "grad_norm": 2.5148253440856934, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8688305616378784, + "num_tokens": 101558163.0, + "step": 2610 + }, + { + "epoch": 0.3321460373998219, + "grad_norm": 2.861046314239502, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.8693162202835083, + "num_tokens": 101594035.0, + "step": 2611 + }, + { + "epoch": 0.33227324767841243, + "grad_norm": 2.831939935684204, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8872188329696655, + "num_tokens": 101629752.0, + "step": 2612 + }, + { + "epoch": 0.3324004579570029, + "grad_norm": 2.6795966625213623, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.8689224720001221, + "num_tokens": 101669359.0, + "step": 2613 + }, + { + "epoch": 0.33252766823559343, + "grad_norm": 2.663346290588379, + "learning_rate": 1e-06, + "loss": 0.4497, + "mean_token_accuracy": 0.8462891578674316, + "num_tokens": 101716391.0, + "step": 2614 + }, + { + "epoch": 0.33265487851418396, + "grad_norm": 2.7914860248565674, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8719398975372314, + "num_tokens": 101751481.0, + "step": 2615 + }, + { + "epoch": 0.33278208879277443, + "grad_norm": 3.2135133743286133, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8631538152694702, + "num_tokens": 101786268.0, + "step": 2616 + }, + { + "epoch": 0.33290929907136496, + "grad_norm": 2.7701871395111084, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8673569560050964, + "num_tokens": 101824153.0, + "step": 2617 + }, + { + "epoch": 0.3330365093499555, + "grad_norm": 2.6549816131591797, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8709841370582581, + "num_tokens": 101867237.0, + "step": 2618 + }, + { + "epoch": 0.33316371962854596, + "grad_norm": 2.7688889503479004, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.884544312953949, + "num_tokens": 101904944.0, + "step": 2619 + }, + { + "epoch": 0.3332909299071365, + "grad_norm": 2.896941900253296, + "learning_rate": 1e-06, + "loss": 0.4228, + "mean_token_accuracy": 0.8523666858673096, + "num_tokens": 101944664.0, + "step": 2620 + }, + { + "epoch": 0.333418140185727, + "grad_norm": 2.960085391998291, + "learning_rate": 1e-06, + "loss": 0.4737, + "mean_token_accuracy": 0.8406563997268677, + "num_tokens": 101986706.0, + "step": 2621 + }, + { + "epoch": 0.3335453504643175, + "grad_norm": 3.384136438369751, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.8635320663452148, + "num_tokens": 102018683.0, + "step": 2622 + }, + { + "epoch": 0.333672560742908, + "grad_norm": 2.7950525283813477, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8722012042999268, + "num_tokens": 102056996.0, + "step": 2623 + }, + { + "epoch": 0.33379977102149855, + "grad_norm": 2.7697606086730957, + "learning_rate": 1e-06, + "loss": 0.3952, + "mean_token_accuracy": 0.8622651100158691, + "num_tokens": 102096276.0, + "step": 2624 + }, + { + "epoch": 0.333926981300089, + "grad_norm": 2.6173908710479736, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8765591382980347, + "num_tokens": 102137167.0, + "step": 2625 + }, + { + "epoch": 0.33405419157867955, + "grad_norm": 2.8897385597229004, + "learning_rate": 1e-06, + "loss": 0.4276, + "mean_token_accuracy": 0.8516266942024231, + "num_tokens": 102174467.0, + "step": 2626 + }, + { + "epoch": 0.3341814018572701, + "grad_norm": 2.4750261306762695, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8766046762466431, + "num_tokens": 102214099.0, + "step": 2627 + }, + { + "epoch": 0.33430861213586055, + "grad_norm": 2.9546804428100586, + "learning_rate": 1e-06, + "loss": 0.3907, + "mean_token_accuracy": 0.866117000579834, + "num_tokens": 102248407.0, + "step": 2628 + }, + { + "epoch": 0.3344358224144511, + "grad_norm": 2.5062592029571533, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8938263654708862, + "num_tokens": 102284970.0, + "step": 2629 + }, + { + "epoch": 0.3345630326930416, + "grad_norm": 2.8823773860931396, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8682464361190796, + "num_tokens": 102319138.0, + "step": 2630 + }, + { + "epoch": 0.33469024297163213, + "grad_norm": 2.8167295455932617, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8734754920005798, + "num_tokens": 102354198.0, + "step": 2631 + }, + { + "epoch": 0.3348174532502226, + "grad_norm": 2.8296072483062744, + "learning_rate": 1e-06, + "loss": 0.395, + "mean_token_accuracy": 0.861331045627594, + "num_tokens": 102391378.0, + "step": 2632 + }, + { + "epoch": 0.33494466352881314, + "grad_norm": 2.526810884475708, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8835941553115845, + "num_tokens": 102431705.0, + "step": 2633 + }, + { + "epoch": 0.33507187380740366, + "grad_norm": 2.665241241455078, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8821585178375244, + "num_tokens": 102467470.0, + "step": 2634 + }, + { + "epoch": 0.33519908408599414, + "grad_norm": 2.832254409790039, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8735268115997314, + "num_tokens": 102504162.0, + "step": 2635 + }, + { + "epoch": 0.33532629436458466, + "grad_norm": 2.664999485015869, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8754786252975464, + "num_tokens": 102544146.0, + "step": 2636 + }, + { + "epoch": 0.3354535046431752, + "grad_norm": 2.8263282775878906, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.87582927942276, + "num_tokens": 102578752.0, + "step": 2637 + }, + { + "epoch": 0.33558071492176567, + "grad_norm": 2.415283679962158, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8757447004318237, + "num_tokens": 102625493.0, + "step": 2638 + }, + { + "epoch": 0.3357079252003562, + "grad_norm": 2.672180414199829, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8756488561630249, + "num_tokens": 102664811.0, + "step": 2639 + }, + { + "epoch": 0.3358351354789467, + "grad_norm": 2.574000597000122, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8782673478126526, + "num_tokens": 102705796.0, + "step": 2640 + }, + { + "epoch": 0.3359623457575372, + "grad_norm": 2.5831046104431152, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8752312660217285, + "num_tokens": 102748695.0, + "step": 2641 + }, + { + "epoch": 0.3360895560361277, + "grad_norm": 2.798370122909546, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8711119890213013, + "num_tokens": 102787188.0, + "step": 2642 + }, + { + "epoch": 0.33621676631471825, + "grad_norm": 2.4666786193847656, + "learning_rate": 1e-06, + "loss": 0.3835, + "mean_token_accuracy": 0.8683264851570129, + "num_tokens": 102830962.0, + "step": 2643 + }, + { + "epoch": 0.3363439765933087, + "grad_norm": 2.7779431343078613, + "learning_rate": 1e-06, + "loss": 0.3957, + "mean_token_accuracy": 0.8639458417892456, + "num_tokens": 102869160.0, + "step": 2644 + }, + { + "epoch": 0.33647118687189925, + "grad_norm": 2.9077179431915283, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8718595504760742, + "num_tokens": 102904804.0, + "step": 2645 + }, + { + "epoch": 0.3365983971504898, + "grad_norm": 2.6078805923461914, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8761193752288818, + "num_tokens": 102945462.0, + "step": 2646 + }, + { + "epoch": 0.33672560742908025, + "grad_norm": 2.8151402473449707, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.869815468788147, + "num_tokens": 102983156.0, + "step": 2647 + }, + { + "epoch": 0.3368528177076708, + "grad_norm": 2.8896844387054443, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8693217039108276, + "num_tokens": 103019337.0, + "step": 2648 + }, + { + "epoch": 0.3369800279862613, + "grad_norm": 2.784507989883423, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8697251081466675, + "num_tokens": 103058001.0, + "step": 2649 + }, + { + "epoch": 0.3371072382648518, + "grad_norm": 2.747284173965454, + "learning_rate": 1e-06, + "loss": 0.3908, + "mean_token_accuracy": 0.8644200563430786, + "num_tokens": 103097165.0, + "step": 2650 + }, + { + "epoch": 0.3372344485434423, + "grad_norm": 2.857719898223877, + "learning_rate": 1e-06, + "loss": 0.3826, + "mean_token_accuracy": 0.8636698722839355, + "num_tokens": 103132706.0, + "step": 2651 + }, + { + "epoch": 0.33736165882203284, + "grad_norm": 2.732119083404541, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8730063438415527, + "num_tokens": 103170990.0, + "step": 2652 + }, + { + "epoch": 0.3374888691006233, + "grad_norm": 2.594902753829956, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8742449879646301, + "num_tokens": 103211228.0, + "step": 2653 + }, + { + "epoch": 0.33761607937921384, + "grad_norm": 2.811018228530884, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8620775938034058, + "num_tokens": 103246977.0, + "step": 2654 + }, + { + "epoch": 0.33774328965780437, + "grad_norm": 2.89485502243042, + "learning_rate": 1e-06, + "loss": 0.4426, + "mean_token_accuracy": 0.8484952449798584, + "num_tokens": 103288651.0, + "step": 2655 + }, + { + "epoch": 0.33787049993639484, + "grad_norm": 2.848742723464966, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.8663134574890137, + "num_tokens": 103330002.0, + "step": 2656 + }, + { + "epoch": 0.33799771021498537, + "grad_norm": 2.6802375316619873, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.873628556728363, + "num_tokens": 103364929.0, + "step": 2657 + }, + { + "epoch": 0.3381249204935759, + "grad_norm": 2.7675743103027344, + "learning_rate": 1e-06, + "loss": 0.4529, + "mean_token_accuracy": 0.8470053672790527, + "num_tokens": 103407029.0, + "step": 2658 + }, + { + "epoch": 0.33825213077216637, + "grad_norm": 2.6817235946655273, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8692511320114136, + "num_tokens": 103445982.0, + "step": 2659 + }, + { + "epoch": 0.3383793410507569, + "grad_norm": 2.8939802646636963, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.868999719619751, + "num_tokens": 103479710.0, + "step": 2660 + }, + { + "epoch": 0.3385065513293474, + "grad_norm": 2.659320831298828, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8758337497711182, + "num_tokens": 103521185.0, + "step": 2661 + }, + { + "epoch": 0.3386337616079379, + "grad_norm": 2.5458028316497803, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8820701837539673, + "num_tokens": 103566176.0, + "step": 2662 + }, + { + "epoch": 0.33876097188652843, + "grad_norm": 2.762573003768921, + "learning_rate": 1e-06, + "loss": 0.4034, + "mean_token_accuracy": 0.8617043495178223, + "num_tokens": 103607962.0, + "step": 2663 + }, + { + "epoch": 0.33888818216511896, + "grad_norm": 2.991250991821289, + "learning_rate": 1e-06, + "loss": 0.4105, + "mean_token_accuracy": 0.8649791479110718, + "num_tokens": 103640794.0, + "step": 2664 + }, + { + "epoch": 0.33901539244370943, + "grad_norm": 2.6990416049957275, + "learning_rate": 1e-06, + "loss": 0.3907, + "mean_token_accuracy": 0.8661494255065918, + "num_tokens": 103684239.0, + "step": 2665 + }, + { + "epoch": 0.33914260272229996, + "grad_norm": 2.8546950817108154, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.8677897453308105, + "num_tokens": 103720152.0, + "step": 2666 + }, + { + "epoch": 0.3392698130008905, + "grad_norm": 3.000110387802124, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8691141605377197, + "num_tokens": 103754874.0, + "step": 2667 + }, + { + "epoch": 0.33939702327948096, + "grad_norm": 2.7092411518096924, + "learning_rate": 1e-06, + "loss": 0.4002, + "mean_token_accuracy": 0.867148756980896, + "num_tokens": 103796479.0, + "step": 2668 + }, + { + "epoch": 0.3395242335580715, + "grad_norm": 3.2044479846954346, + "learning_rate": 1e-06, + "loss": 0.4696, + "mean_token_accuracy": 0.8373438119888306, + "num_tokens": 103836012.0, + "step": 2669 + }, + { + "epoch": 0.339651443836662, + "grad_norm": 2.9852943420410156, + "learning_rate": 1e-06, + "loss": 0.3996, + "mean_token_accuracy": 0.8624697923660278, + "num_tokens": 103877579.0, + "step": 2670 + }, + { + "epoch": 0.3397786541152525, + "grad_norm": 2.672020673751831, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8775153756141663, + "num_tokens": 103919355.0, + "step": 2671 + }, + { + "epoch": 0.339905864393843, + "grad_norm": 2.629915714263916, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8646531105041504, + "num_tokens": 103958830.0, + "step": 2672 + }, + { + "epoch": 0.34003307467243354, + "grad_norm": 2.763822555541992, + "learning_rate": 1e-06, + "loss": 0.4083, + "mean_token_accuracy": 0.8599324822425842, + "num_tokens": 103997330.0, + "step": 2673 + }, + { + "epoch": 0.340160284951024, + "grad_norm": 2.694960832595825, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8757583498954773, + "num_tokens": 104036913.0, + "step": 2674 + }, + { + "epoch": 0.34028749522961454, + "grad_norm": 2.748049736022949, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8804042339324951, + "num_tokens": 104071575.0, + "step": 2675 + }, + { + "epoch": 0.3404147055082051, + "grad_norm": 2.431170701980591, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.881371021270752, + "num_tokens": 104113949.0, + "step": 2676 + }, + { + "epoch": 0.34054191578679555, + "grad_norm": 2.55464768409729, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8699544072151184, + "num_tokens": 104159312.0, + "step": 2677 + }, + { + "epoch": 0.3406691260653861, + "grad_norm": 2.9966673851013184, + "learning_rate": 1e-06, + "loss": 0.4059, + "mean_token_accuracy": 0.8607395887374878, + "num_tokens": 104193558.0, + "step": 2678 + }, + { + "epoch": 0.3407963363439766, + "grad_norm": 2.4422900676727295, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8815000653266907, + "num_tokens": 104237598.0, + "step": 2679 + }, + { + "epoch": 0.34092354662256713, + "grad_norm": 2.604112386703491, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8798570036888123, + "num_tokens": 104277875.0, + "step": 2680 + }, + { + "epoch": 0.3410507569011576, + "grad_norm": 3.344559669494629, + "learning_rate": 1e-06, + "loss": 0.4322, + "mean_token_accuracy": 0.8635592460632324, + "num_tokens": 104307961.0, + "step": 2681 + }, + { + "epoch": 0.34117796717974813, + "grad_norm": 2.9924583435058594, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8751875758171082, + "num_tokens": 104339586.0, + "step": 2682 + }, + { + "epoch": 0.34130517745833866, + "grad_norm": 2.956371307373047, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8754672408103943, + "num_tokens": 104375648.0, + "step": 2683 + }, + { + "epoch": 0.34143238773692913, + "grad_norm": 2.6961612701416016, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8851891756057739, + "num_tokens": 104412504.0, + "step": 2684 + }, + { + "epoch": 0.34155959801551966, + "grad_norm": 2.9361860752105713, + "learning_rate": 1e-06, + "loss": 0.4167, + "mean_token_accuracy": 0.8580440878868103, + "num_tokens": 104450126.0, + "step": 2685 + }, + { + "epoch": 0.3416868082941102, + "grad_norm": 2.8091046810150146, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8737412691116333, + "num_tokens": 104488152.0, + "step": 2686 + }, + { + "epoch": 0.34181401857270066, + "grad_norm": 2.968848705291748, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8665550947189331, + "num_tokens": 104520335.0, + "step": 2687 + }, + { + "epoch": 0.3419412288512912, + "grad_norm": 2.795023202896118, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8783353567123413, + "num_tokens": 104558359.0, + "step": 2688 + }, + { + "epoch": 0.3420684391298817, + "grad_norm": 2.730163097381592, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8757914304733276, + "num_tokens": 104594466.0, + "step": 2689 + }, + { + "epoch": 0.3421956494084722, + "grad_norm": 2.5405662059783936, + "learning_rate": 1e-06, + "loss": 0.3939, + "mean_token_accuracy": 0.8667590022087097, + "num_tokens": 104638335.0, + "step": 2690 + }, + { + "epoch": 0.3423228596870627, + "grad_norm": 2.6965699195861816, + "learning_rate": 1e-06, + "loss": 0.3994, + "mean_token_accuracy": 0.8639985918998718, + "num_tokens": 104678764.0, + "step": 2691 + }, + { + "epoch": 0.34245006996565325, + "grad_norm": 2.6199910640716553, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8740066289901733, + "num_tokens": 104718440.0, + "step": 2692 + }, + { + "epoch": 0.3425772802442437, + "grad_norm": 2.618563413619995, + "learning_rate": 1e-06, + "loss": 0.3945, + "mean_token_accuracy": 0.8630731105804443, + "num_tokens": 104762290.0, + "step": 2693 + }, + { + "epoch": 0.34270449052283425, + "grad_norm": 2.4661037921905518, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8824755549430847, + "num_tokens": 104805424.0, + "step": 2694 + }, + { + "epoch": 0.3428317008014248, + "grad_norm": 2.7283778190612793, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8751929998397827, + "num_tokens": 104844625.0, + "step": 2695 + }, + { + "epoch": 0.34295891108001525, + "grad_norm": 2.697986125946045, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8777592778205872, + "num_tokens": 104882344.0, + "step": 2696 + }, + { + "epoch": 0.3430861213586058, + "grad_norm": 2.693991184234619, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8679811954498291, + "num_tokens": 104924627.0, + "step": 2697 + }, + { + "epoch": 0.3432133316371963, + "grad_norm": 2.7475762367248535, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.877130389213562, + "num_tokens": 104963225.0, + "step": 2698 + }, + { + "epoch": 0.3433405419157868, + "grad_norm": 2.6611461639404297, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.866370439529419, + "num_tokens": 105005417.0, + "step": 2699 + }, + { + "epoch": 0.3434677521943773, + "grad_norm": 2.739032745361328, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8647319078445435, + "num_tokens": 105045810.0, + "step": 2700 + }, + { + "epoch": 0.34359496247296784, + "grad_norm": 2.678107976913452, + "learning_rate": 1e-06, + "loss": 0.41, + "mean_token_accuracy": 0.8609744906425476, + "num_tokens": 105089996.0, + "step": 2701 + }, + { + "epoch": 0.3437221727515583, + "grad_norm": 2.69272780418396, + "learning_rate": 1e-06, + "loss": 0.41, + "mean_token_accuracy": 0.8604103326797485, + "num_tokens": 105132039.0, + "step": 2702 + }, + { + "epoch": 0.34384938303014884, + "grad_norm": 2.958387851715088, + "learning_rate": 1e-06, + "loss": 0.3956, + "mean_token_accuracy": 0.8632293939590454, + "num_tokens": 105166232.0, + "step": 2703 + }, + { + "epoch": 0.34397659330873936, + "grad_norm": 2.694186210632324, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8667941093444824, + "num_tokens": 105203282.0, + "step": 2704 + }, + { + "epoch": 0.34410380358732984, + "grad_norm": 3.0252931118011475, + "learning_rate": 1e-06, + "loss": 0.4336, + "mean_token_accuracy": 0.850654661655426, + "num_tokens": 105238332.0, + "step": 2705 + }, + { + "epoch": 0.34423101386592037, + "grad_norm": 2.751333713531494, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8741408586502075, + "num_tokens": 105273566.0, + "step": 2706 + }, + { + "epoch": 0.3443582241445109, + "grad_norm": 2.6906394958496094, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8814084529876709, + "num_tokens": 105310545.0, + "step": 2707 + }, + { + "epoch": 0.34448543442310137, + "grad_norm": 2.7628912925720215, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8663426637649536, + "num_tokens": 105347241.0, + "step": 2708 + }, + { + "epoch": 0.3446126447016919, + "grad_norm": 2.552276849746704, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8719918727874756, + "num_tokens": 105389584.0, + "step": 2709 + }, + { + "epoch": 0.3447398549802824, + "grad_norm": 2.6769754886627197, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8825005292892456, + "num_tokens": 105423961.0, + "step": 2710 + }, + { + "epoch": 0.3448670652588729, + "grad_norm": 2.6074697971343994, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8813951015472412, + "num_tokens": 105462183.0, + "step": 2711 + }, + { + "epoch": 0.3449942755374634, + "grad_norm": 2.5822622776031494, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8807420134544373, + "num_tokens": 105499900.0, + "step": 2712 + }, + { + "epoch": 0.34512148581605395, + "grad_norm": 2.802950859069824, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8751873970031738, + "num_tokens": 105537355.0, + "step": 2713 + }, + { + "epoch": 0.3452486960946444, + "grad_norm": 2.5894126892089844, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8747321367263794, + "num_tokens": 105579476.0, + "step": 2714 + }, + { + "epoch": 0.34537590637323495, + "grad_norm": 2.712730646133423, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.882798969745636, + "num_tokens": 105618235.0, + "step": 2715 + }, + { + "epoch": 0.3455031166518255, + "grad_norm": 2.6492130756378174, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8823719620704651, + "num_tokens": 105660715.0, + "step": 2716 + }, + { + "epoch": 0.34563032693041595, + "grad_norm": 2.592484712600708, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8807063698768616, + "num_tokens": 105703066.0, + "step": 2717 + }, + { + "epoch": 0.3457575372090065, + "grad_norm": 2.8053202629089355, + "learning_rate": 1e-06, + "loss": 0.3929, + "mean_token_accuracy": 0.8660643696784973, + "num_tokens": 105740354.0, + "step": 2718 + }, + { + "epoch": 0.345884747487597, + "grad_norm": 2.930501937866211, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8835229277610779, + "num_tokens": 105771949.0, + "step": 2719 + }, + { + "epoch": 0.3460119577661875, + "grad_norm": 2.592888832092285, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8728759288787842, + "num_tokens": 105813310.0, + "step": 2720 + }, + { + "epoch": 0.346139168044778, + "grad_norm": 2.9078269004821777, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.867338240146637, + "num_tokens": 105847661.0, + "step": 2721 + }, + { + "epoch": 0.34626637832336854, + "grad_norm": 2.5600745677948, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8720252513885498, + "num_tokens": 105889013.0, + "step": 2722 + }, + { + "epoch": 0.346393588601959, + "grad_norm": 2.6197540760040283, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.883399486541748, + "num_tokens": 105930551.0, + "step": 2723 + }, + { + "epoch": 0.34652079888054954, + "grad_norm": 2.8759379386901855, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8739885687828064, + "num_tokens": 105965366.0, + "step": 2724 + }, + { + "epoch": 0.34664800915914007, + "grad_norm": 2.6782994270324707, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.873319149017334, + "num_tokens": 106003753.0, + "step": 2725 + }, + { + "epoch": 0.34677521943773054, + "grad_norm": 2.6183078289031982, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8679865598678589, + "num_tokens": 106043216.0, + "step": 2726 + }, + { + "epoch": 0.34690242971632107, + "grad_norm": 2.880350112915039, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8714534640312195, + "num_tokens": 106082883.0, + "step": 2727 + }, + { + "epoch": 0.3470296399949116, + "grad_norm": 2.8315811157226562, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8872606754302979, + "num_tokens": 106116435.0, + "step": 2728 + }, + { + "epoch": 0.3471568502735021, + "grad_norm": 2.902038097381592, + "learning_rate": 1e-06, + "loss": 0.4076, + "mean_token_accuracy": 0.8607068061828613, + "num_tokens": 106151295.0, + "step": 2729 + }, + { + "epoch": 0.3472840605520926, + "grad_norm": 2.601689100265503, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8784041404724121, + "num_tokens": 106192186.0, + "step": 2730 + }, + { + "epoch": 0.3474112708306831, + "grad_norm": 2.744874954223633, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8744755387306213, + "num_tokens": 106226820.0, + "step": 2731 + }, + { + "epoch": 0.34753848110927366, + "grad_norm": 2.7686514854431152, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8861313462257385, + "num_tokens": 106262166.0, + "step": 2732 + }, + { + "epoch": 0.34766569138786413, + "grad_norm": 2.892317533493042, + "learning_rate": 1e-06, + "loss": 0.3951, + "mean_token_accuracy": 0.8638681173324585, + "num_tokens": 106298678.0, + "step": 2733 + }, + { + "epoch": 0.34779290166645466, + "grad_norm": 2.919545888900757, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.86375492811203, + "num_tokens": 106335422.0, + "step": 2734 + }, + { + "epoch": 0.3479201119450452, + "grad_norm": 2.7012808322906494, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8748762607574463, + "num_tokens": 106373512.0, + "step": 2735 + }, + { + "epoch": 0.34804732222363566, + "grad_norm": 2.655393362045288, + "learning_rate": 1e-06, + "loss": 0.4256, + "mean_token_accuracy": 0.856440007686615, + "num_tokens": 106416854.0, + "step": 2736 + }, + { + "epoch": 0.3481745325022262, + "grad_norm": 2.7591030597686768, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8858929872512817, + "num_tokens": 106449900.0, + "step": 2737 + }, + { + "epoch": 0.3483017427808167, + "grad_norm": 2.7145028114318848, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8790711760520935, + "num_tokens": 106484370.0, + "step": 2738 + }, + { + "epoch": 0.3484289530594072, + "grad_norm": 2.6299023628234863, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8813655376434326, + "num_tokens": 106520298.0, + "step": 2739 + }, + { + "epoch": 0.3485561633379977, + "grad_norm": 2.456171751022339, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8721980452537537, + "num_tokens": 106564693.0, + "step": 2740 + }, + { + "epoch": 0.34868337361658824, + "grad_norm": 2.6983821392059326, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8753365278244019, + "num_tokens": 106602747.0, + "step": 2741 + }, + { + "epoch": 0.3488105838951787, + "grad_norm": 2.741312265396118, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8716959953308105, + "num_tokens": 106643776.0, + "step": 2742 + }, + { + "epoch": 0.34893779417376924, + "grad_norm": 2.8277781009674072, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8818073272705078, + "num_tokens": 106684736.0, + "step": 2743 + }, + { + "epoch": 0.3490650044523598, + "grad_norm": 2.7548880577087402, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8842176198959351, + "num_tokens": 106719738.0, + "step": 2744 + }, + { + "epoch": 0.34919221473095025, + "grad_norm": 2.72839617729187, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.8717701435089111, + "num_tokens": 106757928.0, + "step": 2745 + }, + { + "epoch": 0.3493194250095408, + "grad_norm": 2.593667507171631, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8802640438079834, + "num_tokens": 106799468.0, + "step": 2746 + }, + { + "epoch": 0.3494466352881313, + "grad_norm": 3.0883913040161133, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8615625500679016, + "num_tokens": 106832209.0, + "step": 2747 + }, + { + "epoch": 0.3495738455667218, + "grad_norm": 2.7543551921844482, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8722553253173828, + "num_tokens": 106872385.0, + "step": 2748 + }, + { + "epoch": 0.3497010558453123, + "grad_norm": 2.3659374713897705, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8957438468933105, + "num_tokens": 106916595.0, + "step": 2749 + }, + { + "epoch": 0.34982826612390283, + "grad_norm": 2.787257432937622, + "learning_rate": 1e-06, + "loss": 0.4373, + "mean_token_accuracy": 0.8524876236915588, + "num_tokens": 106960055.0, + "step": 2750 + }, + { + "epoch": 0.3499554764024933, + "grad_norm": 2.5581817626953125, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8673611879348755, + "num_tokens": 106998873.0, + "step": 2751 + }, + { + "epoch": 0.35008268668108383, + "grad_norm": 2.5525336265563965, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8716373443603516, + "num_tokens": 107039508.0, + "step": 2752 + }, + { + "epoch": 0.35020989695967436, + "grad_norm": 2.7936103343963623, + "learning_rate": 1e-06, + "loss": 0.4098, + "mean_token_accuracy": 0.8588089346885681, + "num_tokens": 107078518.0, + "step": 2753 + }, + { + "epoch": 0.35033710723826483, + "grad_norm": 3.047935724258423, + "learning_rate": 1e-06, + "loss": 0.4252, + "mean_token_accuracy": 0.8586452007293701, + "num_tokens": 107117016.0, + "step": 2754 + }, + { + "epoch": 0.35046431751685536, + "grad_norm": 2.7476701736450195, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8873662948608398, + "num_tokens": 107155465.0, + "step": 2755 + }, + { + "epoch": 0.3505915277954459, + "grad_norm": 2.7536511421203613, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8782749176025391, + "num_tokens": 107188976.0, + "step": 2756 + }, + { + "epoch": 0.35071873807403636, + "grad_norm": 2.6394407749176025, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8690431118011475, + "num_tokens": 107232259.0, + "step": 2757 + }, + { + "epoch": 0.3508459483526269, + "grad_norm": 2.6254634857177734, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8876262307167053, + "num_tokens": 107269991.0, + "step": 2758 + }, + { + "epoch": 0.3509731586312174, + "grad_norm": 3.2243170738220215, + "learning_rate": 1e-06, + "loss": 0.4396, + "mean_token_accuracy": 0.8520084619522095, + "num_tokens": 107304734.0, + "step": 2759 + }, + { + "epoch": 0.3511003689098079, + "grad_norm": 2.8536574840545654, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8728488087654114, + "num_tokens": 107344067.0, + "step": 2760 + }, + { + "epoch": 0.3512275791883984, + "grad_norm": 2.7441179752349854, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.882603108882904, + "num_tokens": 107379500.0, + "step": 2761 + }, + { + "epoch": 0.35135478946698895, + "grad_norm": 2.675915002822876, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8700050115585327, + "num_tokens": 107418590.0, + "step": 2762 + }, + { + "epoch": 0.3514819997455794, + "grad_norm": 3.0304229259490967, + "learning_rate": 1e-06, + "loss": 0.4125, + "mean_token_accuracy": 0.8561356067657471, + "num_tokens": 107453562.0, + "step": 2763 + }, + { + "epoch": 0.35160921002416995, + "grad_norm": 2.8748083114624023, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.869977593421936, + "num_tokens": 107492596.0, + "step": 2764 + }, + { + "epoch": 0.3517364203027605, + "grad_norm": 2.95870041847229, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8645283579826355, + "num_tokens": 107526232.0, + "step": 2765 + }, + { + "epoch": 0.35186363058135095, + "grad_norm": 2.921908378601074, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8653123378753662, + "num_tokens": 107563574.0, + "step": 2766 + }, + { + "epoch": 0.3519908408599415, + "grad_norm": 2.6166720390319824, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8717718124389648, + "num_tokens": 107605536.0, + "step": 2767 + }, + { + "epoch": 0.352118051138532, + "grad_norm": 2.5524723529815674, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8693794012069702, + "num_tokens": 107650076.0, + "step": 2768 + }, + { + "epoch": 0.3522452614171225, + "grad_norm": 2.770840644836426, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8661879897117615, + "num_tokens": 107688838.0, + "step": 2769 + }, + { + "epoch": 0.352372471695713, + "grad_norm": 2.7973811626434326, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8701084852218628, + "num_tokens": 107727740.0, + "step": 2770 + }, + { + "epoch": 0.35249968197430354, + "grad_norm": 2.9353580474853516, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8661583662033081, + "num_tokens": 107762138.0, + "step": 2771 + }, + { + "epoch": 0.352626892252894, + "grad_norm": 2.7678184509277344, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8710136413574219, + "num_tokens": 107802944.0, + "step": 2772 + }, + { + "epoch": 0.35275410253148454, + "grad_norm": 2.702840805053711, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8778451681137085, + "num_tokens": 107843888.0, + "step": 2773 + }, + { + "epoch": 0.35288131281007507, + "grad_norm": 2.489906072616577, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8796999454498291, + "num_tokens": 107887387.0, + "step": 2774 + }, + { + "epoch": 0.35300852308866554, + "grad_norm": 2.7837700843811035, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8746819496154785, + "num_tokens": 107924543.0, + "step": 2775 + }, + { + "epoch": 0.35313573336725607, + "grad_norm": 2.631516695022583, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8721919655799866, + "num_tokens": 107964357.0, + "step": 2776 + }, + { + "epoch": 0.3532629436458466, + "grad_norm": 2.622654914855957, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8694268465042114, + "num_tokens": 108006750.0, + "step": 2777 + }, + { + "epoch": 0.35339015392443707, + "grad_norm": 2.494067668914795, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8661630153656006, + "num_tokens": 108053935.0, + "step": 2778 + }, + { + "epoch": 0.3535173642030276, + "grad_norm": 2.481154441833496, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8736448287963867, + "num_tokens": 108101017.0, + "step": 2779 + }, + { + "epoch": 0.3536445744816181, + "grad_norm": 2.9762117862701416, + "learning_rate": 1e-06, + "loss": 0.4112, + "mean_token_accuracy": 0.8588265180587769, + "num_tokens": 108136076.0, + "step": 2780 + }, + { + "epoch": 0.35377178476020865, + "grad_norm": 2.9871885776519775, + "learning_rate": 1e-06, + "loss": 0.4025, + "mean_token_accuracy": 0.8619070053100586, + "num_tokens": 108170692.0, + "step": 2781 + }, + { + "epoch": 0.3538989950387991, + "grad_norm": 2.5824029445648193, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8869808912277222, + "num_tokens": 108210485.0, + "step": 2782 + }, + { + "epoch": 0.35402620531738965, + "grad_norm": 2.6082890033721924, + "learning_rate": 1e-06, + "loss": 0.4043, + "mean_token_accuracy": 0.8613203167915344, + "num_tokens": 108253270.0, + "step": 2783 + }, + { + "epoch": 0.3541534155959802, + "grad_norm": 2.658263921737671, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8752721548080444, + "num_tokens": 108293393.0, + "step": 2784 + }, + { + "epoch": 0.35428062587457065, + "grad_norm": 2.639009952545166, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8792874813079834, + "num_tokens": 108334070.0, + "step": 2785 + }, + { + "epoch": 0.3544078361531612, + "grad_norm": 2.9583632946014404, + "learning_rate": 1e-06, + "loss": 0.4151, + "mean_token_accuracy": 0.8592484593391418, + "num_tokens": 108368988.0, + "step": 2786 + }, + { + "epoch": 0.3545350464317517, + "grad_norm": 2.4193882942199707, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8833808898925781, + "num_tokens": 108413568.0, + "step": 2787 + }, + { + "epoch": 0.3546622567103422, + "grad_norm": 2.929940938949585, + "learning_rate": 1e-06, + "loss": 0.4153, + "mean_token_accuracy": 0.8661590814590454, + "num_tokens": 108449162.0, + "step": 2788 + }, + { + "epoch": 0.3547894669889327, + "grad_norm": 2.6751081943511963, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8759809136390686, + "num_tokens": 108486042.0, + "step": 2789 + }, + { + "epoch": 0.35491667726752324, + "grad_norm": 2.5004029273986816, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8870965242385864, + "num_tokens": 108526951.0, + "step": 2790 + }, + { + "epoch": 0.3550438875461137, + "grad_norm": 2.631880760192871, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8716421127319336, + "num_tokens": 108567768.0, + "step": 2791 + }, + { + "epoch": 0.35517109782470424, + "grad_norm": 2.7286014556884766, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8700681924819946, + "num_tokens": 108608915.0, + "step": 2792 + }, + { + "epoch": 0.35529830810329477, + "grad_norm": 2.603924036026001, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8755711317062378, + "num_tokens": 108647377.0, + "step": 2793 + }, + { + "epoch": 0.35542551838188524, + "grad_norm": 3.008981704711914, + "learning_rate": 1e-06, + "loss": 0.397, + "mean_token_accuracy": 0.8635146617889404, + "num_tokens": 108686921.0, + "step": 2794 + }, + { + "epoch": 0.35555272866047577, + "grad_norm": 2.9404258728027344, + "learning_rate": 1e-06, + "loss": 0.4049, + "mean_token_accuracy": 0.8609868288040161, + "num_tokens": 108722116.0, + "step": 2795 + }, + { + "epoch": 0.3556799389390663, + "grad_norm": 2.878452777862549, + "learning_rate": 1e-06, + "loss": 0.393, + "mean_token_accuracy": 0.8622990250587463, + "num_tokens": 108757346.0, + "step": 2796 + }, + { + "epoch": 0.35580714921765677, + "grad_norm": 2.6172499656677246, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.875295877456665, + "num_tokens": 108796209.0, + "step": 2797 + }, + { + "epoch": 0.3559343594962473, + "grad_norm": 2.540339708328247, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8925051689147949, + "num_tokens": 108835884.0, + "step": 2798 + }, + { + "epoch": 0.3560615697748378, + "grad_norm": 2.9729366302490234, + "learning_rate": 1e-06, + "loss": 0.4157, + "mean_token_accuracy": 0.8592920899391174, + "num_tokens": 108872823.0, + "step": 2799 + }, + { + "epoch": 0.3561887800534283, + "grad_norm": 2.560567617416382, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8733741044998169, + "num_tokens": 108915855.0, + "step": 2800 + }, + { + "epoch": 0.35631599033201883, + "grad_norm": 2.5367698669433594, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8738429546356201, + "num_tokens": 108962296.0, + "step": 2801 + }, + { + "epoch": 0.35644320061060936, + "grad_norm": 2.548896312713623, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8710124492645264, + "num_tokens": 109005247.0, + "step": 2802 + }, + { + "epoch": 0.35657041088919983, + "grad_norm": 2.6449010372161865, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8720158338546753, + "num_tokens": 109042544.0, + "step": 2803 + }, + { + "epoch": 0.35669762116779036, + "grad_norm": 2.642521619796753, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8694964647293091, + "num_tokens": 109084726.0, + "step": 2804 + }, + { + "epoch": 0.3568248314463809, + "grad_norm": 2.505772113800049, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8749585151672363, + "num_tokens": 109128609.0, + "step": 2805 + }, + { + "epoch": 0.35695204172497136, + "grad_norm": 2.6837778091430664, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8743384480476379, + "num_tokens": 109168464.0, + "step": 2806 + }, + { + "epoch": 0.3570792520035619, + "grad_norm": 2.637601375579834, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.881142258644104, + "num_tokens": 109205861.0, + "step": 2807 + }, + { + "epoch": 0.3572064622821524, + "grad_norm": 2.866912841796875, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8687502145767212, + "num_tokens": 109240158.0, + "step": 2808 + }, + { + "epoch": 0.3573336725607429, + "grad_norm": 2.7754056453704834, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8805131316184998, + "num_tokens": 109277674.0, + "step": 2809 + }, + { + "epoch": 0.3574608828393334, + "grad_norm": 2.6256840229034424, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8794729709625244, + "num_tokens": 109317630.0, + "step": 2810 + }, + { + "epoch": 0.35758809311792394, + "grad_norm": 2.924750804901123, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8775361776351929, + "num_tokens": 109351488.0, + "step": 2811 + }, + { + "epoch": 0.3577153033965144, + "grad_norm": 2.7864229679107666, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8765811324119568, + "num_tokens": 109385329.0, + "step": 2812 + }, + { + "epoch": 0.35784251367510495, + "grad_norm": 2.9611356258392334, + "learning_rate": 1e-06, + "loss": 0.4197, + "mean_token_accuracy": 0.8559268712997437, + "num_tokens": 109420213.0, + "step": 2813 + }, + { + "epoch": 0.3579697239536955, + "grad_norm": 2.5629141330718994, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8703444004058838, + "num_tokens": 109462160.0, + "step": 2814 + }, + { + "epoch": 0.35809693423228595, + "grad_norm": 2.4674925804138184, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8778566122055054, + "num_tokens": 109505991.0, + "step": 2815 + }, + { + "epoch": 0.3582241445108765, + "grad_norm": 2.542039632797241, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.868893027305603, + "num_tokens": 109549091.0, + "step": 2816 + }, + { + "epoch": 0.358351354789467, + "grad_norm": 2.8423659801483154, + "learning_rate": 1e-06, + "loss": 0.4136, + "mean_token_accuracy": 0.8582791686058044, + "num_tokens": 109588253.0, + "step": 2817 + }, + { + "epoch": 0.3584785650680575, + "grad_norm": 2.8698203563690186, + "learning_rate": 1e-06, + "loss": 0.3879, + "mean_token_accuracy": 0.8679993152618408, + "num_tokens": 109623144.0, + "step": 2818 + }, + { + "epoch": 0.358605775346648, + "grad_norm": 2.6698052883148193, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8713558912277222, + "num_tokens": 109665064.0, + "step": 2819 + }, + { + "epoch": 0.35873298562523853, + "grad_norm": 3.1874492168426514, + "learning_rate": 1e-06, + "loss": 0.4452, + "mean_token_accuracy": 0.8473430275917053, + "num_tokens": 109698073.0, + "step": 2820 + }, + { + "epoch": 0.358860195903829, + "grad_norm": 2.9369819164276123, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8758540153503418, + "num_tokens": 109729036.0, + "step": 2821 + }, + { + "epoch": 0.35898740618241953, + "grad_norm": 2.735551118850708, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.875841498374939, + "num_tokens": 109767873.0, + "step": 2822 + }, + { + "epoch": 0.35911461646101006, + "grad_norm": 2.7535440921783447, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.871892511844635, + "num_tokens": 109811682.0, + "step": 2823 + }, + { + "epoch": 0.35924182673960053, + "grad_norm": 2.9093472957611084, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8795071244239807, + "num_tokens": 109845806.0, + "step": 2824 + }, + { + "epoch": 0.35936903701819106, + "grad_norm": 2.623945951461792, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8786249756813049, + "num_tokens": 109885342.0, + "step": 2825 + }, + { + "epoch": 0.3594962472967816, + "grad_norm": 2.7527639865875244, + "learning_rate": 1e-06, + "loss": 0.4124, + "mean_token_accuracy": 0.8598878383636475, + "num_tokens": 109927850.0, + "step": 2826 + }, + { + "epoch": 0.35962345757537206, + "grad_norm": 2.744468927383423, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8693209886550903, + "num_tokens": 109964813.0, + "step": 2827 + }, + { + "epoch": 0.3597506678539626, + "grad_norm": 2.8167459964752197, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8695698380470276, + "num_tokens": 110003005.0, + "step": 2828 + }, + { + "epoch": 0.3598778781325531, + "grad_norm": 2.609621286392212, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.878873348236084, + "num_tokens": 110040065.0, + "step": 2829 + }, + { + "epoch": 0.36000508841114365, + "grad_norm": 2.9407896995544434, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8818123936653137, + "num_tokens": 110077283.0, + "step": 2830 + }, + { + "epoch": 0.3601322986897341, + "grad_norm": 2.6359918117523193, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8745417594909668, + "num_tokens": 110118574.0, + "step": 2831 + }, + { + "epoch": 0.36025950896832465, + "grad_norm": 2.7773303985595703, + "learning_rate": 1e-06, + "loss": 0.4103, + "mean_token_accuracy": 0.8619977235794067, + "num_tokens": 110158336.0, + "step": 2832 + }, + { + "epoch": 0.3603867192469152, + "grad_norm": 2.894242763519287, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8690947890281677, + "num_tokens": 110192609.0, + "step": 2833 + }, + { + "epoch": 0.36051392952550565, + "grad_norm": 3.1466166973114014, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8743887543678284, + "num_tokens": 110226214.0, + "step": 2834 + }, + { + "epoch": 0.3606411398040962, + "grad_norm": 2.602325439453125, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8694480657577515, + "num_tokens": 110270336.0, + "step": 2835 + }, + { + "epoch": 0.3607683500826867, + "grad_norm": 2.7976250648498535, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8802973031997681, + "num_tokens": 110306468.0, + "step": 2836 + }, + { + "epoch": 0.3608955603612772, + "grad_norm": 2.8274688720703125, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8752130270004272, + "num_tokens": 110342820.0, + "step": 2837 + }, + { + "epoch": 0.3610227706398677, + "grad_norm": 2.801292657852173, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8762422800064087, + "num_tokens": 110380383.0, + "step": 2838 + }, + { + "epoch": 0.36114998091845824, + "grad_norm": 3.0438637733459473, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8803299069404602, + "num_tokens": 110410998.0, + "step": 2839 + }, + { + "epoch": 0.3612771911970487, + "grad_norm": 2.8294382095336914, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8676045536994934, + "num_tokens": 110447323.0, + "step": 2840 + }, + { + "epoch": 0.36140440147563924, + "grad_norm": 2.9492053985595703, + "learning_rate": 1e-06, + "loss": 0.417, + "mean_token_accuracy": 0.8551810383796692, + "num_tokens": 110486781.0, + "step": 2841 + }, + { + "epoch": 0.36153161175422976, + "grad_norm": 2.4767513275146484, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.8747163414955139, + "num_tokens": 110531932.0, + "step": 2842 + }, + { + "epoch": 0.36165882203282024, + "grad_norm": 3.114828586578369, + "learning_rate": 1e-06, + "loss": 0.4516, + "mean_token_accuracy": 0.8493112325668335, + "num_tokens": 110568139.0, + "step": 2843 + }, + { + "epoch": 0.36178603231141077, + "grad_norm": 2.7733266353607178, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8772307634353638, + "num_tokens": 110604142.0, + "step": 2844 + }, + { + "epoch": 0.3619132425900013, + "grad_norm": 2.806539297103882, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8753798604011536, + "num_tokens": 110643371.0, + "step": 2845 + }, + { + "epoch": 0.36204045286859177, + "grad_norm": 2.5792431831359863, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8887577056884766, + "num_tokens": 110678883.0, + "step": 2846 + }, + { + "epoch": 0.3621676631471823, + "grad_norm": 2.5918116569519043, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8733329176902771, + "num_tokens": 110719065.0, + "step": 2847 + }, + { + "epoch": 0.3622948734257728, + "grad_norm": 3.553712844848633, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8618718981742859, + "num_tokens": 110751854.0, + "step": 2848 + }, + { + "epoch": 0.3624220837043633, + "grad_norm": 2.516287326812744, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8838371634483337, + "num_tokens": 110791342.0, + "step": 2849 + }, + { + "epoch": 0.3625492939829538, + "grad_norm": 2.815016269683838, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8670092821121216, + "num_tokens": 110830978.0, + "step": 2850 + }, + { + "epoch": 0.36267650426154435, + "grad_norm": 2.758695363998413, + "learning_rate": 1e-06, + "loss": 0.3966, + "mean_token_accuracy": 0.8636048436164856, + "num_tokens": 110869278.0, + "step": 2851 + }, + { + "epoch": 0.3628037145401348, + "grad_norm": 2.7885568141937256, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8735055923461914, + "num_tokens": 110903679.0, + "step": 2852 + }, + { + "epoch": 0.36293092481872535, + "grad_norm": 2.6468992233276367, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8812249898910522, + "num_tokens": 110941669.0, + "step": 2853 + }, + { + "epoch": 0.3630581350973159, + "grad_norm": 2.961178779602051, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8765025734901428, + "num_tokens": 110972452.0, + "step": 2854 + }, + { + "epoch": 0.36318534537590635, + "grad_norm": 2.717571973800659, + "learning_rate": 1e-06, + "loss": 0.4031, + "mean_token_accuracy": 0.8630085587501526, + "num_tokens": 111012344.0, + "step": 2855 + }, + { + "epoch": 0.3633125556544969, + "grad_norm": 2.618103504180908, + "learning_rate": 1e-06, + "loss": 0.3806, + "mean_token_accuracy": 0.8653721213340759, + "num_tokens": 111053932.0, + "step": 2856 + }, + { + "epoch": 0.3634397659330874, + "grad_norm": 2.7378158569335938, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8695271015167236, + "num_tokens": 111091324.0, + "step": 2857 + }, + { + "epoch": 0.3635669762116779, + "grad_norm": 2.776742458343506, + "learning_rate": 1e-06, + "loss": 0.4183, + "mean_token_accuracy": 0.8525233268737793, + "num_tokens": 111131916.0, + "step": 2858 + }, + { + "epoch": 0.3636941864902684, + "grad_norm": 2.8581509590148926, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8672236800193787, + "num_tokens": 111167484.0, + "step": 2859 + }, + { + "epoch": 0.36382139676885894, + "grad_norm": 2.693415403366089, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8765634298324585, + "num_tokens": 111207502.0, + "step": 2860 + }, + { + "epoch": 0.3639486070474494, + "grad_norm": 2.5993359088897705, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8808824419975281, + "num_tokens": 111244218.0, + "step": 2861 + }, + { + "epoch": 0.36407581732603994, + "grad_norm": 2.753218650817871, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8705379962921143, + "num_tokens": 111280221.0, + "step": 2862 + }, + { + "epoch": 0.36420302760463047, + "grad_norm": 2.8346779346466064, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8696662187576294, + "num_tokens": 111317189.0, + "step": 2863 + }, + { + "epoch": 0.36433023788322094, + "grad_norm": 2.872215986251831, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8701549768447876, + "num_tokens": 111350564.0, + "step": 2864 + }, + { + "epoch": 0.36445744816181147, + "grad_norm": 2.8192760944366455, + "learning_rate": 1e-06, + "loss": 0.3881, + "mean_token_accuracy": 0.8669661283493042, + "num_tokens": 111391750.0, + "step": 2865 + }, + { + "epoch": 0.364584658440402, + "grad_norm": 2.8128957748413086, + "learning_rate": 1e-06, + "loss": 0.397, + "mean_token_accuracy": 0.8615564703941345, + "num_tokens": 111431159.0, + "step": 2866 + }, + { + "epoch": 0.36471186871899247, + "grad_norm": 2.6773219108581543, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8813058137893677, + "num_tokens": 111469121.0, + "step": 2867 + }, + { + "epoch": 0.364839078997583, + "grad_norm": 2.7330710887908936, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8873787522315979, + "num_tokens": 111502378.0, + "step": 2868 + }, + { + "epoch": 0.36496628927617353, + "grad_norm": 2.8648924827575684, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8710080981254578, + "num_tokens": 111541667.0, + "step": 2869 + }, + { + "epoch": 0.365093499554764, + "grad_norm": 2.543117046356201, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8935055732727051, + "num_tokens": 111580933.0, + "step": 2870 + }, + { + "epoch": 0.36522070983335453, + "grad_norm": 2.7172415256500244, + "learning_rate": 1e-06, + "loss": 0.4082, + "mean_token_accuracy": 0.8615175485610962, + "num_tokens": 111625807.0, + "step": 2871 + }, + { + "epoch": 0.36534792011194506, + "grad_norm": 2.6827285289764404, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8728524446487427, + "num_tokens": 111663495.0, + "step": 2872 + }, + { + "epoch": 0.36547513039053553, + "grad_norm": 2.8723936080932617, + "learning_rate": 1e-06, + "loss": 0.3955, + "mean_token_accuracy": 0.8635200262069702, + "num_tokens": 111699199.0, + "step": 2873 + }, + { + "epoch": 0.36560234066912606, + "grad_norm": 2.621466636657715, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8735561966896057, + "num_tokens": 111741463.0, + "step": 2874 + }, + { + "epoch": 0.3657295509477166, + "grad_norm": 2.7007997035980225, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.874264121055603, + "num_tokens": 111777806.0, + "step": 2875 + }, + { + "epoch": 0.36585676122630706, + "grad_norm": 2.6031758785247803, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8717060089111328, + "num_tokens": 111816988.0, + "step": 2876 + }, + { + "epoch": 0.3659839715048976, + "grad_norm": 2.6206657886505127, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.884040355682373, + "num_tokens": 111858088.0, + "step": 2877 + }, + { + "epoch": 0.3661111817834881, + "grad_norm": 2.3761167526245117, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8877701759338379, + "num_tokens": 111898289.0, + "step": 2878 + }, + { + "epoch": 0.3662383920620786, + "grad_norm": 2.451981544494629, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8704271912574768, + "num_tokens": 111942756.0, + "step": 2879 + }, + { + "epoch": 0.3663656023406691, + "grad_norm": 2.628133773803711, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8659517765045166, + "num_tokens": 111984025.0, + "step": 2880 + }, + { + "epoch": 0.36649281261925964, + "grad_norm": 2.421294689178467, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8790310621261597, + "num_tokens": 112029161.0, + "step": 2881 + }, + { + "epoch": 0.3666200228978502, + "grad_norm": 2.5062215328216553, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8713054060935974, + "num_tokens": 112071304.0, + "step": 2882 + }, + { + "epoch": 0.36674723317644065, + "grad_norm": 2.6160824298858643, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8928471207618713, + "num_tokens": 112107059.0, + "step": 2883 + }, + { + "epoch": 0.3668744434550312, + "grad_norm": 2.7245676517486572, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8821411728858948, + "num_tokens": 112141301.0, + "step": 2884 + }, + { + "epoch": 0.3670016537336217, + "grad_norm": 2.4527437686920166, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.885819673538208, + "num_tokens": 112181375.0, + "step": 2885 + }, + { + "epoch": 0.3671288640122122, + "grad_norm": 2.7176904678344727, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.884010910987854, + "num_tokens": 112217369.0, + "step": 2886 + }, + { + "epoch": 0.3672560742908027, + "grad_norm": 2.825892686843872, + "learning_rate": 1e-06, + "loss": 0.4109, + "mean_token_accuracy": 0.8577190637588501, + "num_tokens": 112257188.0, + "step": 2887 + }, + { + "epoch": 0.36738328456939323, + "grad_norm": 2.472931385040283, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8748149871826172, + "num_tokens": 112301286.0, + "step": 2888 + }, + { + "epoch": 0.3675104948479837, + "grad_norm": 2.8935444355010986, + "learning_rate": 1e-06, + "loss": 0.4038, + "mean_token_accuracy": 0.8583930730819702, + "num_tokens": 112336640.0, + "step": 2889 + }, + { + "epoch": 0.36763770512657423, + "grad_norm": 2.9368648529052734, + "learning_rate": 1e-06, + "loss": 0.385, + "mean_token_accuracy": 0.8665964603424072, + "num_tokens": 112374970.0, + "step": 2890 + }, + { + "epoch": 0.36776491540516476, + "grad_norm": 2.6704843044281006, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8780317306518555, + "num_tokens": 112411964.0, + "step": 2891 + }, + { + "epoch": 0.36789212568375523, + "grad_norm": 2.8854222297668457, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8785495758056641, + "num_tokens": 112446522.0, + "step": 2892 + }, + { + "epoch": 0.36801933596234576, + "grad_norm": 2.768267869949341, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.874654233455658, + "num_tokens": 112484541.0, + "step": 2893 + }, + { + "epoch": 0.3681465462409363, + "grad_norm": 2.540400266647339, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8859586119651794, + "num_tokens": 112525226.0, + "step": 2894 + }, + { + "epoch": 0.36827375651952676, + "grad_norm": 2.688025951385498, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8709882497787476, + "num_tokens": 112561891.0, + "step": 2895 + }, + { + "epoch": 0.3684009667981173, + "grad_norm": 2.5088415145874023, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8669877052307129, + "num_tokens": 112610426.0, + "step": 2896 + }, + { + "epoch": 0.3685281770767078, + "grad_norm": 2.8458874225616455, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8748294711112976, + "num_tokens": 112643318.0, + "step": 2897 + }, + { + "epoch": 0.3686553873552983, + "grad_norm": 2.7276318073272705, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8777877688407898, + "num_tokens": 112677802.0, + "step": 2898 + }, + { + "epoch": 0.3687825976338888, + "grad_norm": 2.80623722076416, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8662047982215881, + "num_tokens": 112715412.0, + "step": 2899 + }, + { + "epoch": 0.36890980791247935, + "grad_norm": 2.738863945007324, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8602937459945679, + "num_tokens": 112759234.0, + "step": 2900 + }, + { + "epoch": 0.3690370181910698, + "grad_norm": 2.7555155754089355, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8756858110427856, + "num_tokens": 112799440.0, + "step": 2901 + }, + { + "epoch": 0.36916422846966035, + "grad_norm": 2.900973081588745, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.871224045753479, + "num_tokens": 112835588.0, + "step": 2902 + }, + { + "epoch": 0.3692914387482509, + "grad_norm": 2.5363917350769043, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8792954683303833, + "num_tokens": 112876338.0, + "step": 2903 + }, + { + "epoch": 0.36941864902684135, + "grad_norm": 2.7622416019439697, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8708257675170898, + "num_tokens": 112912687.0, + "step": 2904 + }, + { + "epoch": 0.3695458593054319, + "grad_norm": 2.8535640239715576, + "learning_rate": 1e-06, + "loss": 0.4207, + "mean_token_accuracy": 0.8578320741653442, + "num_tokens": 112951119.0, + "step": 2905 + }, + { + "epoch": 0.3696730695840224, + "grad_norm": 2.814950942993164, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.8669019937515259, + "num_tokens": 112989125.0, + "step": 2906 + }, + { + "epoch": 0.3698002798626129, + "grad_norm": 2.7089426517486572, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8689217567443848, + "num_tokens": 113029034.0, + "step": 2907 + }, + { + "epoch": 0.3699274901412034, + "grad_norm": 4.283156871795654, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8725031614303589, + "num_tokens": 113068867.0, + "step": 2908 + }, + { + "epoch": 0.37005470041979394, + "grad_norm": 2.7215771675109863, + "learning_rate": 1e-06, + "loss": 0.4056, + "mean_token_accuracy": 0.8605523109436035, + "num_tokens": 113111218.0, + "step": 2909 + }, + { + "epoch": 0.3701819106983844, + "grad_norm": 2.840535879135132, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8723762035369873, + "num_tokens": 113148507.0, + "step": 2910 + }, + { + "epoch": 0.37030912097697494, + "grad_norm": 2.6609132289886475, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.881270170211792, + "num_tokens": 113184527.0, + "step": 2911 + }, + { + "epoch": 0.37043633125556547, + "grad_norm": 2.8645195960998535, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8669853806495667, + "num_tokens": 113220102.0, + "step": 2912 + }, + { + "epoch": 0.37056354153415594, + "grad_norm": 3.1167805194854736, + "learning_rate": 1e-06, + "loss": 0.4201, + "mean_token_accuracy": 0.8584945201873779, + "num_tokens": 113255057.0, + "step": 2913 + }, + { + "epoch": 0.37069075181274647, + "grad_norm": 2.819668769836426, + "learning_rate": 1e-06, + "loss": 0.3942, + "mean_token_accuracy": 0.8632661700248718, + "num_tokens": 113292904.0, + "step": 2914 + }, + { + "epoch": 0.370817962091337, + "grad_norm": 2.7699942588806152, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8747944235801697, + "num_tokens": 113329552.0, + "step": 2915 + }, + { + "epoch": 0.37094517236992747, + "grad_norm": 2.9283599853515625, + "learning_rate": 1e-06, + "loss": 0.3995, + "mean_token_accuracy": 0.8640046715736389, + "num_tokens": 113368905.0, + "step": 2916 + }, + { + "epoch": 0.371072382648518, + "grad_norm": 2.729161262512207, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8736624121665955, + "num_tokens": 113407799.0, + "step": 2917 + }, + { + "epoch": 0.3711995929271085, + "grad_norm": 2.60013484954834, + "learning_rate": 1e-06, + "loss": 0.4048, + "mean_token_accuracy": 0.861208438873291, + "num_tokens": 113454876.0, + "step": 2918 + }, + { + "epoch": 0.371326803205699, + "grad_norm": 2.6460793018341064, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8725048303604126, + "num_tokens": 113494860.0, + "step": 2919 + }, + { + "epoch": 0.3714540134842895, + "grad_norm": 2.664698839187622, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8746817111968994, + "num_tokens": 113539060.0, + "step": 2920 + }, + { + "epoch": 0.37158122376288005, + "grad_norm": 2.5984761714935303, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8803881406784058, + "num_tokens": 113575865.0, + "step": 2921 + }, + { + "epoch": 0.3717084340414705, + "grad_norm": 2.918041229248047, + "learning_rate": 1e-06, + "loss": 0.4406, + "mean_token_accuracy": 0.8494756817817688, + "num_tokens": 113616727.0, + "step": 2922 + }, + { + "epoch": 0.37183564432006105, + "grad_norm": 3.074178695678711, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8735712766647339, + "num_tokens": 113654516.0, + "step": 2923 + }, + { + "epoch": 0.3719628545986516, + "grad_norm": 3.0218920707702637, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8630236983299255, + "num_tokens": 113692126.0, + "step": 2924 + }, + { + "epoch": 0.37209006487724205, + "grad_norm": 2.598414897918701, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8882231116294861, + "num_tokens": 113729914.0, + "step": 2925 + }, + { + "epoch": 0.3722172751558326, + "grad_norm": 2.5945606231689453, + "learning_rate": 1e-06, + "loss": 0.4046, + "mean_token_accuracy": 0.8587969541549683, + "num_tokens": 113774976.0, + "step": 2926 + }, + { + "epoch": 0.3723444854344231, + "grad_norm": 2.5618960857391357, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8835666179656982, + "num_tokens": 113815844.0, + "step": 2927 + }, + { + "epoch": 0.3724716957130136, + "grad_norm": 2.7598752975463867, + "learning_rate": 1e-06, + "loss": 0.4138, + "mean_token_accuracy": 0.8557373285293579, + "num_tokens": 113858822.0, + "step": 2928 + }, + { + "epoch": 0.3725989059916041, + "grad_norm": 2.7933382987976074, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8740113377571106, + "num_tokens": 113894592.0, + "step": 2929 + }, + { + "epoch": 0.37272611627019464, + "grad_norm": 2.677438497543335, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8862147331237793, + "num_tokens": 113929939.0, + "step": 2930 + }, + { + "epoch": 0.37285332654878517, + "grad_norm": 2.832354784011841, + "learning_rate": 1e-06, + "loss": 0.4201, + "mean_token_accuracy": 0.8570654988288879, + "num_tokens": 113970073.0, + "step": 2931 + }, + { + "epoch": 0.37298053682737564, + "grad_norm": 2.769361972808838, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8763116598129272, + "num_tokens": 114007465.0, + "step": 2932 + }, + { + "epoch": 0.37310774710596617, + "grad_norm": 3.004762649536133, + "learning_rate": 1e-06, + "loss": 0.4117, + "mean_token_accuracy": 0.862067461013794, + "num_tokens": 114045388.0, + "step": 2933 + }, + { + "epoch": 0.3732349573845567, + "grad_norm": 2.885305166244507, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8807070255279541, + "num_tokens": 114080117.0, + "step": 2934 + }, + { + "epoch": 0.37336216766314717, + "grad_norm": 2.743445634841919, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8722872734069824, + "num_tokens": 114121454.0, + "step": 2935 + }, + { + "epoch": 0.3734893779417377, + "grad_norm": 2.5696115493774414, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8670873641967773, + "num_tokens": 114165392.0, + "step": 2936 + }, + { + "epoch": 0.3736165882203282, + "grad_norm": 2.469860315322876, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8799200654029846, + "num_tokens": 114213036.0, + "step": 2937 + }, + { + "epoch": 0.3737437984989187, + "grad_norm": 2.649207353591919, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.866769552230835, + "num_tokens": 114254993.0, + "step": 2938 + }, + { + "epoch": 0.37387100877750923, + "grad_norm": 2.744300127029419, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.872124433517456, + "num_tokens": 114292901.0, + "step": 2939 + }, + { + "epoch": 0.37399821905609976, + "grad_norm": 2.744072437286377, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8892634510993958, + "num_tokens": 114328094.0, + "step": 2940 + }, + { + "epoch": 0.37412542933469023, + "grad_norm": 2.593127965927124, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8851211667060852, + "num_tokens": 114365855.0, + "step": 2941 + }, + { + "epoch": 0.37425263961328076, + "grad_norm": 2.855036497116089, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8646029233932495, + "num_tokens": 114405145.0, + "step": 2942 + }, + { + "epoch": 0.3743798498918713, + "grad_norm": 2.741610050201416, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.872072696685791, + "num_tokens": 114442056.0, + "step": 2943 + }, + { + "epoch": 0.37450706017046176, + "grad_norm": 3.0383424758911133, + "learning_rate": 1e-06, + "loss": 0.409, + "mean_token_accuracy": 0.8597887754440308, + "num_tokens": 114476711.0, + "step": 2944 + }, + { + "epoch": 0.3746342704490523, + "grad_norm": 2.908604860305786, + "learning_rate": 1e-06, + "loss": 0.391, + "mean_token_accuracy": 0.8685914278030396, + "num_tokens": 114511325.0, + "step": 2945 + }, + { + "epoch": 0.3747614807276428, + "grad_norm": 3.1091301441192627, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8729251623153687, + "num_tokens": 114550663.0, + "step": 2946 + }, + { + "epoch": 0.3748886910062333, + "grad_norm": 2.6448636054992676, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8815733194351196, + "num_tokens": 114589301.0, + "step": 2947 + }, + { + "epoch": 0.3750159012848238, + "grad_norm": 2.9014015197753906, + "learning_rate": 1e-06, + "loss": 0.3964, + "mean_token_accuracy": 0.8629660606384277, + "num_tokens": 114628424.0, + "step": 2948 + }, + { + "epoch": 0.37514311156341434, + "grad_norm": 2.8421504497528076, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8690757751464844, + "num_tokens": 114666619.0, + "step": 2949 + }, + { + "epoch": 0.3752703218420048, + "grad_norm": 2.7928192615509033, + "learning_rate": 1e-06, + "loss": 0.4568, + "mean_token_accuracy": 0.8464339971542358, + "num_tokens": 114712008.0, + "step": 2950 + }, + { + "epoch": 0.37539753212059535, + "grad_norm": 2.720313310623169, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8746601939201355, + "num_tokens": 114751422.0, + "step": 2951 + }, + { + "epoch": 0.3755247423991859, + "grad_norm": 2.842042922973633, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.875672459602356, + "num_tokens": 114784541.0, + "step": 2952 + }, + { + "epoch": 0.37565195267777635, + "grad_norm": 2.7232375144958496, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.865490198135376, + "num_tokens": 114823617.0, + "step": 2953 + }, + { + "epoch": 0.3757791629563669, + "grad_norm": 2.746950149536133, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.877295970916748, + "num_tokens": 114858537.0, + "step": 2954 + }, + { + "epoch": 0.3759063732349574, + "grad_norm": 2.4735562801361084, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.880374550819397, + "num_tokens": 114898899.0, + "step": 2955 + }, + { + "epoch": 0.3760335835135479, + "grad_norm": 3.0246288776397705, + "learning_rate": 1e-06, + "loss": 0.4644, + "mean_token_accuracy": 0.8418849110603333, + "num_tokens": 114935678.0, + "step": 2956 + }, + { + "epoch": 0.3761607937921384, + "grad_norm": 2.5265913009643555, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8867586851119995, + "num_tokens": 114974501.0, + "step": 2957 + }, + { + "epoch": 0.37628800407072893, + "grad_norm": 2.564167022705078, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8718772530555725, + "num_tokens": 115016159.0, + "step": 2958 + }, + { + "epoch": 0.3764152143493194, + "grad_norm": 2.529616355895996, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8812274932861328, + "num_tokens": 115057379.0, + "step": 2959 + }, + { + "epoch": 0.37654242462790993, + "grad_norm": 2.578164577484131, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8694660663604736, + "num_tokens": 115101326.0, + "step": 2960 + }, + { + "epoch": 0.37666963490650046, + "grad_norm": 2.8601326942443848, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8711659908294678, + "num_tokens": 115138544.0, + "step": 2961 + }, + { + "epoch": 0.37679684518509093, + "grad_norm": 2.8438494205474854, + "learning_rate": 1e-06, + "loss": 0.4354, + "mean_token_accuracy": 0.8483926057815552, + "num_tokens": 115178588.0, + "step": 2962 + }, + { + "epoch": 0.37692405546368146, + "grad_norm": 2.5771331787109375, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8822969198226929, + "num_tokens": 115220951.0, + "step": 2963 + }, + { + "epoch": 0.377051265742272, + "grad_norm": 2.751913070678711, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8771542310714722, + "num_tokens": 115258684.0, + "step": 2964 + }, + { + "epoch": 0.37717847602086246, + "grad_norm": 2.606476306915283, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8758856058120728, + "num_tokens": 115302684.0, + "step": 2965 + }, + { + "epoch": 0.377305686299453, + "grad_norm": 2.5249855518341064, + "learning_rate": 1e-06, + "loss": 0.3926, + "mean_token_accuracy": 0.8670623898506165, + "num_tokens": 115350419.0, + "step": 2966 + }, + { + "epoch": 0.3774328965780435, + "grad_norm": 2.750469923019409, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8785791397094727, + "num_tokens": 115389726.0, + "step": 2967 + }, + { + "epoch": 0.377560106856634, + "grad_norm": 2.547987222671509, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8767156600952148, + "num_tokens": 115431848.0, + "step": 2968 + }, + { + "epoch": 0.3776873171352245, + "grad_norm": 2.6686925888061523, + "learning_rate": 1e-06, + "loss": 0.3911, + "mean_token_accuracy": 0.8654515743255615, + "num_tokens": 115472636.0, + "step": 2969 + }, + { + "epoch": 0.37781452741381505, + "grad_norm": 2.7415454387664795, + "learning_rate": 1e-06, + "loss": 0.3948, + "mean_token_accuracy": 0.8623939752578735, + "num_tokens": 115514301.0, + "step": 2970 + }, + { + "epoch": 0.3779417376924055, + "grad_norm": 2.642470121383667, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8722478747367859, + "num_tokens": 115558350.0, + "step": 2971 + }, + { + "epoch": 0.37806894797099605, + "grad_norm": 2.6759912967681885, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8692365288734436, + "num_tokens": 115597887.0, + "step": 2972 + }, + { + "epoch": 0.3781961582495866, + "grad_norm": 2.540208578109741, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8745986223220825, + "num_tokens": 115641058.0, + "step": 2973 + }, + { + "epoch": 0.37832336852817705, + "grad_norm": 2.782405376434326, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8696470260620117, + "num_tokens": 115679573.0, + "step": 2974 + }, + { + "epoch": 0.3784505788067676, + "grad_norm": 2.6318302154541016, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8678292036056519, + "num_tokens": 115724152.0, + "step": 2975 + }, + { + "epoch": 0.3785777890853581, + "grad_norm": 2.8841323852539062, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.870851993560791, + "num_tokens": 115756927.0, + "step": 2976 + }, + { + "epoch": 0.3787049993639486, + "grad_norm": 2.632366418838501, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8703360557556152, + "num_tokens": 115800480.0, + "step": 2977 + }, + { + "epoch": 0.3788322096425391, + "grad_norm": 3.0438873767852783, + "learning_rate": 1e-06, + "loss": 0.4013, + "mean_token_accuracy": 0.8643704652786255, + "num_tokens": 115832890.0, + "step": 2978 + }, + { + "epoch": 0.37895941992112964, + "grad_norm": 3.0385818481445312, + "learning_rate": 1e-06, + "loss": 0.3832, + "mean_token_accuracy": 0.8671299815177917, + "num_tokens": 115864832.0, + "step": 2979 + }, + { + "epoch": 0.37908663019972016, + "grad_norm": 2.6634981632232666, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8706961870193481, + "num_tokens": 115903996.0, + "step": 2980 + }, + { + "epoch": 0.37921384047831064, + "grad_norm": 2.778273820877075, + "learning_rate": 1e-06, + "loss": 0.4449, + "mean_token_accuracy": 0.8518835306167603, + "num_tokens": 115948025.0, + "step": 2981 + }, + { + "epoch": 0.37934105075690117, + "grad_norm": 2.6019649505615234, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8756842613220215, + "num_tokens": 115990579.0, + "step": 2982 + }, + { + "epoch": 0.3794682610354917, + "grad_norm": 2.681999921798706, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8658074140548706, + "num_tokens": 116037838.0, + "step": 2983 + }, + { + "epoch": 0.37959547131408217, + "grad_norm": 2.7568063735961914, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8777957558631897, + "num_tokens": 116075989.0, + "step": 2984 + }, + { + "epoch": 0.3797226815926727, + "grad_norm": 2.738887310028076, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.865776538848877, + "num_tokens": 116116817.0, + "step": 2985 + }, + { + "epoch": 0.3798498918712632, + "grad_norm": 2.609501361846924, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8753654956817627, + "num_tokens": 116159657.0, + "step": 2986 + }, + { + "epoch": 0.3799771021498537, + "grad_norm": 2.61727237701416, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8897659778594971, + "num_tokens": 116201729.0, + "step": 2987 + }, + { + "epoch": 0.3801043124284442, + "grad_norm": 2.811633825302124, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8694650530815125, + "num_tokens": 116237665.0, + "step": 2988 + }, + { + "epoch": 0.38023152270703475, + "grad_norm": 2.498858690261841, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8817548751831055, + "num_tokens": 116279404.0, + "step": 2989 + }, + { + "epoch": 0.3803587329856252, + "grad_norm": 2.6104788780212402, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8735966682434082, + "num_tokens": 116321689.0, + "step": 2990 + }, + { + "epoch": 0.38048594326421575, + "grad_norm": 2.716945171356201, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8683966398239136, + "num_tokens": 116359974.0, + "step": 2991 + }, + { + "epoch": 0.3806131535428063, + "grad_norm": 2.76611065864563, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8732663989067078, + "num_tokens": 116397260.0, + "step": 2992 + }, + { + "epoch": 0.38074036382139675, + "grad_norm": 2.8513684272766113, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8795335292816162, + "num_tokens": 116434644.0, + "step": 2993 + }, + { + "epoch": 0.3808675740999873, + "grad_norm": 2.684227228164673, + "learning_rate": 1e-06, + "loss": 0.4145, + "mean_token_accuracy": 0.8595180511474609, + "num_tokens": 116476162.0, + "step": 2994 + }, + { + "epoch": 0.3809947843785778, + "grad_norm": 2.7833571434020996, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8838348984718323, + "num_tokens": 116510493.0, + "step": 2995 + }, + { + "epoch": 0.3811219946571683, + "grad_norm": 2.5772836208343506, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8710687160491943, + "num_tokens": 116552129.0, + "step": 2996 + }, + { + "epoch": 0.3812492049357588, + "grad_norm": 2.789679765701294, + "learning_rate": 1e-06, + "loss": 0.4022, + "mean_token_accuracy": 0.8644726872444153, + "num_tokens": 116591944.0, + "step": 2997 + }, + { + "epoch": 0.38137641521434934, + "grad_norm": 2.634312629699707, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8760400414466858, + "num_tokens": 116631084.0, + "step": 2998 + }, + { + "epoch": 0.3815036254929398, + "grad_norm": 2.6500205993652344, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8715758323669434, + "num_tokens": 116670605.0, + "step": 2999 + }, + { + "epoch": 0.38163083577153034, + "grad_norm": 2.5941948890686035, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8696759939193726, + "num_tokens": 116712808.0, + "step": 3000 + }, + { + "epoch": 0.38175804605012087, + "grad_norm": 2.6205434799194336, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8854063749313354, + "num_tokens": 116752114.0, + "step": 3001 + }, + { + "epoch": 0.38188525632871134, + "grad_norm": 2.6015024185180664, + "learning_rate": 1e-06, + "loss": 0.3942, + "mean_token_accuracy": 0.8653395175933838, + "num_tokens": 116793955.0, + "step": 3002 + }, + { + "epoch": 0.38201246660730187, + "grad_norm": 2.7467708587646484, + "learning_rate": 1e-06, + "loss": 0.3863, + "mean_token_accuracy": 0.8665722608566284, + "num_tokens": 116832822.0, + "step": 3003 + }, + { + "epoch": 0.3821396768858924, + "grad_norm": 2.587890625, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8858555555343628, + "num_tokens": 116872993.0, + "step": 3004 + }, + { + "epoch": 0.38226688716448287, + "grad_norm": 2.5737240314483643, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.872664749622345, + "num_tokens": 116914826.0, + "step": 3005 + }, + { + "epoch": 0.3823940974430734, + "grad_norm": 3.268488645553589, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.866938054561615, + "num_tokens": 116954793.0, + "step": 3006 + }, + { + "epoch": 0.38252130772166393, + "grad_norm": 2.6559507846832275, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.866972029209137, + "num_tokens": 116996257.0, + "step": 3007 + }, + { + "epoch": 0.3826485180002544, + "grad_norm": 2.7120654582977295, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8640028238296509, + "num_tokens": 117037643.0, + "step": 3008 + }, + { + "epoch": 0.38277572827884493, + "grad_norm": 2.499570608139038, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8767117261886597, + "num_tokens": 117083022.0, + "step": 3009 + }, + { + "epoch": 0.38290293855743546, + "grad_norm": 2.7624642848968506, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8846195936203003, + "num_tokens": 117118486.0, + "step": 3010 + }, + { + "epoch": 0.38303014883602593, + "grad_norm": 2.8340795040130615, + "learning_rate": 1e-06, + "loss": 0.3911, + "mean_token_accuracy": 0.8627271056175232, + "num_tokens": 117155240.0, + "step": 3011 + }, + { + "epoch": 0.38315735911461646, + "grad_norm": 2.7124812602996826, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8776111006736755, + "num_tokens": 117191212.0, + "step": 3012 + }, + { + "epoch": 0.383284569393207, + "grad_norm": 2.8991825580596924, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8731706142425537, + "num_tokens": 117221902.0, + "step": 3013 + }, + { + "epoch": 0.38341177967179746, + "grad_norm": 2.7709827423095703, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8883886337280273, + "num_tokens": 117256532.0, + "step": 3014 + }, + { + "epoch": 0.383538989950388, + "grad_norm": 2.6962971687316895, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.8743627071380615, + "num_tokens": 117297171.0, + "step": 3015 + }, + { + "epoch": 0.3836662002289785, + "grad_norm": 2.905320167541504, + "learning_rate": 1e-06, + "loss": 0.4278, + "mean_token_accuracy": 0.8546477556228638, + "num_tokens": 117335082.0, + "step": 3016 + }, + { + "epoch": 0.383793410507569, + "grad_norm": 2.7020790576934814, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8717229962348938, + "num_tokens": 117375622.0, + "step": 3017 + }, + { + "epoch": 0.3839206207861595, + "grad_norm": 2.836050033569336, + "learning_rate": 1e-06, + "loss": 0.4136, + "mean_token_accuracy": 0.8618132472038269, + "num_tokens": 117412966.0, + "step": 3018 + }, + { + "epoch": 0.38404783106475004, + "grad_norm": 2.783350944519043, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.882161557674408, + "num_tokens": 117451329.0, + "step": 3019 + }, + { + "epoch": 0.3841750413433405, + "grad_norm": 2.6059939861297607, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8744354248046875, + "num_tokens": 117492238.0, + "step": 3020 + }, + { + "epoch": 0.38430225162193105, + "grad_norm": 2.5288949012756348, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8751809597015381, + "num_tokens": 117533657.0, + "step": 3021 + }, + { + "epoch": 0.3844294619005216, + "grad_norm": 2.7239692211151123, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8821731805801392, + "num_tokens": 117570913.0, + "step": 3022 + }, + { + "epoch": 0.38455667217911205, + "grad_norm": 2.274857997894287, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8893564939498901, + "num_tokens": 117616846.0, + "step": 3023 + }, + { + "epoch": 0.3846838824577026, + "grad_norm": 2.6241047382354736, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8781280517578125, + "num_tokens": 117656198.0, + "step": 3024 + }, + { + "epoch": 0.3848110927362931, + "grad_norm": 2.9384515285491943, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.881312906742096, + "num_tokens": 117690711.0, + "step": 3025 + }, + { + "epoch": 0.3849383030148836, + "grad_norm": 2.6391890048980713, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8616535663604736, + "num_tokens": 117731687.0, + "step": 3026 + }, + { + "epoch": 0.3850655132934741, + "grad_norm": 2.7578694820404053, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8868976831436157, + "num_tokens": 117768353.0, + "step": 3027 + }, + { + "epoch": 0.38519272357206463, + "grad_norm": 2.567903518676758, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8842253684997559, + "num_tokens": 117806671.0, + "step": 3028 + }, + { + "epoch": 0.3853199338506551, + "grad_norm": 2.55015230178833, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.883472740650177, + "num_tokens": 117849003.0, + "step": 3029 + }, + { + "epoch": 0.38544714412924563, + "grad_norm": 2.7925331592559814, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8788466453552246, + "num_tokens": 117884035.0, + "step": 3030 + }, + { + "epoch": 0.38557435440783616, + "grad_norm": 2.845931053161621, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.868653416633606, + "num_tokens": 117923616.0, + "step": 3031 + }, + { + "epoch": 0.3857015646864267, + "grad_norm": 2.7335731983184814, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8742128610610962, + "num_tokens": 117959051.0, + "step": 3032 + }, + { + "epoch": 0.38582877496501716, + "grad_norm": 2.96053409576416, + "learning_rate": 1e-06, + "loss": 0.3959, + "mean_token_accuracy": 0.8628362417221069, + "num_tokens": 117995499.0, + "step": 3033 + }, + { + "epoch": 0.3859559852436077, + "grad_norm": 2.790492296218872, + "learning_rate": 1e-06, + "loss": 0.4311, + "mean_token_accuracy": 0.8512558341026306, + "num_tokens": 118035623.0, + "step": 3034 + }, + { + "epoch": 0.3860831955221982, + "grad_norm": 2.9170351028442383, + "learning_rate": 1e-06, + "loss": 0.4125, + "mean_token_accuracy": 0.8556758165359497, + "num_tokens": 118076525.0, + "step": 3035 + }, + { + "epoch": 0.3862104058007887, + "grad_norm": 2.7238786220550537, + "learning_rate": 1e-06, + "loss": 0.3919, + "mean_token_accuracy": 0.8651418685913086, + "num_tokens": 118117300.0, + "step": 3036 + }, + { + "epoch": 0.3863376160793792, + "grad_norm": 2.8875091075897217, + "learning_rate": 1e-06, + "loss": 0.4279, + "mean_token_accuracy": 0.8563290238380432, + "num_tokens": 118157719.0, + "step": 3037 + }, + { + "epoch": 0.38646482635796975, + "grad_norm": 2.806727886199951, + "learning_rate": 1e-06, + "loss": 0.4134, + "mean_token_accuracy": 0.8593364357948303, + "num_tokens": 118195408.0, + "step": 3038 + }, + { + "epoch": 0.3865920366365602, + "grad_norm": 3.0467493534088135, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8717892169952393, + "num_tokens": 118227670.0, + "step": 3039 + }, + { + "epoch": 0.38671924691515075, + "grad_norm": 2.5571885108947754, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8756920695304871, + "num_tokens": 118268696.0, + "step": 3040 + }, + { + "epoch": 0.3868464571937413, + "grad_norm": 2.5256130695343018, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8812375664710999, + "num_tokens": 118310535.0, + "step": 3041 + }, + { + "epoch": 0.38697366747233175, + "grad_norm": 2.83717679977417, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8911933302879333, + "num_tokens": 118342336.0, + "step": 3042 + }, + { + "epoch": 0.3871008777509223, + "grad_norm": 2.7748751640319824, + "learning_rate": 1e-06, + "loss": 0.41, + "mean_token_accuracy": 0.8609550595283508, + "num_tokens": 118383058.0, + "step": 3043 + }, + { + "epoch": 0.3872280880295128, + "grad_norm": 2.634031295776367, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.8652591705322266, + "num_tokens": 118425809.0, + "step": 3044 + }, + { + "epoch": 0.3873552983081033, + "grad_norm": 2.6054203510284424, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8714182376861572, + "num_tokens": 118465927.0, + "step": 3045 + }, + { + "epoch": 0.3874825085866938, + "grad_norm": 2.649822950363159, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8777389526367188, + "num_tokens": 118503947.0, + "step": 3046 + }, + { + "epoch": 0.38760971886528434, + "grad_norm": 2.595916271209717, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8671740293502808, + "num_tokens": 118547246.0, + "step": 3047 + }, + { + "epoch": 0.3877369291438748, + "grad_norm": 2.806605100631714, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.8702048063278198, + "num_tokens": 118582062.0, + "step": 3048 + }, + { + "epoch": 0.38786413942246534, + "grad_norm": 2.9520039558410645, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8785111904144287, + "num_tokens": 118615610.0, + "step": 3049 + }, + { + "epoch": 0.38799134970105587, + "grad_norm": 2.74554705619812, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8738133907318115, + "num_tokens": 118652519.0, + "step": 3050 + }, + { + "epoch": 0.38811855997964634, + "grad_norm": 2.7444159984588623, + "learning_rate": 1e-06, + "loss": 0.4127, + "mean_token_accuracy": 0.8621716499328613, + "num_tokens": 118690683.0, + "step": 3051 + }, + { + "epoch": 0.38824577025823687, + "grad_norm": 2.4791605472564697, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8884892463684082, + "num_tokens": 118731443.0, + "step": 3052 + }, + { + "epoch": 0.3883729805368274, + "grad_norm": 2.818966865539551, + "learning_rate": 1e-06, + "loss": 0.3992, + "mean_token_accuracy": 0.8616171479225159, + "num_tokens": 118769559.0, + "step": 3053 + }, + { + "epoch": 0.38850019081541787, + "grad_norm": 2.63228702545166, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8719191551208496, + "num_tokens": 118808696.0, + "step": 3054 + }, + { + "epoch": 0.3886274010940084, + "grad_norm": 2.7950973510742188, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8722723722457886, + "num_tokens": 118842176.0, + "step": 3055 + }, + { + "epoch": 0.3887546113725989, + "grad_norm": 2.771494150161743, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8702410459518433, + "num_tokens": 118878643.0, + "step": 3056 + }, + { + "epoch": 0.3888818216511894, + "grad_norm": 2.999736785888672, + "learning_rate": 1e-06, + "loss": 0.3876, + "mean_token_accuracy": 0.8650515079498291, + "num_tokens": 118915055.0, + "step": 3057 + }, + { + "epoch": 0.3890090319297799, + "grad_norm": 2.658714532852173, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8678680658340454, + "num_tokens": 118957449.0, + "step": 3058 + }, + { + "epoch": 0.38913624220837045, + "grad_norm": 2.9349400997161865, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8705875873565674, + "num_tokens": 118994100.0, + "step": 3059 + }, + { + "epoch": 0.3892634524869609, + "grad_norm": 2.627589464187622, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.863447368144989, + "num_tokens": 119036404.0, + "step": 3060 + }, + { + "epoch": 0.38939066276555145, + "grad_norm": 4.012998104095459, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8797170519828796, + "num_tokens": 119079405.0, + "step": 3061 + }, + { + "epoch": 0.389517873044142, + "grad_norm": 2.553314685821533, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8750708103179932, + "num_tokens": 119121973.0, + "step": 3062 + }, + { + "epoch": 0.38964508332273246, + "grad_norm": 2.873102903366089, + "learning_rate": 1e-06, + "loss": 0.4396, + "mean_token_accuracy": 0.8458558917045593, + "num_tokens": 119162684.0, + "step": 3063 + }, + { + "epoch": 0.389772293601323, + "grad_norm": 2.7080163955688477, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8777886033058167, + "num_tokens": 119197161.0, + "step": 3064 + }, + { + "epoch": 0.3898995038799135, + "grad_norm": 2.893477439880371, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8723773956298828, + "num_tokens": 119234657.0, + "step": 3065 + }, + { + "epoch": 0.390026714158504, + "grad_norm": 3.036409616470337, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8759214282035828, + "num_tokens": 119267628.0, + "step": 3066 + }, + { + "epoch": 0.3901539244370945, + "grad_norm": 2.7453935146331787, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8731105327606201, + "num_tokens": 119306352.0, + "step": 3067 + }, + { + "epoch": 0.39028113471568504, + "grad_norm": 2.644843339920044, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8724052906036377, + "num_tokens": 119345910.0, + "step": 3068 + }, + { + "epoch": 0.3904083449942755, + "grad_norm": 2.8046493530273438, + "learning_rate": 1e-06, + "loss": 0.4115, + "mean_token_accuracy": 0.8603789806365967, + "num_tokens": 119385555.0, + "step": 3069 + }, + { + "epoch": 0.39053555527286604, + "grad_norm": 2.672682046890259, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8743821382522583, + "num_tokens": 119425663.0, + "step": 3070 + }, + { + "epoch": 0.39066276555145657, + "grad_norm": 2.836019992828369, + "learning_rate": 1e-06, + "loss": 0.4386, + "mean_token_accuracy": 0.8455204367637634, + "num_tokens": 119467814.0, + "step": 3071 + }, + { + "epoch": 0.39078997583004704, + "grad_norm": 2.7208213806152344, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8754969835281372, + "num_tokens": 119504652.0, + "step": 3072 + }, + { + "epoch": 0.39091718610863757, + "grad_norm": 2.7285029888153076, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.869991660118103, + "num_tokens": 119544056.0, + "step": 3073 + }, + { + "epoch": 0.3910443963872281, + "grad_norm": 2.815535068511963, + "learning_rate": 1e-06, + "loss": 0.4204, + "mean_token_accuracy": 0.855840802192688, + "num_tokens": 119583990.0, + "step": 3074 + }, + { + "epoch": 0.39117160666581857, + "grad_norm": 2.7327635288238525, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8872050046920776, + "num_tokens": 119618401.0, + "step": 3075 + }, + { + "epoch": 0.3912988169444091, + "grad_norm": 2.751917839050293, + "learning_rate": 1e-06, + "loss": 0.3949, + "mean_token_accuracy": 0.8652924299240112, + "num_tokens": 119658622.0, + "step": 3076 + }, + { + "epoch": 0.39142602722299963, + "grad_norm": 2.7630395889282227, + "learning_rate": 1e-06, + "loss": 0.4113, + "mean_token_accuracy": 0.861143171787262, + "num_tokens": 119701720.0, + "step": 3077 + }, + { + "epoch": 0.3915532375015901, + "grad_norm": 2.6157915592193604, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8639232516288757, + "num_tokens": 119743968.0, + "step": 3078 + }, + { + "epoch": 0.39168044778018063, + "grad_norm": 2.543907642364502, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8914414048194885, + "num_tokens": 119783604.0, + "step": 3079 + }, + { + "epoch": 0.39180765805877116, + "grad_norm": 2.9224259853363037, + "learning_rate": 1e-06, + "loss": 0.4025, + "mean_token_accuracy": 0.857708215713501, + "num_tokens": 119819994.0, + "step": 3080 + }, + { + "epoch": 0.3919348683373617, + "grad_norm": 2.7646501064300537, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.864248514175415, + "num_tokens": 119856520.0, + "step": 3081 + }, + { + "epoch": 0.39206207861595216, + "grad_norm": 2.5734803676605225, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8909165263175964, + "num_tokens": 119894059.0, + "step": 3082 + }, + { + "epoch": 0.3921892888945427, + "grad_norm": 2.7700865268707275, + "learning_rate": 1e-06, + "loss": 0.3786, + "mean_token_accuracy": 0.8694562911987305, + "num_tokens": 119931859.0, + "step": 3083 + }, + { + "epoch": 0.3923164991731332, + "grad_norm": 2.898360013961792, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.884361207485199, + "num_tokens": 119967275.0, + "step": 3084 + }, + { + "epoch": 0.3924437094517237, + "grad_norm": 2.7528560161590576, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8685942888259888, + "num_tokens": 120002163.0, + "step": 3085 + }, + { + "epoch": 0.3925709197303142, + "grad_norm": 2.647261619567871, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.8672182559967041, + "num_tokens": 120042938.0, + "step": 3086 + }, + { + "epoch": 0.39269813000890474, + "grad_norm": 2.7025041580200195, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8822849988937378, + "num_tokens": 120082474.0, + "step": 3087 + }, + { + "epoch": 0.3928253402874952, + "grad_norm": 2.938401699066162, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8720035552978516, + "num_tokens": 120117074.0, + "step": 3088 + }, + { + "epoch": 0.39295255056608575, + "grad_norm": 2.6669061183929443, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8878841400146484, + "num_tokens": 120153096.0, + "step": 3089 + }, + { + "epoch": 0.3930797608446763, + "grad_norm": 2.8992433547973633, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.8711706399917603, + "num_tokens": 120190180.0, + "step": 3090 + }, + { + "epoch": 0.39320697112326675, + "grad_norm": 2.917239189147949, + "learning_rate": 1e-06, + "loss": 0.4154, + "mean_token_accuracy": 0.8573459386825562, + "num_tokens": 120229690.0, + "step": 3091 + }, + { + "epoch": 0.3933341814018573, + "grad_norm": 2.5222415924072266, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8903487920761108, + "num_tokens": 120271074.0, + "step": 3092 + }, + { + "epoch": 0.3934613916804478, + "grad_norm": 2.861685276031494, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8671380281448364, + "num_tokens": 120307232.0, + "step": 3093 + }, + { + "epoch": 0.3935886019590383, + "grad_norm": 2.848860740661621, + "learning_rate": 1e-06, + "loss": 0.3979, + "mean_token_accuracy": 0.8639304041862488, + "num_tokens": 120343617.0, + "step": 3094 + }, + { + "epoch": 0.3937158122376288, + "grad_norm": 2.6725172996520996, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8727750778198242, + "num_tokens": 120383845.0, + "step": 3095 + }, + { + "epoch": 0.39384302251621933, + "grad_norm": 2.543609142303467, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8748847246170044, + "num_tokens": 120426269.0, + "step": 3096 + }, + { + "epoch": 0.3939702327948098, + "grad_norm": 2.484391689300537, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8716878890991211, + "num_tokens": 120471280.0, + "step": 3097 + }, + { + "epoch": 0.39409744307340033, + "grad_norm": 2.850695848464966, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8776464462280273, + "num_tokens": 120505704.0, + "step": 3098 + }, + { + "epoch": 0.39422465335199086, + "grad_norm": 2.6503543853759766, + "learning_rate": 1e-06, + "loss": 0.3956, + "mean_token_accuracy": 0.8625500202178955, + "num_tokens": 120547215.0, + "step": 3099 + }, + { + "epoch": 0.39435186363058133, + "grad_norm": 2.7248101234436035, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8790351748466492, + "num_tokens": 120581389.0, + "step": 3100 + }, + { + "epoch": 0.39447907390917186, + "grad_norm": 2.6095473766326904, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8645915985107422, + "num_tokens": 120621351.0, + "step": 3101 + }, + { + "epoch": 0.3946062841877624, + "grad_norm": 2.648334264755249, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8749542236328125, + "num_tokens": 120662335.0, + "step": 3102 + }, + { + "epoch": 0.39473349446635286, + "grad_norm": 2.511052131652832, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8770864605903625, + "num_tokens": 120705569.0, + "step": 3103 + }, + { + "epoch": 0.3948607047449434, + "grad_norm": 2.718066453933716, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.8743907809257507, + "num_tokens": 120746518.0, + "step": 3104 + }, + { + "epoch": 0.3949879150235339, + "grad_norm": 2.388085126876831, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8859950304031372, + "num_tokens": 120791441.0, + "step": 3105 + }, + { + "epoch": 0.3951151253021244, + "grad_norm": 2.6132757663726807, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8748117685317993, + "num_tokens": 120830169.0, + "step": 3106 + }, + { + "epoch": 0.3952423355807149, + "grad_norm": 2.9959471225738525, + "learning_rate": 1e-06, + "loss": 0.4369, + "mean_token_accuracy": 0.8510444164276123, + "num_tokens": 120870426.0, + "step": 3107 + }, + { + "epoch": 0.39536954585930545, + "grad_norm": 2.7647886276245117, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.881948709487915, + "num_tokens": 120907316.0, + "step": 3108 + }, + { + "epoch": 0.3954967561378959, + "grad_norm": 2.83695387840271, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.872901201248169, + "num_tokens": 120942531.0, + "step": 3109 + }, + { + "epoch": 0.39562396641648645, + "grad_norm": 2.440016031265259, + "learning_rate": 1e-06, + "loss": 0.3835, + "mean_token_accuracy": 0.867400050163269, + "num_tokens": 120988196.0, + "step": 3110 + }, + { + "epoch": 0.395751176695077, + "grad_norm": 2.6590232849121094, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.8662626147270203, + "num_tokens": 121029384.0, + "step": 3111 + }, + { + "epoch": 0.39587838697366745, + "grad_norm": 2.568356990814209, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8745155334472656, + "num_tokens": 121070412.0, + "step": 3112 + }, + { + "epoch": 0.396005597252258, + "grad_norm": 3.1047489643096924, + "learning_rate": 1e-06, + "loss": 0.3925, + "mean_token_accuracy": 0.866569995880127, + "num_tokens": 121102458.0, + "step": 3113 + }, + { + "epoch": 0.3961328075308485, + "grad_norm": 2.722507953643799, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8713558316230774, + "num_tokens": 121138197.0, + "step": 3114 + }, + { + "epoch": 0.396260017809439, + "grad_norm": 3.228269100189209, + "learning_rate": 1e-06, + "loss": 0.4084, + "mean_token_accuracy": 0.8646981716156006, + "num_tokens": 121170774.0, + "step": 3115 + }, + { + "epoch": 0.3963872280880295, + "grad_norm": 2.7668659687042236, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8801564574241638, + "num_tokens": 121205708.0, + "step": 3116 + }, + { + "epoch": 0.39651443836662004, + "grad_norm": 2.7133429050445557, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8746066689491272, + "num_tokens": 121242477.0, + "step": 3117 + }, + { + "epoch": 0.3966416486452105, + "grad_norm": 2.719651460647583, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.895827054977417, + "num_tokens": 121274753.0, + "step": 3118 + }, + { + "epoch": 0.39676885892380104, + "grad_norm": 2.6499202251434326, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8833795189857483, + "num_tokens": 121313701.0, + "step": 3119 + }, + { + "epoch": 0.39689606920239157, + "grad_norm": 2.7664034366607666, + "learning_rate": 1e-06, + "loss": 0.4024, + "mean_token_accuracy": 0.8620598316192627, + "num_tokens": 121356389.0, + "step": 3120 + }, + { + "epoch": 0.39702327948098204, + "grad_norm": 2.3298275470733643, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8890434503555298, + "num_tokens": 121394611.0, + "step": 3121 + }, + { + "epoch": 0.39715048975957257, + "grad_norm": 2.878272533416748, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8700481653213501, + "num_tokens": 121428847.0, + "step": 3122 + }, + { + "epoch": 0.3972777000381631, + "grad_norm": 2.653470039367676, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8714455366134644, + "num_tokens": 121471128.0, + "step": 3123 + }, + { + "epoch": 0.39740491031675357, + "grad_norm": 2.512019395828247, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8663408756256104, + "num_tokens": 121514038.0, + "step": 3124 + }, + { + "epoch": 0.3975321205953441, + "grad_norm": 2.776740789413452, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8683044910430908, + "num_tokens": 121555541.0, + "step": 3125 + }, + { + "epoch": 0.3976593308739346, + "grad_norm": 2.949673891067505, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8660092949867249, + "num_tokens": 121589841.0, + "step": 3126 + }, + { + "epoch": 0.3977865411525251, + "grad_norm": 2.5663440227508545, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8942992091178894, + "num_tokens": 121626287.0, + "step": 3127 + }, + { + "epoch": 0.3979137514311156, + "grad_norm": 2.423729658126831, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8761143684387207, + "num_tokens": 121670817.0, + "step": 3128 + }, + { + "epoch": 0.39804096170970615, + "grad_norm": 2.632850170135498, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8774616718292236, + "num_tokens": 121711647.0, + "step": 3129 + }, + { + "epoch": 0.3981681719882967, + "grad_norm": 2.7556164264678955, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8892161250114441, + "num_tokens": 121743842.0, + "step": 3130 + }, + { + "epoch": 0.39829538226688715, + "grad_norm": 2.6576991081237793, + "learning_rate": 1e-06, + "loss": 0.4027, + "mean_token_accuracy": 0.8652709126472473, + "num_tokens": 121788648.0, + "step": 3131 + }, + { + "epoch": 0.3984225925454777, + "grad_norm": 3.0405404567718506, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8672882318496704, + "num_tokens": 121822981.0, + "step": 3132 + }, + { + "epoch": 0.3985498028240682, + "grad_norm": 2.4650864601135254, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8878398537635803, + "num_tokens": 121867506.0, + "step": 3133 + }, + { + "epoch": 0.3986770131026587, + "grad_norm": 2.775965929031372, + "learning_rate": 1e-06, + "loss": 0.3886, + "mean_token_accuracy": 0.8623254299163818, + "num_tokens": 121906857.0, + "step": 3134 + }, + { + "epoch": 0.3988042233812492, + "grad_norm": 3.3210771083831787, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8657268285751343, + "num_tokens": 121944196.0, + "step": 3135 + }, + { + "epoch": 0.39893143365983974, + "grad_norm": 2.7528138160705566, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8810006976127625, + "num_tokens": 121979901.0, + "step": 3136 + }, + { + "epoch": 0.3990586439384302, + "grad_norm": 2.7618424892425537, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8687493801116943, + "num_tokens": 122018618.0, + "step": 3137 + }, + { + "epoch": 0.39918585421702074, + "grad_norm": 2.622948408126831, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8858739137649536, + "num_tokens": 122058472.0, + "step": 3138 + }, + { + "epoch": 0.39931306449561127, + "grad_norm": 2.459981918334961, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.889404296875, + "num_tokens": 122099906.0, + "step": 3139 + }, + { + "epoch": 0.39944027477420174, + "grad_norm": 2.7973744869232178, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8698405027389526, + "num_tokens": 122136289.0, + "step": 3140 + }, + { + "epoch": 0.39956748505279227, + "grad_norm": 2.6087734699249268, + "learning_rate": 1e-06, + "loss": 0.3786, + "mean_token_accuracy": 0.8669070601463318, + "num_tokens": 122175318.0, + "step": 3141 + }, + { + "epoch": 0.3996946953313828, + "grad_norm": 2.5857417583465576, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8760345578193665, + "num_tokens": 122216813.0, + "step": 3142 + }, + { + "epoch": 0.39982190560997327, + "grad_norm": 2.938275098800659, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8670966625213623, + "num_tokens": 122253524.0, + "step": 3143 + }, + { + "epoch": 0.3999491158885638, + "grad_norm": 2.6496121883392334, + "learning_rate": 1e-06, + "loss": 0.4134, + "mean_token_accuracy": 0.8592572212219238, + "num_tokens": 122298382.0, + "step": 3144 + }, + { + "epoch": 0.40007632616715433, + "grad_norm": 2.8533594608306885, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8689064979553223, + "num_tokens": 122335671.0, + "step": 3145 + }, + { + "epoch": 0.4002035364457448, + "grad_norm": 2.4032559394836426, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8698569536209106, + "num_tokens": 122381546.0, + "step": 3146 + }, + { + "epoch": 0.40033074672433533, + "grad_norm": 2.794368028640747, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.8588187098503113, + "num_tokens": 122417984.0, + "step": 3147 + }, + { + "epoch": 0.40045795700292586, + "grad_norm": 2.502626895904541, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8777878284454346, + "num_tokens": 122459885.0, + "step": 3148 + }, + { + "epoch": 0.40058516728151633, + "grad_norm": 2.917279005050659, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8720651268959045, + "num_tokens": 122493489.0, + "step": 3149 + }, + { + "epoch": 0.40071237756010686, + "grad_norm": 2.788208484649658, + "learning_rate": 1e-06, + "loss": 0.3936, + "mean_token_accuracy": 0.8607548475265503, + "num_tokens": 122529615.0, + "step": 3150 + }, + { + "epoch": 0.4008395878386974, + "grad_norm": 2.9507393836975098, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8672342896461487, + "num_tokens": 122565881.0, + "step": 3151 + }, + { + "epoch": 0.40096679811728786, + "grad_norm": 2.8030216693878174, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8729913234710693, + "num_tokens": 122600949.0, + "step": 3152 + }, + { + "epoch": 0.4010940083958784, + "grad_norm": 2.5778894424438477, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8733516931533813, + "num_tokens": 122644297.0, + "step": 3153 + }, + { + "epoch": 0.4012212186744689, + "grad_norm": 2.5341243743896484, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.869598925113678, + "num_tokens": 122690654.0, + "step": 3154 + }, + { + "epoch": 0.4013484289530594, + "grad_norm": 3.0022573471069336, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8674171566963196, + "num_tokens": 122725585.0, + "step": 3155 + }, + { + "epoch": 0.4014756392316499, + "grad_norm": 2.7601916790008545, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8718674182891846, + "num_tokens": 122761759.0, + "step": 3156 + }, + { + "epoch": 0.40160284951024044, + "grad_norm": 2.716374635696411, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.866812527179718, + "num_tokens": 122801837.0, + "step": 3157 + }, + { + "epoch": 0.4017300597888309, + "grad_norm": 2.6141364574432373, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8744697570800781, + "num_tokens": 122843530.0, + "step": 3158 + }, + { + "epoch": 0.40185727006742145, + "grad_norm": 2.663888931274414, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8738983869552612, + "num_tokens": 122882310.0, + "step": 3159 + }, + { + "epoch": 0.401984480346012, + "grad_norm": 2.691336154937744, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8749045133590698, + "num_tokens": 122919376.0, + "step": 3160 + }, + { + "epoch": 0.40211169062460245, + "grad_norm": 2.6993069648742676, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8752995729446411, + "num_tokens": 122958680.0, + "step": 3161 + }, + { + "epoch": 0.402238900903193, + "grad_norm": 2.698280096054077, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.8673866391181946, + "num_tokens": 122997236.0, + "step": 3162 + }, + { + "epoch": 0.4023661111817835, + "grad_norm": 2.770630121231079, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.866216778755188, + "num_tokens": 123033830.0, + "step": 3163 + }, + { + "epoch": 0.402493321460374, + "grad_norm": 2.645782709121704, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8758072853088379, + "num_tokens": 123073770.0, + "step": 3164 + }, + { + "epoch": 0.4026205317389645, + "grad_norm": 2.685844898223877, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8765624761581421, + "num_tokens": 123113617.0, + "step": 3165 + }, + { + "epoch": 0.40274774201755503, + "grad_norm": 3.0506272315979004, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.873581051826477, + "num_tokens": 123145426.0, + "step": 3166 + }, + { + "epoch": 0.4028749522961455, + "grad_norm": 2.937070369720459, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8663021922111511, + "num_tokens": 123177183.0, + "step": 3167 + }, + { + "epoch": 0.40300216257473603, + "grad_norm": 2.8975937366485596, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8675075173377991, + "num_tokens": 123219243.0, + "step": 3168 + }, + { + "epoch": 0.40312937285332656, + "grad_norm": 2.701068878173828, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8711254000663757, + "num_tokens": 123257876.0, + "step": 3169 + }, + { + "epoch": 0.40325658313191703, + "grad_norm": 2.5767998695373535, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.869292140007019, + "num_tokens": 123300237.0, + "step": 3170 + }, + { + "epoch": 0.40338379341050756, + "grad_norm": 2.6140975952148438, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8768059611320496, + "num_tokens": 123341971.0, + "step": 3171 + }, + { + "epoch": 0.4035110036890981, + "grad_norm": 2.7718443870544434, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8717010021209717, + "num_tokens": 123381005.0, + "step": 3172 + }, + { + "epoch": 0.40363821396768856, + "grad_norm": 2.574066638946533, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8829611539840698, + "num_tokens": 123423100.0, + "step": 3173 + }, + { + "epoch": 0.4037654242462791, + "grad_norm": 2.768700122833252, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8731541633605957, + "num_tokens": 123464750.0, + "step": 3174 + }, + { + "epoch": 0.4038926345248696, + "grad_norm": 2.595515489578247, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8861142992973328, + "num_tokens": 123502265.0, + "step": 3175 + }, + { + "epoch": 0.4040198448034601, + "grad_norm": 3.4280598163604736, + "learning_rate": 1e-06, + "loss": 0.4659, + "mean_token_accuracy": 0.844652533531189, + "num_tokens": 123534704.0, + "step": 3176 + }, + { + "epoch": 0.4041470550820506, + "grad_norm": 2.887937307357788, + "learning_rate": 1e-06, + "loss": 0.3975, + "mean_token_accuracy": 0.8635378479957581, + "num_tokens": 123573102.0, + "step": 3177 + }, + { + "epoch": 0.40427426536064115, + "grad_norm": 2.736882448196411, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8691059947013855, + "num_tokens": 123611185.0, + "step": 3178 + }, + { + "epoch": 0.4044014756392316, + "grad_norm": 3.0638246536254883, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8660320043563843, + "num_tokens": 123650780.0, + "step": 3179 + }, + { + "epoch": 0.40452868591782215, + "grad_norm": 2.826523542404175, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8709759712219238, + "num_tokens": 123687491.0, + "step": 3180 + }, + { + "epoch": 0.4046558961964127, + "grad_norm": 2.746857166290283, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8758230209350586, + "num_tokens": 123724375.0, + "step": 3181 + }, + { + "epoch": 0.4047831064750032, + "grad_norm": 2.751741647720337, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8868571519851685, + "num_tokens": 123760189.0, + "step": 3182 + }, + { + "epoch": 0.4049103167535937, + "grad_norm": 3.3121447563171387, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.876705527305603, + "num_tokens": 123786923.0, + "step": 3183 + }, + { + "epoch": 0.4050375270321842, + "grad_norm": 2.6707279682159424, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8825342655181885, + "num_tokens": 123824810.0, + "step": 3184 + }, + { + "epoch": 0.40516473731077474, + "grad_norm": 2.5493621826171875, + "learning_rate": 1e-06, + "loss": 0.4018, + "mean_token_accuracy": 0.8632724285125732, + "num_tokens": 123872426.0, + "step": 3185 + }, + { + "epoch": 0.4052919475893652, + "grad_norm": 2.8543901443481445, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8716140985488892, + "num_tokens": 123911889.0, + "step": 3186 + }, + { + "epoch": 0.40541915786795574, + "grad_norm": 2.5469868183135986, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8727747201919556, + "num_tokens": 123956156.0, + "step": 3187 + }, + { + "epoch": 0.40554636814654627, + "grad_norm": 2.8821492195129395, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8793866038322449, + "num_tokens": 123990645.0, + "step": 3188 + }, + { + "epoch": 0.40567357842513674, + "grad_norm": 2.7227256298065186, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8744332790374756, + "num_tokens": 124027865.0, + "step": 3189 + }, + { + "epoch": 0.40580078870372727, + "grad_norm": 2.8224387168884277, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8704078197479248, + "num_tokens": 124066434.0, + "step": 3190 + }, + { + "epoch": 0.4059279989823178, + "grad_norm": 2.6870553493499756, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8836865425109863, + "num_tokens": 124107160.0, + "step": 3191 + }, + { + "epoch": 0.40605520926090827, + "grad_norm": 2.6015570163726807, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.884579062461853, + "num_tokens": 124145597.0, + "step": 3192 + }, + { + "epoch": 0.4061824195394988, + "grad_norm": 2.6373910903930664, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8761352896690369, + "num_tokens": 124183174.0, + "step": 3193 + }, + { + "epoch": 0.4063096298180893, + "grad_norm": 2.7465362548828125, + "learning_rate": 1e-06, + "loss": 0.3936, + "mean_token_accuracy": 0.8682690858840942, + "num_tokens": 124222587.0, + "step": 3194 + }, + { + "epoch": 0.4064368400966798, + "grad_norm": 2.7185590267181396, + "learning_rate": 1e-06, + "loss": 0.4182, + "mean_token_accuracy": 0.8575520515441895, + "num_tokens": 124265381.0, + "step": 3195 + }, + { + "epoch": 0.4065640503752703, + "grad_norm": 2.6234726905822754, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8864970803260803, + "num_tokens": 124302343.0, + "step": 3196 + }, + { + "epoch": 0.40669126065386085, + "grad_norm": 2.712942600250244, + "learning_rate": 1e-06, + "loss": 0.4108, + "mean_token_accuracy": 0.8544090986251831, + "num_tokens": 124340859.0, + "step": 3197 + }, + { + "epoch": 0.4068184709324513, + "grad_norm": 2.899831533432007, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8718023300170898, + "num_tokens": 124376126.0, + "step": 3198 + }, + { + "epoch": 0.40694568121104185, + "grad_norm": 2.811321973800659, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8756927847862244, + "num_tokens": 124412599.0, + "step": 3199 + }, + { + "epoch": 0.4070728914896324, + "grad_norm": 2.595564603805542, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8769935965538025, + "num_tokens": 124452542.0, + "step": 3200 + }, + { + "epoch": 0.40720010176822286, + "grad_norm": 2.6521308422088623, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8719755411148071, + "num_tokens": 124495717.0, + "step": 3201 + }, + { + "epoch": 0.4073273120468134, + "grad_norm": 2.9245800971984863, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8686336874961853, + "num_tokens": 124534224.0, + "step": 3202 + }, + { + "epoch": 0.4074545223254039, + "grad_norm": 2.9372804164886475, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8706272840499878, + "num_tokens": 124567220.0, + "step": 3203 + }, + { + "epoch": 0.4075817326039944, + "grad_norm": 2.724684476852417, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8838587999343872, + "num_tokens": 124601729.0, + "step": 3204 + }, + { + "epoch": 0.4077089428825849, + "grad_norm": 2.736614227294922, + "learning_rate": 1e-06, + "loss": 0.4056, + "mean_token_accuracy": 0.860434353351593, + "num_tokens": 124640131.0, + "step": 3205 + }, + { + "epoch": 0.40783615316117544, + "grad_norm": 2.8800249099731445, + "learning_rate": 1e-06, + "loss": 0.4022, + "mean_token_accuracy": 0.8632938265800476, + "num_tokens": 124678689.0, + "step": 3206 + }, + { + "epoch": 0.4079633634397659, + "grad_norm": 2.470646381378174, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.8668005466461182, + "num_tokens": 124721574.0, + "step": 3207 + }, + { + "epoch": 0.40809057371835644, + "grad_norm": 2.876359701156616, + "learning_rate": 1e-06, + "loss": 0.3895, + "mean_token_accuracy": 0.8666872978210449, + "num_tokens": 124758706.0, + "step": 3208 + }, + { + "epoch": 0.40821778399694697, + "grad_norm": 2.739088296890259, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8656584024429321, + "num_tokens": 124799688.0, + "step": 3209 + }, + { + "epoch": 0.40834499427553744, + "grad_norm": 2.8194429874420166, + "learning_rate": 1e-06, + "loss": 0.4151, + "mean_token_accuracy": 0.8606790900230408, + "num_tokens": 124840897.0, + "step": 3210 + }, + { + "epoch": 0.40847220455412797, + "grad_norm": 2.9420807361602783, + "learning_rate": 1e-06, + "loss": 0.3919, + "mean_token_accuracy": 0.8657016754150391, + "num_tokens": 124880167.0, + "step": 3211 + }, + { + "epoch": 0.4085994148327185, + "grad_norm": 2.8516552448272705, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8651344776153564, + "num_tokens": 124915157.0, + "step": 3212 + }, + { + "epoch": 0.40872662511130897, + "grad_norm": 2.809983730316162, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8736067414283752, + "num_tokens": 124950337.0, + "step": 3213 + }, + { + "epoch": 0.4088538353898995, + "grad_norm": 2.7106828689575195, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8764077425003052, + "num_tokens": 124987321.0, + "step": 3214 + }, + { + "epoch": 0.40898104566849003, + "grad_norm": 3.101209878921509, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8703045845031738, + "num_tokens": 125019308.0, + "step": 3215 + }, + { + "epoch": 0.4091082559470805, + "grad_norm": 2.5539636611938477, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8726050853729248, + "num_tokens": 125062526.0, + "step": 3216 + }, + { + "epoch": 0.40923546622567103, + "grad_norm": 2.707595109939575, + "learning_rate": 1e-06, + "loss": 0.3917, + "mean_token_accuracy": 0.8617140054702759, + "num_tokens": 125102791.0, + "step": 3217 + }, + { + "epoch": 0.40936267650426156, + "grad_norm": 2.572845697402954, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8931247591972351, + "num_tokens": 125139820.0, + "step": 3218 + }, + { + "epoch": 0.40948988678285203, + "grad_norm": 2.5013887882232666, + "learning_rate": 1e-06, + "loss": 0.4153, + "mean_token_accuracy": 0.856730043888092, + "num_tokens": 125185325.0, + "step": 3219 + }, + { + "epoch": 0.40961709706144256, + "grad_norm": 2.8946597576141357, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8686397075653076, + "num_tokens": 125217964.0, + "step": 3220 + }, + { + "epoch": 0.4097443073400331, + "grad_norm": 2.901108741760254, + "learning_rate": 1e-06, + "loss": 0.4068, + "mean_token_accuracy": 0.8628185987472534, + "num_tokens": 125256529.0, + "step": 3221 + }, + { + "epoch": 0.40987151761862356, + "grad_norm": 2.4866175651550293, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8820973634719849, + "num_tokens": 125301135.0, + "step": 3222 + }, + { + "epoch": 0.4099987278972141, + "grad_norm": 2.4895243644714355, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8722184300422668, + "num_tokens": 125346448.0, + "step": 3223 + }, + { + "epoch": 0.4101259381758046, + "grad_norm": 2.638197422027588, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8835455179214478, + "num_tokens": 125385904.0, + "step": 3224 + }, + { + "epoch": 0.4102531484543951, + "grad_norm": 2.6657583713531494, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8718838691711426, + "num_tokens": 125427111.0, + "step": 3225 + }, + { + "epoch": 0.4103803587329856, + "grad_norm": 2.5245909690856934, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8767573833465576, + "num_tokens": 125472128.0, + "step": 3226 + }, + { + "epoch": 0.41050756901157615, + "grad_norm": 2.664825439453125, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8701837062835693, + "num_tokens": 125513728.0, + "step": 3227 + }, + { + "epoch": 0.4106347792901666, + "grad_norm": 2.729118585586548, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8846211433410645, + "num_tokens": 125544789.0, + "step": 3228 + }, + { + "epoch": 0.41076198956875715, + "grad_norm": 2.641007661819458, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8650404214859009, + "num_tokens": 125585759.0, + "step": 3229 + }, + { + "epoch": 0.4108891998473477, + "grad_norm": 2.809027671813965, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.8684779405593872, + "num_tokens": 125623941.0, + "step": 3230 + }, + { + "epoch": 0.4110164101259382, + "grad_norm": 2.9371132850646973, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8750303983688354, + "num_tokens": 125658971.0, + "step": 3231 + }, + { + "epoch": 0.4111436204045287, + "grad_norm": 3.0742480754852295, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8688556551933289, + "num_tokens": 125690148.0, + "step": 3232 + }, + { + "epoch": 0.4112708306831192, + "grad_norm": 2.639024257659912, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8781565427780151, + "num_tokens": 125728070.0, + "step": 3233 + }, + { + "epoch": 0.41139804096170973, + "grad_norm": 2.658639907836914, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.874589204788208, + "num_tokens": 125766749.0, + "step": 3234 + }, + { + "epoch": 0.4115252512403002, + "grad_norm": 2.555262327194214, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8773097395896912, + "num_tokens": 125804718.0, + "step": 3235 + }, + { + "epoch": 0.41165246151889073, + "grad_norm": 2.70632004737854, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.880871593952179, + "num_tokens": 125838877.0, + "step": 3236 + }, + { + "epoch": 0.41177967179748126, + "grad_norm": 2.551893949508667, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8850679397583008, + "num_tokens": 125880411.0, + "step": 3237 + }, + { + "epoch": 0.41190688207607173, + "grad_norm": 2.7660152912139893, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8745362758636475, + "num_tokens": 125915303.0, + "step": 3238 + }, + { + "epoch": 0.41203409235466226, + "grad_norm": 2.618715763092041, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8798264265060425, + "num_tokens": 125951053.0, + "step": 3239 + }, + { + "epoch": 0.4121613026332528, + "grad_norm": 2.718303918838501, + "learning_rate": 1e-06, + "loss": 0.4024, + "mean_token_accuracy": 0.8587208986282349, + "num_tokens": 125991985.0, + "step": 3240 + }, + { + "epoch": 0.41228851291184326, + "grad_norm": 2.788896083831787, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.867952823638916, + "num_tokens": 126028098.0, + "step": 3241 + }, + { + "epoch": 0.4124157231904338, + "grad_norm": 2.429328441619873, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8934733867645264, + "num_tokens": 126068747.0, + "step": 3242 + }, + { + "epoch": 0.4125429334690243, + "grad_norm": 2.655414581298828, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.8725910186767578, + "num_tokens": 126108119.0, + "step": 3243 + }, + { + "epoch": 0.4126701437476148, + "grad_norm": 2.6106984615325928, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8823365569114685, + "num_tokens": 126145652.0, + "step": 3244 + }, + { + "epoch": 0.4127973540262053, + "grad_norm": 2.8953583240509033, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8790369033813477, + "num_tokens": 126178749.0, + "step": 3245 + }, + { + "epoch": 0.41292456430479585, + "grad_norm": 2.6957833766937256, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8751952052116394, + "num_tokens": 126217622.0, + "step": 3246 + }, + { + "epoch": 0.4130517745833863, + "grad_norm": 2.9779932498931885, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.866144061088562, + "num_tokens": 126251394.0, + "step": 3247 + }, + { + "epoch": 0.41317898486197685, + "grad_norm": 2.6923635005950928, + "learning_rate": 1e-06, + "loss": 0.3909, + "mean_token_accuracy": 0.8677729368209839, + "num_tokens": 126293763.0, + "step": 3248 + }, + { + "epoch": 0.4133061951405674, + "grad_norm": 2.7568674087524414, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.873521625995636, + "num_tokens": 126332472.0, + "step": 3249 + }, + { + "epoch": 0.41343340541915785, + "grad_norm": 2.9066264629364014, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8702201247215271, + "num_tokens": 126367313.0, + "step": 3250 + }, + { + "epoch": 0.4135606156977484, + "grad_norm": 2.930532217025757, + "learning_rate": 1e-06, + "loss": 0.3994, + "mean_token_accuracy": 0.8647385239601135, + "num_tokens": 126403597.0, + "step": 3251 + }, + { + "epoch": 0.4136878259763389, + "grad_norm": 3.0304417610168457, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8663503527641296, + "num_tokens": 126440450.0, + "step": 3252 + }, + { + "epoch": 0.4138150362549294, + "grad_norm": 2.749797821044922, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8804731369018555, + "num_tokens": 126478651.0, + "step": 3253 + }, + { + "epoch": 0.4139422465335199, + "grad_norm": 2.7408831119537354, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8762378096580505, + "num_tokens": 126517837.0, + "step": 3254 + }, + { + "epoch": 0.41406945681211044, + "grad_norm": 2.6587085723876953, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8754021525382996, + "num_tokens": 126555501.0, + "step": 3255 + }, + { + "epoch": 0.4141966670907009, + "grad_norm": 2.578068733215332, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.863975465297699, + "num_tokens": 126599170.0, + "step": 3256 + }, + { + "epoch": 0.41432387736929144, + "grad_norm": 2.4828271865844727, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8712911605834961, + "num_tokens": 126643194.0, + "step": 3257 + }, + { + "epoch": 0.41445108764788197, + "grad_norm": 2.764127492904663, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8814221620559692, + "num_tokens": 126680968.0, + "step": 3258 + }, + { + "epoch": 0.41457829792647244, + "grad_norm": 2.9602513313293457, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8616992235183716, + "num_tokens": 126715386.0, + "step": 3259 + }, + { + "epoch": 0.41470550820506297, + "grad_norm": 3.04068922996521, + "learning_rate": 1e-06, + "loss": 0.4067, + "mean_token_accuracy": 0.8602217435836792, + "num_tokens": 126748113.0, + "step": 3260 + }, + { + "epoch": 0.4148327184836535, + "grad_norm": 2.644974708557129, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8783636689186096, + "num_tokens": 126784497.0, + "step": 3261 + }, + { + "epoch": 0.41495992876224397, + "grad_norm": 2.5235657691955566, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8796965479850769, + "num_tokens": 126829345.0, + "step": 3262 + }, + { + "epoch": 0.4150871390408345, + "grad_norm": 3.062335968017578, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8691256046295166, + "num_tokens": 126862280.0, + "step": 3263 + }, + { + "epoch": 0.415214349319425, + "grad_norm": 2.6706480979919434, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8716421127319336, + "num_tokens": 126902655.0, + "step": 3264 + }, + { + "epoch": 0.4153415595980155, + "grad_norm": 2.4702436923980713, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8831058740615845, + "num_tokens": 126945546.0, + "step": 3265 + }, + { + "epoch": 0.415468769876606, + "grad_norm": 2.801518201828003, + "learning_rate": 1e-06, + "loss": 0.4062, + "mean_token_accuracy": 0.8604803085327148, + "num_tokens": 126985687.0, + "step": 3266 + }, + { + "epoch": 0.41559598015519655, + "grad_norm": 2.58603835105896, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8783467411994934, + "num_tokens": 127025659.0, + "step": 3267 + }, + { + "epoch": 0.415723190433787, + "grad_norm": 2.7498950958251953, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8921017646789551, + "num_tokens": 127059329.0, + "step": 3268 + }, + { + "epoch": 0.41585040071237755, + "grad_norm": 2.636669635772705, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8870385885238647, + "num_tokens": 127097304.0, + "step": 3269 + }, + { + "epoch": 0.4159776109909681, + "grad_norm": 2.6811375617980957, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8747137784957886, + "num_tokens": 127135034.0, + "step": 3270 + }, + { + "epoch": 0.41610482126955856, + "grad_norm": 2.421818971633911, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8713998198509216, + "num_tokens": 127177102.0, + "step": 3271 + }, + { + "epoch": 0.4162320315481491, + "grad_norm": 2.69665789604187, + "learning_rate": 1e-06, + "loss": 0.3952, + "mean_token_accuracy": 0.8629817366600037, + "num_tokens": 127215492.0, + "step": 3272 + }, + { + "epoch": 0.4163592418267396, + "grad_norm": 2.7808775901794434, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8715939521789551, + "num_tokens": 127250117.0, + "step": 3273 + }, + { + "epoch": 0.4164864521053301, + "grad_norm": 2.6300666332244873, + "learning_rate": 1e-06, + "loss": 0.4243, + "mean_token_accuracy": 0.8525017499923706, + "num_tokens": 127295304.0, + "step": 3274 + }, + { + "epoch": 0.4166136623839206, + "grad_norm": 2.562803030014038, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8669553995132446, + "num_tokens": 127337837.0, + "step": 3275 + }, + { + "epoch": 0.41674087266251114, + "grad_norm": 2.6998515129089355, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8832964897155762, + "num_tokens": 127372289.0, + "step": 3276 + }, + { + "epoch": 0.4168680829411016, + "grad_norm": 2.552391767501831, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8922746181488037, + "num_tokens": 127409455.0, + "step": 3277 + }, + { + "epoch": 0.41699529321969214, + "grad_norm": 2.4333906173706055, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8889260292053223, + "num_tokens": 127451760.0, + "step": 3278 + }, + { + "epoch": 0.41712250349828267, + "grad_norm": 2.829249382019043, + "learning_rate": 1e-06, + "loss": 0.3869, + "mean_token_accuracy": 0.8666309714317322, + "num_tokens": 127491255.0, + "step": 3279 + }, + { + "epoch": 0.4172497137768732, + "grad_norm": 2.6916604042053223, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8785861730575562, + "num_tokens": 127532412.0, + "step": 3280 + }, + { + "epoch": 0.41737692405546367, + "grad_norm": 2.537381410598755, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8720566034317017, + "num_tokens": 127577478.0, + "step": 3281 + }, + { + "epoch": 0.4175041343340542, + "grad_norm": 2.7457635402679443, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8717020750045776, + "num_tokens": 127616489.0, + "step": 3282 + }, + { + "epoch": 0.41763134461264473, + "grad_norm": 2.633545160293579, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8844969272613525, + "num_tokens": 127656202.0, + "step": 3283 + }, + { + "epoch": 0.4177585548912352, + "grad_norm": 2.5625174045562744, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8749012351036072, + "num_tokens": 127695436.0, + "step": 3284 + }, + { + "epoch": 0.41788576516982573, + "grad_norm": 2.7210137844085693, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8730059862136841, + "num_tokens": 127733586.0, + "step": 3285 + }, + { + "epoch": 0.41801297544841626, + "grad_norm": 2.999295711517334, + "learning_rate": 1e-06, + "loss": 0.3952, + "mean_token_accuracy": 0.8615204095840454, + "num_tokens": 127766548.0, + "step": 3286 + }, + { + "epoch": 0.41814018572700673, + "grad_norm": 2.830723762512207, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8698906302452087, + "num_tokens": 127801915.0, + "step": 3287 + }, + { + "epoch": 0.41826739600559726, + "grad_norm": 2.4837071895599365, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.875335156917572, + "num_tokens": 127843227.0, + "step": 3288 + }, + { + "epoch": 0.4183946062841878, + "grad_norm": 2.6201040744781494, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8770818710327148, + "num_tokens": 127882221.0, + "step": 3289 + }, + { + "epoch": 0.41852181656277826, + "grad_norm": 2.812241554260254, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8667516708374023, + "num_tokens": 127921522.0, + "step": 3290 + }, + { + "epoch": 0.4186490268413688, + "grad_norm": 2.647298574447632, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8700289726257324, + "num_tokens": 127962383.0, + "step": 3291 + }, + { + "epoch": 0.4187762371199593, + "grad_norm": 2.7672953605651855, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8722394108772278, + "num_tokens": 127998379.0, + "step": 3292 + }, + { + "epoch": 0.4189034473985498, + "grad_norm": 2.5156161785125732, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8695194721221924, + "num_tokens": 128042741.0, + "step": 3293 + }, + { + "epoch": 0.4190306576771403, + "grad_norm": 2.7413763999938965, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8775417804718018, + "num_tokens": 128080020.0, + "step": 3294 + }, + { + "epoch": 0.41915786795573085, + "grad_norm": 2.8806002140045166, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.875518262386322, + "num_tokens": 128113627.0, + "step": 3295 + }, + { + "epoch": 0.4192850782343213, + "grad_norm": 2.6974496841430664, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8706536293029785, + "num_tokens": 128153523.0, + "step": 3296 + }, + { + "epoch": 0.41941228851291185, + "grad_norm": 2.7097971439361572, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8752536773681641, + "num_tokens": 128188519.0, + "step": 3297 + }, + { + "epoch": 0.4195394987915024, + "grad_norm": 2.818629264831543, + "learning_rate": 1e-06, + "loss": 0.3981, + "mean_token_accuracy": 0.8669871091842651, + "num_tokens": 128226767.0, + "step": 3298 + }, + { + "epoch": 0.41966670907009285, + "grad_norm": 2.6631650924682617, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8719648122787476, + "num_tokens": 128263716.0, + "step": 3299 + }, + { + "epoch": 0.4197939193486834, + "grad_norm": 2.674220323562622, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8789860010147095, + "num_tokens": 128301782.0, + "step": 3300 + }, + { + "epoch": 0.4199211296272739, + "grad_norm": 2.8026719093322754, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8719978332519531, + "num_tokens": 128341086.0, + "step": 3301 + }, + { + "epoch": 0.4200483399058644, + "grad_norm": 2.7754967212677, + "learning_rate": 1e-06, + "loss": 0.4042, + "mean_token_accuracy": 0.8644570708274841, + "num_tokens": 128379834.0, + "step": 3302 + }, + { + "epoch": 0.4201755501844549, + "grad_norm": 2.750913143157959, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8638553023338318, + "num_tokens": 128421849.0, + "step": 3303 + }, + { + "epoch": 0.42030276046304543, + "grad_norm": 2.5871033668518066, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8762202858924866, + "num_tokens": 128462378.0, + "step": 3304 + }, + { + "epoch": 0.4204299707416359, + "grad_norm": 2.6953752040863037, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.8645579814910889, + "num_tokens": 128501913.0, + "step": 3305 + }, + { + "epoch": 0.42055718102022643, + "grad_norm": 2.709073543548584, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8692691326141357, + "num_tokens": 128543664.0, + "step": 3306 + }, + { + "epoch": 0.42068439129881696, + "grad_norm": 2.5295872688293457, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8679702281951904, + "num_tokens": 128586436.0, + "step": 3307 + }, + { + "epoch": 0.42081160157740743, + "grad_norm": 2.8339853286743164, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8881092071533203, + "num_tokens": 128620170.0, + "step": 3308 + }, + { + "epoch": 0.42093881185599796, + "grad_norm": 2.856955051422119, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.8708527684211731, + "num_tokens": 128656065.0, + "step": 3309 + }, + { + "epoch": 0.4210660221345885, + "grad_norm": 2.794511079788208, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.8666355013847351, + "num_tokens": 128693859.0, + "step": 3310 + }, + { + "epoch": 0.42119323241317896, + "grad_norm": 2.5230610370635986, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8811123371124268, + "num_tokens": 128736937.0, + "step": 3311 + }, + { + "epoch": 0.4213204426917695, + "grad_norm": 2.3833844661712646, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8772004842758179, + "num_tokens": 128781781.0, + "step": 3312 + }, + { + "epoch": 0.42144765297036, + "grad_norm": 2.5827322006225586, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8796195983886719, + "num_tokens": 128821372.0, + "step": 3313 + }, + { + "epoch": 0.4215748632489505, + "grad_norm": 2.6315808296203613, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8742578029632568, + "num_tokens": 128859156.0, + "step": 3314 + }, + { + "epoch": 0.421702073527541, + "grad_norm": 2.582277297973633, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8774881362915039, + "num_tokens": 128900867.0, + "step": 3315 + }, + { + "epoch": 0.42182928380613155, + "grad_norm": 2.68611741065979, + "learning_rate": 1e-06, + "loss": 0.3962, + "mean_token_accuracy": 0.8607361316680908, + "num_tokens": 128939963.0, + "step": 3316 + }, + { + "epoch": 0.421956494084722, + "grad_norm": 2.581315040588379, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8826927542686462, + "num_tokens": 128976029.0, + "step": 3317 + }, + { + "epoch": 0.42208370436331255, + "grad_norm": 2.615877389907837, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.8614205121994019, + "num_tokens": 129017442.0, + "step": 3318 + }, + { + "epoch": 0.4222109146419031, + "grad_norm": 2.640390634536743, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8667123317718506, + "num_tokens": 129059137.0, + "step": 3319 + }, + { + "epoch": 0.42233812492049355, + "grad_norm": 3.0627176761627197, + "learning_rate": 1e-06, + "loss": 0.391, + "mean_token_accuracy": 0.865654706954956, + "num_tokens": 129092620.0, + "step": 3320 + }, + { + "epoch": 0.4224653351990841, + "grad_norm": 2.870436906814575, + "learning_rate": 1e-06, + "loss": 0.4084, + "mean_token_accuracy": 0.8630543947219849, + "num_tokens": 129132313.0, + "step": 3321 + }, + { + "epoch": 0.4225925454776746, + "grad_norm": 2.9823343753814697, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.8675341606140137, + "num_tokens": 129171258.0, + "step": 3322 + }, + { + "epoch": 0.4227197557562651, + "grad_norm": 2.630566120147705, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.8608913421630859, + "num_tokens": 129215009.0, + "step": 3323 + }, + { + "epoch": 0.4228469660348556, + "grad_norm": 2.989760160446167, + "learning_rate": 1e-06, + "loss": 0.4315, + "mean_token_accuracy": 0.853959321975708, + "num_tokens": 129250397.0, + "step": 3324 + }, + { + "epoch": 0.42297417631344614, + "grad_norm": 2.7107925415039062, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.88204425573349, + "num_tokens": 129287892.0, + "step": 3325 + }, + { + "epoch": 0.4231013865920366, + "grad_norm": 2.586120367050171, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8668781518936157, + "num_tokens": 129328062.0, + "step": 3326 + }, + { + "epoch": 0.42322859687062714, + "grad_norm": 2.7119643688201904, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8844285607337952, + "num_tokens": 129363486.0, + "step": 3327 + }, + { + "epoch": 0.42335580714921767, + "grad_norm": 2.87429141998291, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.877380907535553, + "num_tokens": 129396211.0, + "step": 3328 + }, + { + "epoch": 0.42348301742780814, + "grad_norm": 2.6556825637817383, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8810359239578247, + "num_tokens": 129433702.0, + "step": 3329 + }, + { + "epoch": 0.42361022770639867, + "grad_norm": 2.30553936958313, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8774341344833374, + "num_tokens": 129479216.0, + "step": 3330 + }, + { + "epoch": 0.4237374379849892, + "grad_norm": 2.6869566440582275, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8863633871078491, + "num_tokens": 129512863.0, + "step": 3331 + }, + { + "epoch": 0.4238646482635797, + "grad_norm": 2.675508737564087, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8706812858581543, + "num_tokens": 129554334.0, + "step": 3332 + }, + { + "epoch": 0.4239918585421702, + "grad_norm": 2.8597846031188965, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8668053150177002, + "num_tokens": 129589380.0, + "step": 3333 + }, + { + "epoch": 0.4241190688207607, + "grad_norm": 2.921069622039795, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8797792792320251, + "num_tokens": 129623143.0, + "step": 3334 + }, + { + "epoch": 0.42424627909935125, + "grad_norm": 2.5299782752990723, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8826929926872253, + "num_tokens": 129659202.0, + "step": 3335 + }, + { + "epoch": 0.4243734893779417, + "grad_norm": 2.603156328201294, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8775915503501892, + "num_tokens": 129696880.0, + "step": 3336 + }, + { + "epoch": 0.42450069965653225, + "grad_norm": 2.5188045501708984, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8771270513534546, + "num_tokens": 129737512.0, + "step": 3337 + }, + { + "epoch": 0.4246279099351228, + "grad_norm": 2.8871819972991943, + "learning_rate": 1e-06, + "loss": 0.4047, + "mean_token_accuracy": 0.8613083958625793, + "num_tokens": 129772557.0, + "step": 3338 + }, + { + "epoch": 0.42475512021371326, + "grad_norm": 2.5877761840820312, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8781378269195557, + "num_tokens": 129817643.0, + "step": 3339 + }, + { + "epoch": 0.4248823304923038, + "grad_norm": 2.5074005126953125, + "learning_rate": 1e-06, + "loss": 0.3786, + "mean_token_accuracy": 0.8665362596511841, + "num_tokens": 129862639.0, + "step": 3340 + }, + { + "epoch": 0.4250095407708943, + "grad_norm": 2.7403619289398193, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8775068521499634, + "num_tokens": 129899388.0, + "step": 3341 + }, + { + "epoch": 0.4251367510494848, + "grad_norm": 2.5286595821380615, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8780189752578735, + "num_tokens": 129942761.0, + "step": 3342 + }, + { + "epoch": 0.4252639613280753, + "grad_norm": 2.528582811355591, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8814194202423096, + "num_tokens": 129984500.0, + "step": 3343 + }, + { + "epoch": 0.42539117160666584, + "grad_norm": 2.6492369174957275, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.8698591589927673, + "num_tokens": 130028268.0, + "step": 3344 + }, + { + "epoch": 0.4255183818852563, + "grad_norm": 2.6485886573791504, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8813338875770569, + "num_tokens": 130066094.0, + "step": 3345 + }, + { + "epoch": 0.42564559216384684, + "grad_norm": 2.8177201747894287, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8834719657897949, + "num_tokens": 130098308.0, + "step": 3346 + }, + { + "epoch": 0.42577280244243737, + "grad_norm": 2.7667059898376465, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8834869861602783, + "num_tokens": 130130672.0, + "step": 3347 + }, + { + "epoch": 0.42590001272102784, + "grad_norm": 2.667515277862549, + "learning_rate": 1e-06, + "loss": 0.3854, + "mean_token_accuracy": 0.8650389909744263, + "num_tokens": 130168377.0, + "step": 3348 + }, + { + "epoch": 0.42602722299961837, + "grad_norm": 2.8335049152374268, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8686906099319458, + "num_tokens": 130205419.0, + "step": 3349 + }, + { + "epoch": 0.4261544332782089, + "grad_norm": 2.5844950675964355, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8890820741653442, + "num_tokens": 130246324.0, + "step": 3350 + }, + { + "epoch": 0.4262816435567994, + "grad_norm": 2.664407730102539, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8715077638626099, + "num_tokens": 130283024.0, + "step": 3351 + }, + { + "epoch": 0.4264088538353899, + "grad_norm": 2.5172078609466553, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8820530772209167, + "num_tokens": 130325191.0, + "step": 3352 + }, + { + "epoch": 0.42653606411398043, + "grad_norm": 2.778672933578491, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8756338953971863, + "num_tokens": 130363319.0, + "step": 3353 + }, + { + "epoch": 0.4266632743925709, + "grad_norm": 2.5024936199188232, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8822043538093567, + "num_tokens": 130402633.0, + "step": 3354 + }, + { + "epoch": 0.42679048467116143, + "grad_norm": 2.7995822429656982, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.8642659187316895, + "num_tokens": 130440345.0, + "step": 3355 + }, + { + "epoch": 0.42691769494975196, + "grad_norm": 2.934713840484619, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8722184896469116, + "num_tokens": 130475503.0, + "step": 3356 + }, + { + "epoch": 0.42704490522834243, + "grad_norm": 2.829974412918091, + "learning_rate": 1e-06, + "loss": 0.3937, + "mean_token_accuracy": 0.864069402217865, + "num_tokens": 130513430.0, + "step": 3357 + }, + { + "epoch": 0.42717211550693296, + "grad_norm": 2.618189573287964, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8809095025062561, + "num_tokens": 130551510.0, + "step": 3358 + }, + { + "epoch": 0.4272993257855235, + "grad_norm": 2.6435816287994385, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8754321336746216, + "num_tokens": 130590684.0, + "step": 3359 + }, + { + "epoch": 0.42742653606411396, + "grad_norm": 2.5061793327331543, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8861469626426697, + "num_tokens": 130627853.0, + "step": 3360 + }, + { + "epoch": 0.4275537463427045, + "grad_norm": 2.825617790222168, + "learning_rate": 1e-06, + "loss": 0.4034, + "mean_token_accuracy": 0.8623676300048828, + "num_tokens": 130666653.0, + "step": 3361 + }, + { + "epoch": 0.427680956621295, + "grad_norm": 2.9859001636505127, + "learning_rate": 1e-06, + "loss": 0.4121, + "mean_token_accuracy": 0.8605768084526062, + "num_tokens": 130699792.0, + "step": 3362 + }, + { + "epoch": 0.4278081668998855, + "grad_norm": 2.528041362762451, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8837252855300903, + "num_tokens": 130741864.0, + "step": 3363 + }, + { + "epoch": 0.427935377178476, + "grad_norm": 2.742889881134033, + "learning_rate": 1e-06, + "loss": 0.4027, + "mean_token_accuracy": 0.8624942898750305, + "num_tokens": 130781485.0, + "step": 3364 + }, + { + "epoch": 0.42806258745706655, + "grad_norm": 3.0237958431243896, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8851399421691895, + "num_tokens": 130810250.0, + "step": 3365 + }, + { + "epoch": 0.428189797735657, + "grad_norm": 2.702399730682373, + "learning_rate": 1e-06, + "loss": 0.3775, + "mean_token_accuracy": 0.8667422533035278, + "num_tokens": 130849483.0, + "step": 3366 + }, + { + "epoch": 0.42831700801424755, + "grad_norm": 2.947240114212036, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8767552375793457, + "num_tokens": 130885606.0, + "step": 3367 + }, + { + "epoch": 0.4284442182928381, + "grad_norm": 2.733299732208252, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.877714216709137, + "num_tokens": 130921314.0, + "step": 3368 + }, + { + "epoch": 0.42857142857142855, + "grad_norm": 2.4251322746276855, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.881793737411499, + "num_tokens": 130967028.0, + "step": 3369 + }, + { + "epoch": 0.4286986388500191, + "grad_norm": 2.7606701850891113, + "learning_rate": 1e-06, + "loss": 0.4069, + "mean_token_accuracy": 0.8632059097290039, + "num_tokens": 131005357.0, + "step": 3370 + }, + { + "epoch": 0.4288258491286096, + "grad_norm": 2.6565322875976562, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8773809671401978, + "num_tokens": 131044495.0, + "step": 3371 + }, + { + "epoch": 0.4289530594072001, + "grad_norm": 2.5591273307800293, + "learning_rate": 1e-06, + "loss": 0.4425, + "mean_token_accuracy": 0.8511782884597778, + "num_tokens": 131089102.0, + "step": 3372 + }, + { + "epoch": 0.4290802696857906, + "grad_norm": 2.882336378097534, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8775992393493652, + "num_tokens": 131126554.0, + "step": 3373 + }, + { + "epoch": 0.42920747996438113, + "grad_norm": 2.539322853088379, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8781031370162964, + "num_tokens": 131171488.0, + "step": 3374 + }, + { + "epoch": 0.4293346902429716, + "grad_norm": 2.9326698780059814, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8768198490142822, + "num_tokens": 131208623.0, + "step": 3375 + }, + { + "epoch": 0.42946190052156213, + "grad_norm": 2.816946029663086, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.875152587890625, + "num_tokens": 131244830.0, + "step": 3376 + }, + { + "epoch": 0.42958911080015266, + "grad_norm": 2.8527352809906006, + "learning_rate": 1e-06, + "loss": 0.4001, + "mean_token_accuracy": 0.8634374141693115, + "num_tokens": 131282363.0, + "step": 3377 + }, + { + "epoch": 0.42971632107874314, + "grad_norm": 2.6198906898498535, + "learning_rate": 1e-06, + "loss": 0.4, + "mean_token_accuracy": 0.8615509271621704, + "num_tokens": 131322869.0, + "step": 3378 + }, + { + "epoch": 0.42984353135733366, + "grad_norm": 2.666966199874878, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8827397227287292, + "num_tokens": 131357838.0, + "step": 3379 + }, + { + "epoch": 0.4299707416359242, + "grad_norm": 2.5987789630889893, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8779647350311279, + "num_tokens": 131400480.0, + "step": 3380 + }, + { + "epoch": 0.4300979519145147, + "grad_norm": 2.7175133228302, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8821457624435425, + "num_tokens": 131438895.0, + "step": 3381 + }, + { + "epoch": 0.4302251621931052, + "grad_norm": 2.766007423400879, + "learning_rate": 1e-06, + "loss": 0.4232, + "mean_token_accuracy": 0.8567802906036377, + "num_tokens": 131480137.0, + "step": 3382 + }, + { + "epoch": 0.4303523724716957, + "grad_norm": 2.5962839126586914, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8793277144432068, + "num_tokens": 131518927.0, + "step": 3383 + }, + { + "epoch": 0.43047958275028625, + "grad_norm": 2.6817049980163574, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8682474493980408, + "num_tokens": 131558523.0, + "step": 3384 + }, + { + "epoch": 0.4306067930288767, + "grad_norm": 2.724180221557617, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8822476267814636, + "num_tokens": 131596138.0, + "step": 3385 + }, + { + "epoch": 0.43073400330746725, + "grad_norm": 2.8225224018096924, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8809280395507812, + "num_tokens": 131628172.0, + "step": 3386 + }, + { + "epoch": 0.4308612135860578, + "grad_norm": 2.7996363639831543, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8755798935890198, + "num_tokens": 131664726.0, + "step": 3387 + }, + { + "epoch": 0.43098842386464825, + "grad_norm": 2.5187010765075684, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8725970983505249, + "num_tokens": 131707649.0, + "step": 3388 + }, + { + "epoch": 0.4311156341432388, + "grad_norm": 2.8456640243530273, + "learning_rate": 1e-06, + "loss": 0.4065, + "mean_token_accuracy": 0.8635761737823486, + "num_tokens": 131750277.0, + "step": 3389 + }, + { + "epoch": 0.4312428444218293, + "grad_norm": 2.5618698596954346, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8776836395263672, + "num_tokens": 131790443.0, + "step": 3390 + }, + { + "epoch": 0.4313700547004198, + "grad_norm": 2.7296361923217773, + "learning_rate": 1e-06, + "loss": 0.4067, + "mean_token_accuracy": 0.857631504535675, + "num_tokens": 131829294.0, + "step": 3391 + }, + { + "epoch": 0.4314972649790103, + "grad_norm": 2.5295379161834717, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.877091109752655, + "num_tokens": 131871944.0, + "step": 3392 + }, + { + "epoch": 0.43162447525760084, + "grad_norm": 2.6754188537597656, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8747619986534119, + "num_tokens": 131911255.0, + "step": 3393 + }, + { + "epoch": 0.4317516855361913, + "grad_norm": 2.6031622886657715, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8731849193572998, + "num_tokens": 131955197.0, + "step": 3394 + }, + { + "epoch": 0.43187889581478184, + "grad_norm": 2.7041022777557373, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8883084058761597, + "num_tokens": 131988876.0, + "step": 3395 + }, + { + "epoch": 0.43200610609337237, + "grad_norm": 2.801762580871582, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8685497045516968, + "num_tokens": 132025990.0, + "step": 3396 + }, + { + "epoch": 0.43213331637196284, + "grad_norm": 2.512401819229126, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8793853521347046, + "num_tokens": 132069437.0, + "step": 3397 + }, + { + "epoch": 0.43226052665055337, + "grad_norm": 2.7030935287475586, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8712877631187439, + "num_tokens": 132107633.0, + "step": 3398 + }, + { + "epoch": 0.4323877369291439, + "grad_norm": 2.669243097305298, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8693275451660156, + "num_tokens": 132148733.0, + "step": 3399 + }, + { + "epoch": 0.43251494720773437, + "grad_norm": 2.752814292907715, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8770469427108765, + "num_tokens": 132188377.0, + "step": 3400 + }, + { + "epoch": 0.4326421574863249, + "grad_norm": 2.827099323272705, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8720013499259949, + "num_tokens": 132225958.0, + "step": 3401 + }, + { + "epoch": 0.4327693677649154, + "grad_norm": 2.5442745685577393, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8796674013137817, + "num_tokens": 132271835.0, + "step": 3402 + }, + { + "epoch": 0.4328965780435059, + "grad_norm": 2.883232831954956, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8676553964614868, + "num_tokens": 132307452.0, + "step": 3403 + }, + { + "epoch": 0.4330237883220964, + "grad_norm": 2.698350191116333, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8749091625213623, + "num_tokens": 132347478.0, + "step": 3404 + }, + { + "epoch": 0.43315099860068695, + "grad_norm": 2.5379035472869873, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8787828683853149, + "num_tokens": 132388910.0, + "step": 3405 + }, + { + "epoch": 0.4332782088792774, + "grad_norm": 2.9006640911102295, + "learning_rate": 1e-06, + "loss": 0.3935, + "mean_token_accuracy": 0.864452600479126, + "num_tokens": 132423610.0, + "step": 3406 + }, + { + "epoch": 0.43340541915786795, + "grad_norm": 2.616405487060547, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8857451677322388, + "num_tokens": 132463959.0, + "step": 3407 + }, + { + "epoch": 0.4335326294364585, + "grad_norm": 2.7556240558624268, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8893426060676575, + "num_tokens": 132500475.0, + "step": 3408 + }, + { + "epoch": 0.43365983971504896, + "grad_norm": 2.6862587928771973, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8732198476791382, + "num_tokens": 132540968.0, + "step": 3409 + }, + { + "epoch": 0.4337870499936395, + "grad_norm": 2.520707130432129, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8715640306472778, + "num_tokens": 132585782.0, + "step": 3410 + }, + { + "epoch": 0.43391426027223, + "grad_norm": 2.9190802574157715, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8644849061965942, + "num_tokens": 132625321.0, + "step": 3411 + }, + { + "epoch": 0.4340414705508205, + "grad_norm": 2.8009040355682373, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.869363009929657, + "num_tokens": 132662163.0, + "step": 3412 + }, + { + "epoch": 0.434168680829411, + "grad_norm": 2.5097086429595947, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8740061521530151, + "num_tokens": 132706335.0, + "step": 3413 + }, + { + "epoch": 0.43429589110800154, + "grad_norm": 2.7744555473327637, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8741733431816101, + "num_tokens": 132744163.0, + "step": 3414 + }, + { + "epoch": 0.434423101386592, + "grad_norm": 2.4704980850219727, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8694376349449158, + "num_tokens": 132791383.0, + "step": 3415 + }, + { + "epoch": 0.43455031166518254, + "grad_norm": 2.492907762527466, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8821977376937866, + "num_tokens": 132834924.0, + "step": 3416 + }, + { + "epoch": 0.43467752194377307, + "grad_norm": 2.5353314876556396, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8761345148086548, + "num_tokens": 132878221.0, + "step": 3417 + }, + { + "epoch": 0.43480473222236354, + "grad_norm": 2.639052152633667, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8746538162231445, + "num_tokens": 132917141.0, + "step": 3418 + }, + { + "epoch": 0.43493194250095407, + "grad_norm": 2.6995606422424316, + "learning_rate": 1e-06, + "loss": 0.4093, + "mean_token_accuracy": 0.859247088432312, + "num_tokens": 132960837.0, + "step": 3419 + }, + { + "epoch": 0.4350591527795446, + "grad_norm": 2.4603779315948486, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8862585425376892, + "num_tokens": 133003982.0, + "step": 3420 + }, + { + "epoch": 0.4351863630581351, + "grad_norm": 2.855888843536377, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8713291883468628, + "num_tokens": 133039930.0, + "step": 3421 + }, + { + "epoch": 0.4353135733367256, + "grad_norm": 2.742415428161621, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8744296431541443, + "num_tokens": 133075500.0, + "step": 3422 + }, + { + "epoch": 0.43544078361531613, + "grad_norm": 2.962596893310547, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.875861644744873, + "num_tokens": 133104416.0, + "step": 3423 + }, + { + "epoch": 0.4355679938939066, + "grad_norm": 2.5701024532318115, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8731892704963684, + "num_tokens": 133145256.0, + "step": 3424 + }, + { + "epoch": 0.43569520417249713, + "grad_norm": 2.8170487880706787, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8824270963668823, + "num_tokens": 133183984.0, + "step": 3425 + }, + { + "epoch": 0.43582241445108766, + "grad_norm": 2.920010805130005, + "learning_rate": 1e-06, + "loss": 0.4137, + "mean_token_accuracy": 0.8584570288658142, + "num_tokens": 133223215.0, + "step": 3426 + }, + { + "epoch": 0.43594962472967813, + "grad_norm": 2.6982085704803467, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8663358092308044, + "num_tokens": 133263659.0, + "step": 3427 + }, + { + "epoch": 0.43607683500826866, + "grad_norm": 2.607388734817505, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8726745843887329, + "num_tokens": 133306649.0, + "step": 3428 + }, + { + "epoch": 0.4362040452868592, + "grad_norm": 2.8673641681671143, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8721288442611694, + "num_tokens": 133343436.0, + "step": 3429 + }, + { + "epoch": 0.4363312555654497, + "grad_norm": 2.693394422531128, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8766053915023804, + "num_tokens": 133379952.0, + "step": 3430 + }, + { + "epoch": 0.4364584658440402, + "grad_norm": 2.7076799869537354, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8692848682403564, + "num_tokens": 133419745.0, + "step": 3431 + }, + { + "epoch": 0.4365856761226307, + "grad_norm": 2.8362226486206055, + "learning_rate": 1e-06, + "loss": 0.4128, + "mean_token_accuracy": 0.8604522943496704, + "num_tokens": 133455579.0, + "step": 3432 + }, + { + "epoch": 0.43671288640122125, + "grad_norm": 2.5687379837036133, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8671363592147827, + "num_tokens": 133498178.0, + "step": 3433 + }, + { + "epoch": 0.4368400966798117, + "grad_norm": 3.015814781188965, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8767637610435486, + "num_tokens": 133527932.0, + "step": 3434 + }, + { + "epoch": 0.43696730695840225, + "grad_norm": 2.6608104705810547, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8723440170288086, + "num_tokens": 133567631.0, + "step": 3435 + }, + { + "epoch": 0.4370945172369928, + "grad_norm": 2.5421438217163086, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8910043239593506, + "num_tokens": 133603298.0, + "step": 3436 + }, + { + "epoch": 0.43722172751558325, + "grad_norm": 2.942274570465088, + "learning_rate": 1e-06, + "loss": 0.4527, + "mean_token_accuracy": 0.8415495157241821, + "num_tokens": 133642012.0, + "step": 3437 + }, + { + "epoch": 0.4373489377941738, + "grad_norm": 2.318983316421509, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8900748491287231, + "num_tokens": 133686343.0, + "step": 3438 + }, + { + "epoch": 0.4374761480727643, + "grad_norm": 2.6673810482025146, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.8632853031158447, + "num_tokens": 133729984.0, + "step": 3439 + }, + { + "epoch": 0.4376033583513548, + "grad_norm": 2.4447858333587646, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8781378269195557, + "num_tokens": 133772285.0, + "step": 3440 + }, + { + "epoch": 0.4377305686299453, + "grad_norm": 2.569031000137329, + "learning_rate": 1e-06, + "loss": 0.3912, + "mean_token_accuracy": 0.8622770309448242, + "num_tokens": 133815481.0, + "step": 3441 + }, + { + "epoch": 0.43785777890853583, + "grad_norm": 2.365344762802124, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8778533935546875, + "num_tokens": 133859390.0, + "step": 3442 + }, + { + "epoch": 0.4379849891871263, + "grad_norm": 2.38633394241333, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8828211426734924, + "num_tokens": 133906036.0, + "step": 3443 + }, + { + "epoch": 0.43811219946571683, + "grad_norm": 2.6538233757019043, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8775650262832642, + "num_tokens": 133942675.0, + "step": 3444 + }, + { + "epoch": 0.43823940974430736, + "grad_norm": 2.7524595260620117, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8742532134056091, + "num_tokens": 133980305.0, + "step": 3445 + }, + { + "epoch": 0.43836662002289783, + "grad_norm": 2.8571224212646484, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8702990412712097, + "num_tokens": 134015911.0, + "step": 3446 + }, + { + "epoch": 0.43849383030148836, + "grad_norm": 2.8776657581329346, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8838579654693604, + "num_tokens": 134044250.0, + "step": 3447 + }, + { + "epoch": 0.4386210405800789, + "grad_norm": 2.640523910522461, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8817917108535767, + "num_tokens": 134081346.0, + "step": 3448 + }, + { + "epoch": 0.43874825085866936, + "grad_norm": 2.6488451957702637, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8809466361999512, + "num_tokens": 134117711.0, + "step": 3449 + }, + { + "epoch": 0.4388754611372599, + "grad_norm": 2.7272231578826904, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.864940881729126, + "num_tokens": 134159185.0, + "step": 3450 + }, + { + "epoch": 0.4390026714158504, + "grad_norm": 2.7955546379089355, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8769919872283936, + "num_tokens": 134193895.0, + "step": 3451 + }, + { + "epoch": 0.4391298816944409, + "grad_norm": 2.7340807914733887, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8716270923614502, + "num_tokens": 134233154.0, + "step": 3452 + }, + { + "epoch": 0.4392570919730314, + "grad_norm": 2.781735897064209, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8738361597061157, + "num_tokens": 134270555.0, + "step": 3453 + }, + { + "epoch": 0.43938430225162195, + "grad_norm": 2.827918767929077, + "learning_rate": 1e-06, + "loss": 0.4059, + "mean_token_accuracy": 0.8605045080184937, + "num_tokens": 134312179.0, + "step": 3454 + }, + { + "epoch": 0.4395115125302124, + "grad_norm": 2.6342592239379883, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8716222643852234, + "num_tokens": 134350003.0, + "step": 3455 + }, + { + "epoch": 0.43963872280880295, + "grad_norm": 2.7147727012634277, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8757990598678589, + "num_tokens": 134386335.0, + "step": 3456 + }, + { + "epoch": 0.4397659330873935, + "grad_norm": 2.8450536727905273, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8725485801696777, + "num_tokens": 134423562.0, + "step": 3457 + }, + { + "epoch": 0.43989314336598395, + "grad_norm": 2.4519691467285156, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8772321939468384, + "num_tokens": 134467894.0, + "step": 3458 + }, + { + "epoch": 0.4400203536445745, + "grad_norm": 2.705146551132202, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8721839189529419, + "num_tokens": 134506140.0, + "step": 3459 + }, + { + "epoch": 0.440147563923165, + "grad_norm": 2.617647647857666, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8787856101989746, + "num_tokens": 134544275.0, + "step": 3460 + }, + { + "epoch": 0.4402747742017555, + "grad_norm": 2.523555040359497, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.863054633140564, + "num_tokens": 134589360.0, + "step": 3461 + }, + { + "epoch": 0.440401984480346, + "grad_norm": 3.008450984954834, + "learning_rate": 1e-06, + "loss": 0.3815, + "mean_token_accuracy": 0.8676176071166992, + "num_tokens": 134622788.0, + "step": 3462 + }, + { + "epoch": 0.44052919475893654, + "grad_norm": 2.709320545196533, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8651062846183777, + "num_tokens": 134662181.0, + "step": 3463 + }, + { + "epoch": 0.440656405037527, + "grad_norm": 2.641747236251831, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8727847337722778, + "num_tokens": 134702659.0, + "step": 3464 + }, + { + "epoch": 0.44078361531611754, + "grad_norm": 2.5065739154815674, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8763342499732971, + "num_tokens": 134743978.0, + "step": 3465 + }, + { + "epoch": 0.44091082559470807, + "grad_norm": 2.5045669078826904, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8829745054244995, + "num_tokens": 134782784.0, + "step": 3466 + }, + { + "epoch": 0.44103803587329854, + "grad_norm": 2.6854970455169678, + "learning_rate": 1e-06, + "loss": 0.409, + "mean_token_accuracy": 0.8610724210739136, + "num_tokens": 134825104.0, + "step": 3467 + }, + { + "epoch": 0.44116524615188907, + "grad_norm": 2.7030811309814453, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.86516934633255, + "num_tokens": 134864284.0, + "step": 3468 + }, + { + "epoch": 0.4412924564304796, + "grad_norm": 2.472644567489624, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8611769080162048, + "num_tokens": 134913867.0, + "step": 3469 + }, + { + "epoch": 0.44141966670907007, + "grad_norm": 2.7095370292663574, + "learning_rate": 1e-06, + "loss": 0.4036, + "mean_token_accuracy": 0.8631500005722046, + "num_tokens": 134957945.0, + "step": 3470 + }, + { + "epoch": 0.4415468769876606, + "grad_norm": 2.5214335918426514, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8855215311050415, + "num_tokens": 134994892.0, + "step": 3471 + }, + { + "epoch": 0.4416740872662511, + "grad_norm": 2.708414077758789, + "learning_rate": 1e-06, + "loss": 0.4276, + "mean_token_accuracy": 0.8527075052261353, + "num_tokens": 135037506.0, + "step": 3472 + }, + { + "epoch": 0.4418012975448416, + "grad_norm": 2.5539052486419678, + "learning_rate": 1e-06, + "loss": 0.4002, + "mean_token_accuracy": 0.8648066520690918, + "num_tokens": 135081247.0, + "step": 3473 + }, + { + "epoch": 0.4419285078234321, + "grad_norm": 2.5424137115478516, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8822575807571411, + "num_tokens": 135122204.0, + "step": 3474 + }, + { + "epoch": 0.44205571810202265, + "grad_norm": 2.779144525527954, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.870750904083252, + "num_tokens": 135160272.0, + "step": 3475 + }, + { + "epoch": 0.4421829283806131, + "grad_norm": 2.7427849769592285, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8645133972167969, + "num_tokens": 135201199.0, + "step": 3476 + }, + { + "epoch": 0.44231013865920366, + "grad_norm": 2.5188539028167725, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8719208240509033, + "num_tokens": 135245144.0, + "step": 3477 + }, + { + "epoch": 0.4424373489377942, + "grad_norm": 2.592280864715576, + "learning_rate": 1e-06, + "loss": 0.4132, + "mean_token_accuracy": 0.8598384261131287, + "num_tokens": 135288668.0, + "step": 3478 + }, + { + "epoch": 0.44256455921638466, + "grad_norm": 2.65700626373291, + "learning_rate": 1e-06, + "loss": 0.4254, + "mean_token_accuracy": 0.8548660278320312, + "num_tokens": 135333116.0, + "step": 3479 + }, + { + "epoch": 0.4426917694949752, + "grad_norm": 2.780083656311035, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8737967014312744, + "num_tokens": 135370941.0, + "step": 3480 + }, + { + "epoch": 0.4428189797735657, + "grad_norm": 2.7428033351898193, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8714749813079834, + "num_tokens": 135407915.0, + "step": 3481 + }, + { + "epoch": 0.44294619005215624, + "grad_norm": 2.810753345489502, + "learning_rate": 1e-06, + "loss": 0.3964, + "mean_token_accuracy": 0.8625319600105286, + "num_tokens": 135443861.0, + "step": 3482 + }, + { + "epoch": 0.4430734003307467, + "grad_norm": 2.545657157897949, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8807567954063416, + "num_tokens": 135483399.0, + "step": 3483 + }, + { + "epoch": 0.44320061060933724, + "grad_norm": 2.8914530277252197, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8692320585250854, + "num_tokens": 135520195.0, + "step": 3484 + }, + { + "epoch": 0.44332782088792777, + "grad_norm": 2.7217071056365967, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8683438301086426, + "num_tokens": 135559357.0, + "step": 3485 + }, + { + "epoch": 0.44345503116651824, + "grad_norm": 2.689650297164917, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8926952481269836, + "num_tokens": 135592507.0, + "step": 3486 + }, + { + "epoch": 0.44358224144510877, + "grad_norm": 2.863814353942871, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.866611897945404, + "num_tokens": 135627653.0, + "step": 3487 + }, + { + "epoch": 0.4437094517236993, + "grad_norm": 2.8355112075805664, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8800866603851318, + "num_tokens": 135662690.0, + "step": 3488 + }, + { + "epoch": 0.4438366620022898, + "grad_norm": 2.6004128456115723, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8714935779571533, + "num_tokens": 135703148.0, + "step": 3489 + }, + { + "epoch": 0.4439638722808803, + "grad_norm": 2.7220678329467773, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8647339940071106, + "num_tokens": 135743141.0, + "step": 3490 + }, + { + "epoch": 0.44409108255947083, + "grad_norm": 2.743128776550293, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8894569873809814, + "num_tokens": 135775815.0, + "step": 3491 + }, + { + "epoch": 0.4442182928380613, + "grad_norm": 2.6903204917907715, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8738136291503906, + "num_tokens": 135814909.0, + "step": 3492 + }, + { + "epoch": 0.44434550311665183, + "grad_norm": 2.77105712890625, + "learning_rate": 1e-06, + "loss": 0.4016, + "mean_token_accuracy": 0.8620404601097107, + "num_tokens": 135851332.0, + "step": 3493 + }, + { + "epoch": 0.44447271339524236, + "grad_norm": 2.5254642963409424, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8801610469818115, + "num_tokens": 135889436.0, + "step": 3494 + }, + { + "epoch": 0.44459992367383283, + "grad_norm": 2.603802442550659, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8764263987541199, + "num_tokens": 135930215.0, + "step": 3495 + }, + { + "epoch": 0.44472713395242336, + "grad_norm": 2.6891844272613525, + "learning_rate": 1e-06, + "loss": 0.4127, + "mean_token_accuracy": 0.8592727184295654, + "num_tokens": 135971707.0, + "step": 3496 + }, + { + "epoch": 0.4448543442310139, + "grad_norm": 2.5613646507263184, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8747363090515137, + "num_tokens": 136014049.0, + "step": 3497 + }, + { + "epoch": 0.44498155450960436, + "grad_norm": 2.7236621379852295, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.87019282579422, + "num_tokens": 136051521.0, + "step": 3498 + }, + { + "epoch": 0.4451087647881949, + "grad_norm": 2.8734188079833984, + "learning_rate": 1e-06, + "loss": 0.4249, + "mean_token_accuracy": 0.8587772846221924, + "num_tokens": 136090000.0, + "step": 3499 + }, + { + "epoch": 0.4452359750667854, + "grad_norm": 2.47794246673584, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8873205780982971, + "num_tokens": 136130449.0, + "step": 3500 + }, + { + "epoch": 0.4453631853453759, + "grad_norm": 2.398050546646118, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8724002838134766, + "num_tokens": 136175253.0, + "step": 3501 + }, + { + "epoch": 0.4454903956239664, + "grad_norm": 2.78235125541687, + "learning_rate": 1e-06, + "loss": 0.388, + "mean_token_accuracy": 0.8684326410293579, + "num_tokens": 136215556.0, + "step": 3502 + }, + { + "epoch": 0.44561760590255695, + "grad_norm": 2.606802225112915, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8710118532180786, + "num_tokens": 136256959.0, + "step": 3503 + }, + { + "epoch": 0.4457448161811474, + "grad_norm": 2.6888914108276367, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8870654106140137, + "num_tokens": 136291023.0, + "step": 3504 + }, + { + "epoch": 0.44587202645973795, + "grad_norm": 2.9601240158081055, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8714854717254639, + "num_tokens": 136326452.0, + "step": 3505 + }, + { + "epoch": 0.4459992367383285, + "grad_norm": 2.636591672897339, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8800134062767029, + "num_tokens": 136364984.0, + "step": 3506 + }, + { + "epoch": 0.44612644701691895, + "grad_norm": 2.397322177886963, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8939571380615234, + "num_tokens": 136404859.0, + "step": 3507 + }, + { + "epoch": 0.4462536572955095, + "grad_norm": 2.586679458618164, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8672825694084167, + "num_tokens": 136447317.0, + "step": 3508 + }, + { + "epoch": 0.4463808675741, + "grad_norm": 2.8769490718841553, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8747605681419373, + "num_tokens": 136480499.0, + "step": 3509 + }, + { + "epoch": 0.4465080778526905, + "grad_norm": 2.7954673767089844, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8755484819412231, + "num_tokens": 136519635.0, + "step": 3510 + }, + { + "epoch": 0.446635288131281, + "grad_norm": 2.5438365936279297, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8867770433425903, + "num_tokens": 136558166.0, + "step": 3511 + }, + { + "epoch": 0.44676249840987153, + "grad_norm": 2.5200207233428955, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8755032420158386, + "num_tokens": 136599463.0, + "step": 3512 + }, + { + "epoch": 0.446889708688462, + "grad_norm": 2.739720106124878, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8690178990364075, + "num_tokens": 136637504.0, + "step": 3513 + }, + { + "epoch": 0.44701691896705253, + "grad_norm": 2.543818950653076, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8896937370300293, + "num_tokens": 136675417.0, + "step": 3514 + }, + { + "epoch": 0.44714412924564306, + "grad_norm": 2.5837934017181396, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8947404623031616, + "num_tokens": 136711847.0, + "step": 3515 + }, + { + "epoch": 0.44727133952423354, + "grad_norm": 2.566987991333008, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8899261355400085, + "num_tokens": 136751578.0, + "step": 3516 + }, + { + "epoch": 0.44739854980282406, + "grad_norm": 2.992156505584717, + "learning_rate": 1e-06, + "loss": 0.4277, + "mean_token_accuracy": 0.8556511402130127, + "num_tokens": 136787965.0, + "step": 3517 + }, + { + "epoch": 0.4475257600814146, + "grad_norm": 2.5564398765563965, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8784064054489136, + "num_tokens": 136831614.0, + "step": 3518 + }, + { + "epoch": 0.44765297036000506, + "grad_norm": 2.9935522079467773, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.8709695339202881, + "num_tokens": 136862720.0, + "step": 3519 + }, + { + "epoch": 0.4477801806385956, + "grad_norm": 2.599008321762085, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8776569366455078, + "num_tokens": 136899938.0, + "step": 3520 + }, + { + "epoch": 0.4479073909171861, + "grad_norm": 2.4962382316589355, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.879959225654602, + "num_tokens": 136943550.0, + "step": 3521 + }, + { + "epoch": 0.4480346011957766, + "grad_norm": 2.606990098953247, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8796294331550598, + "num_tokens": 136981980.0, + "step": 3522 + }, + { + "epoch": 0.4481618114743671, + "grad_norm": 2.487623691558838, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8916200995445251, + "num_tokens": 137019144.0, + "step": 3523 + }, + { + "epoch": 0.44828902175295765, + "grad_norm": 2.5373387336730957, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.872916579246521, + "num_tokens": 137063061.0, + "step": 3524 + }, + { + "epoch": 0.4484162320315481, + "grad_norm": 2.811004400253296, + "learning_rate": 1e-06, + "loss": 0.4131, + "mean_token_accuracy": 0.8581633567810059, + "num_tokens": 137100943.0, + "step": 3525 + }, + { + "epoch": 0.44854344231013865, + "grad_norm": 2.6775221824645996, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.8623600006103516, + "num_tokens": 137142987.0, + "step": 3526 + }, + { + "epoch": 0.4486706525887292, + "grad_norm": 2.474431037902832, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8869882822036743, + "num_tokens": 137183120.0, + "step": 3527 + }, + { + "epoch": 0.44879786286731965, + "grad_norm": 2.7679243087768555, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8720808029174805, + "num_tokens": 137219734.0, + "step": 3528 + }, + { + "epoch": 0.4489250731459102, + "grad_norm": 2.5270373821258545, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8795707821846008, + "num_tokens": 137261637.0, + "step": 3529 + }, + { + "epoch": 0.4490522834245007, + "grad_norm": 2.820333957672119, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8714549541473389, + "num_tokens": 137300897.0, + "step": 3530 + }, + { + "epoch": 0.44917949370309124, + "grad_norm": 2.6265265941619873, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8743414878845215, + "num_tokens": 137341670.0, + "step": 3531 + }, + { + "epoch": 0.4493067039816817, + "grad_norm": 2.501291513442993, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8760861158370972, + "num_tokens": 137381119.0, + "step": 3532 + }, + { + "epoch": 0.44943391426027224, + "grad_norm": 2.664227247238159, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8715517520904541, + "num_tokens": 137421222.0, + "step": 3533 + }, + { + "epoch": 0.44956112453886277, + "grad_norm": 2.8735692501068115, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8627562522888184, + "num_tokens": 137462469.0, + "step": 3534 + }, + { + "epoch": 0.44968833481745324, + "grad_norm": 2.8154964447021484, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8670787811279297, + "num_tokens": 137499519.0, + "step": 3535 + }, + { + "epoch": 0.44981554509604377, + "grad_norm": 2.5505359172821045, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.887161374092102, + "num_tokens": 137539672.0, + "step": 3536 + }, + { + "epoch": 0.4499427553746343, + "grad_norm": 2.768245220184326, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8720520734786987, + "num_tokens": 137579957.0, + "step": 3537 + }, + { + "epoch": 0.45006996565322477, + "grad_norm": 2.785459518432617, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8697679042816162, + "num_tokens": 137620942.0, + "step": 3538 + }, + { + "epoch": 0.4501971759318153, + "grad_norm": 2.673701763153076, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.8645623326301575, + "num_tokens": 137662079.0, + "step": 3539 + }, + { + "epoch": 0.4503243862104058, + "grad_norm": 2.382054567337036, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.885066568851471, + "num_tokens": 137705166.0, + "step": 3540 + }, + { + "epoch": 0.4504515964889963, + "grad_norm": 2.8417694568634033, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8797296285629272, + "num_tokens": 137735958.0, + "step": 3541 + }, + { + "epoch": 0.4505788067675868, + "grad_norm": 2.642543077468872, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.8705089092254639, + "num_tokens": 137779353.0, + "step": 3542 + }, + { + "epoch": 0.45070601704617735, + "grad_norm": 2.6944568157196045, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8666313886642456, + "num_tokens": 137820111.0, + "step": 3543 + }, + { + "epoch": 0.4508332273247678, + "grad_norm": 2.3076891899108887, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8881227374076843, + "num_tokens": 137862544.0, + "step": 3544 + }, + { + "epoch": 0.45096043760335836, + "grad_norm": 2.9121041297912598, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.877778172492981, + "num_tokens": 137899549.0, + "step": 3545 + }, + { + "epoch": 0.4510876478819489, + "grad_norm": 2.6047449111938477, + "learning_rate": 1e-06, + "loss": 0.3824, + "mean_token_accuracy": 0.8662441372871399, + "num_tokens": 137939468.0, + "step": 3546 + }, + { + "epoch": 0.45121485816053936, + "grad_norm": 2.7734577655792236, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8803815245628357, + "num_tokens": 137972415.0, + "step": 3547 + }, + { + "epoch": 0.4513420684391299, + "grad_norm": 2.695974349975586, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8897868394851685, + "num_tokens": 138009057.0, + "step": 3548 + }, + { + "epoch": 0.4514692787177204, + "grad_norm": 2.6540729999542236, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8826813697814941, + "num_tokens": 138045799.0, + "step": 3549 + }, + { + "epoch": 0.4515964889963109, + "grad_norm": 2.3902933597564697, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8812599182128906, + "num_tokens": 138089056.0, + "step": 3550 + }, + { + "epoch": 0.4517236992749014, + "grad_norm": 2.653177499771118, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8740077614784241, + "num_tokens": 138127010.0, + "step": 3551 + }, + { + "epoch": 0.45185090955349194, + "grad_norm": 2.422994375228882, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8838582038879395, + "num_tokens": 138167134.0, + "step": 3552 + }, + { + "epoch": 0.4519781198320824, + "grad_norm": 2.5544064044952393, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.8726596832275391, + "num_tokens": 138207427.0, + "step": 3553 + }, + { + "epoch": 0.45210533011067294, + "grad_norm": 2.6863420009613037, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8765110969543457, + "num_tokens": 138246266.0, + "step": 3554 + }, + { + "epoch": 0.45223254038926347, + "grad_norm": 2.9080395698547363, + "learning_rate": 1e-06, + "loss": 0.4404, + "mean_token_accuracy": 0.848398745059967, + "num_tokens": 138284972.0, + "step": 3555 + }, + { + "epoch": 0.45235975066785394, + "grad_norm": 2.9843273162841797, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8798742890357971, + "num_tokens": 138316460.0, + "step": 3556 + }, + { + "epoch": 0.45248696094644447, + "grad_norm": 2.3661911487579346, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8765050172805786, + "num_tokens": 138363346.0, + "step": 3557 + }, + { + "epoch": 0.452614171225035, + "grad_norm": 2.9196462631225586, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.8618003129959106, + "num_tokens": 138398393.0, + "step": 3558 + }, + { + "epoch": 0.4527413815036255, + "grad_norm": 2.7685325145721436, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8706475496292114, + "num_tokens": 138438701.0, + "step": 3559 + }, + { + "epoch": 0.452868591782216, + "grad_norm": 2.9013986587524414, + "learning_rate": 1e-06, + "loss": 0.4262, + "mean_token_accuracy": 0.8534508347511292, + "num_tokens": 138481967.0, + "step": 3560 + }, + { + "epoch": 0.45299580206080653, + "grad_norm": 2.770132541656494, + "learning_rate": 1e-06, + "loss": 0.3904, + "mean_token_accuracy": 0.867694079875946, + "num_tokens": 138522864.0, + "step": 3561 + }, + { + "epoch": 0.453123012339397, + "grad_norm": 2.539501190185547, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9001166224479675, + "num_tokens": 138557270.0, + "step": 3562 + }, + { + "epoch": 0.45325022261798753, + "grad_norm": 2.5206096172332764, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8827618956565857, + "num_tokens": 138596652.0, + "step": 3563 + }, + { + "epoch": 0.45337743289657806, + "grad_norm": 2.4015707969665527, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8722797632217407, + "num_tokens": 138645005.0, + "step": 3564 + }, + { + "epoch": 0.45350464317516853, + "grad_norm": 2.53437876701355, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.875580906867981, + "num_tokens": 138686588.0, + "step": 3565 + }, + { + "epoch": 0.45363185345375906, + "grad_norm": 2.612959146499634, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.8670451641082764, + "num_tokens": 138731964.0, + "step": 3566 + }, + { + "epoch": 0.4537590637323496, + "grad_norm": 2.763718605041504, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8917187452316284, + "num_tokens": 138767370.0, + "step": 3567 + }, + { + "epoch": 0.45388627401094006, + "grad_norm": 2.4688336849212646, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8888731002807617, + "num_tokens": 138808873.0, + "step": 3568 + }, + { + "epoch": 0.4540134842895306, + "grad_norm": 2.494990348815918, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8827751874923706, + "num_tokens": 138849897.0, + "step": 3569 + }, + { + "epoch": 0.4541406945681211, + "grad_norm": 2.6534388065338135, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8761571645736694, + "num_tokens": 138889452.0, + "step": 3570 + }, + { + "epoch": 0.4542679048467116, + "grad_norm": 2.81266713142395, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8747600317001343, + "num_tokens": 138924685.0, + "step": 3571 + }, + { + "epoch": 0.4543951151253021, + "grad_norm": 3.1523990631103516, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.8652043342590332, + "num_tokens": 138957889.0, + "step": 3572 + }, + { + "epoch": 0.45452232540389265, + "grad_norm": 2.732785701751709, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.869768500328064, + "num_tokens": 138999161.0, + "step": 3573 + }, + { + "epoch": 0.4546495356824831, + "grad_norm": 2.935188055038452, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8727349042892456, + "num_tokens": 139032717.0, + "step": 3574 + }, + { + "epoch": 0.45477674596107365, + "grad_norm": 2.7012643814086914, + "learning_rate": 1e-06, + "loss": 0.4148, + "mean_token_accuracy": 0.8578013181686401, + "num_tokens": 139075839.0, + "step": 3575 + }, + { + "epoch": 0.4549039562396642, + "grad_norm": 2.9739291667938232, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8847833275794983, + "num_tokens": 139106621.0, + "step": 3576 + }, + { + "epoch": 0.45503116651825465, + "grad_norm": 2.8199875354766846, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8753185868263245, + "num_tokens": 139140963.0, + "step": 3577 + }, + { + "epoch": 0.4551583767968452, + "grad_norm": 2.7498159408569336, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8828469514846802, + "num_tokens": 139176131.0, + "step": 3578 + }, + { + "epoch": 0.4552855870754357, + "grad_norm": 2.720621109008789, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8760230541229248, + "num_tokens": 139212627.0, + "step": 3579 + }, + { + "epoch": 0.4554127973540262, + "grad_norm": 2.689359188079834, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8709760904312134, + "num_tokens": 139252739.0, + "step": 3580 + }, + { + "epoch": 0.4555400076326167, + "grad_norm": 2.8108460903167725, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.858971893787384, + "num_tokens": 139293336.0, + "step": 3581 + }, + { + "epoch": 0.45566721791120723, + "grad_norm": 2.594165563583374, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8851362466812134, + "num_tokens": 139331300.0, + "step": 3582 + }, + { + "epoch": 0.45579442818979776, + "grad_norm": 2.8118090629577637, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8721898794174194, + "num_tokens": 139366938.0, + "step": 3583 + }, + { + "epoch": 0.45592163846838824, + "grad_norm": 2.6468541622161865, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.86710524559021, + "num_tokens": 139408871.0, + "step": 3584 + }, + { + "epoch": 0.45604884874697876, + "grad_norm": 2.6367976665496826, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.8618932962417603, + "num_tokens": 139449623.0, + "step": 3585 + }, + { + "epoch": 0.4561760590255693, + "grad_norm": 2.8446097373962402, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8697793483734131, + "num_tokens": 139485454.0, + "step": 3586 + }, + { + "epoch": 0.45630326930415976, + "grad_norm": 2.709672689437866, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8760756254196167, + "num_tokens": 139521074.0, + "step": 3587 + }, + { + "epoch": 0.4564304795827503, + "grad_norm": 2.5780577659606934, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8928454518318176, + "num_tokens": 139554589.0, + "step": 3588 + }, + { + "epoch": 0.4565576898613408, + "grad_norm": 2.592937469482422, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8844642639160156, + "num_tokens": 139593580.0, + "step": 3589 + }, + { + "epoch": 0.4566849001399313, + "grad_norm": 2.5081191062927246, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8842787146568298, + "num_tokens": 139635597.0, + "step": 3590 + }, + { + "epoch": 0.4568121104185218, + "grad_norm": 2.843782424926758, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.881716251373291, + "num_tokens": 139667929.0, + "step": 3591 + }, + { + "epoch": 0.45693932069711235, + "grad_norm": 2.7527928352355957, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8732245564460754, + "num_tokens": 139706538.0, + "step": 3592 + }, + { + "epoch": 0.4570665309757028, + "grad_norm": 3.0614161491394043, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.8661390542984009, + "num_tokens": 139739306.0, + "step": 3593 + }, + { + "epoch": 0.45719374125429335, + "grad_norm": 2.779346466064453, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8789982795715332, + "num_tokens": 139772482.0, + "step": 3594 + }, + { + "epoch": 0.4573209515328839, + "grad_norm": 2.736097574234009, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8833439350128174, + "num_tokens": 139808938.0, + "step": 3595 + }, + { + "epoch": 0.45744816181147435, + "grad_norm": 2.9848737716674805, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.8682314157485962, + "num_tokens": 139842733.0, + "step": 3596 + }, + { + "epoch": 0.4575753720900649, + "grad_norm": 2.7198116779327393, + "learning_rate": 1e-06, + "loss": 0.3985, + "mean_token_accuracy": 0.8652188777923584, + "num_tokens": 139884452.0, + "step": 3597 + }, + { + "epoch": 0.4577025823686554, + "grad_norm": 2.6615638732910156, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8668582439422607, + "num_tokens": 139927297.0, + "step": 3598 + }, + { + "epoch": 0.4578297926472459, + "grad_norm": 2.6421566009521484, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.8680802583694458, + "num_tokens": 139969184.0, + "step": 3599 + }, + { + "epoch": 0.4579570029258364, + "grad_norm": 2.801851987838745, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8739912509918213, + "num_tokens": 140004059.0, + "step": 3600 + }, + { + "epoch": 0.45808421320442694, + "grad_norm": 2.4959847927093506, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8764108419418335, + "num_tokens": 140046706.0, + "step": 3601 + }, + { + "epoch": 0.4582114234830174, + "grad_norm": 2.5047967433929443, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.8674805164337158, + "num_tokens": 140089482.0, + "step": 3602 + }, + { + "epoch": 0.45833863376160794, + "grad_norm": 2.7160465717315674, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8700737357139587, + "num_tokens": 140126322.0, + "step": 3603 + }, + { + "epoch": 0.45846584404019847, + "grad_norm": 2.3949806690216064, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8910922408103943, + "num_tokens": 140170107.0, + "step": 3604 + }, + { + "epoch": 0.45859305431878894, + "grad_norm": 2.52488112449646, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.882379412651062, + "num_tokens": 140213236.0, + "step": 3605 + }, + { + "epoch": 0.45872026459737947, + "grad_norm": 2.7800042629241943, + "learning_rate": 1e-06, + "loss": 0.3867, + "mean_token_accuracy": 0.869314432144165, + "num_tokens": 140253632.0, + "step": 3606 + }, + { + "epoch": 0.45884747487597, + "grad_norm": 2.6335337162017822, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8668762445449829, + "num_tokens": 140294713.0, + "step": 3607 + }, + { + "epoch": 0.45897468515456047, + "grad_norm": 2.611110210418701, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8774208426475525, + "num_tokens": 140334105.0, + "step": 3608 + }, + { + "epoch": 0.459101895433151, + "grad_norm": 2.7490034103393555, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8754746913909912, + "num_tokens": 140367369.0, + "step": 3609 + }, + { + "epoch": 0.4592291057117415, + "grad_norm": 2.532578468322754, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8742618560791016, + "num_tokens": 140409169.0, + "step": 3610 + }, + { + "epoch": 0.459356315990332, + "grad_norm": 3.0015809535980225, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8666622042655945, + "num_tokens": 140442801.0, + "step": 3611 + }, + { + "epoch": 0.4594835262689225, + "grad_norm": 2.5440566539764404, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.876472532749176, + "num_tokens": 140483925.0, + "step": 3612 + }, + { + "epoch": 0.45961073654751305, + "grad_norm": 2.46055269241333, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8858715295791626, + "num_tokens": 140526342.0, + "step": 3613 + }, + { + "epoch": 0.4597379468261035, + "grad_norm": 2.8663387298583984, + "learning_rate": 1e-06, + "loss": 0.4172, + "mean_token_accuracy": 0.8560668230056763, + "num_tokens": 140562760.0, + "step": 3614 + }, + { + "epoch": 0.45986515710469406, + "grad_norm": 2.722358226776123, + "learning_rate": 1e-06, + "loss": 0.3929, + "mean_token_accuracy": 0.8645215034484863, + "num_tokens": 140604559.0, + "step": 3615 + }, + { + "epoch": 0.4599923673832846, + "grad_norm": 2.8422067165374756, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.877698540687561, + "num_tokens": 140638549.0, + "step": 3616 + }, + { + "epoch": 0.46011957766187506, + "grad_norm": 2.476728677749634, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8782476186752319, + "num_tokens": 140682517.0, + "step": 3617 + }, + { + "epoch": 0.4602467879404656, + "grad_norm": 2.6982617378234863, + "learning_rate": 1e-06, + "loss": 0.4077, + "mean_token_accuracy": 0.8641944527626038, + "num_tokens": 140723296.0, + "step": 3618 + }, + { + "epoch": 0.4603739982190561, + "grad_norm": 2.6875925064086914, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8705995678901672, + "num_tokens": 140762865.0, + "step": 3619 + }, + { + "epoch": 0.4605012084976466, + "grad_norm": 2.7048702239990234, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8770852088928223, + "num_tokens": 140799291.0, + "step": 3620 + }, + { + "epoch": 0.4606284187762371, + "grad_norm": 2.3770222663879395, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8938724398612976, + "num_tokens": 140838147.0, + "step": 3621 + }, + { + "epoch": 0.46075562905482764, + "grad_norm": 2.8708066940307617, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8785492181777954, + "num_tokens": 140870019.0, + "step": 3622 + }, + { + "epoch": 0.4608828393334181, + "grad_norm": 2.812875509262085, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8780750036239624, + "num_tokens": 140901568.0, + "step": 3623 + }, + { + "epoch": 0.46101004961200864, + "grad_norm": 2.5460116863250732, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8848316073417664, + "num_tokens": 140939078.0, + "step": 3624 + }, + { + "epoch": 0.46113725989059917, + "grad_norm": 2.7829666137695312, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8732218146324158, + "num_tokens": 140977070.0, + "step": 3625 + }, + { + "epoch": 0.46126447016918964, + "grad_norm": 2.5310444831848145, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8880166411399841, + "num_tokens": 141018791.0, + "step": 3626 + }, + { + "epoch": 0.4613916804477802, + "grad_norm": 2.4430861473083496, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8837134838104248, + "num_tokens": 141061752.0, + "step": 3627 + }, + { + "epoch": 0.4615188907263707, + "grad_norm": 2.708177328109741, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8673436045646667, + "num_tokens": 141100633.0, + "step": 3628 + }, + { + "epoch": 0.4616461010049612, + "grad_norm": 2.7475335597991943, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8662861585617065, + "num_tokens": 141137371.0, + "step": 3629 + }, + { + "epoch": 0.4617733112835517, + "grad_norm": 3.0712387561798096, + "learning_rate": 1e-06, + "loss": 0.4384, + "mean_token_accuracy": 0.8543773293495178, + "num_tokens": 141171344.0, + "step": 3630 + }, + { + "epoch": 0.46190052156214223, + "grad_norm": 2.5525853633880615, + "learning_rate": 1e-06, + "loss": 0.4228, + "mean_token_accuracy": 0.8556900024414062, + "num_tokens": 141215752.0, + "step": 3631 + }, + { + "epoch": 0.46202773184073276, + "grad_norm": 2.6870810985565186, + "learning_rate": 1e-06, + "loss": 0.4043, + "mean_token_accuracy": 0.860674262046814, + "num_tokens": 141258530.0, + "step": 3632 + }, + { + "epoch": 0.46215494211932323, + "grad_norm": 2.745839834213257, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8711751103401184, + "num_tokens": 141298168.0, + "step": 3633 + }, + { + "epoch": 0.46228215239791376, + "grad_norm": 2.530966281890869, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8867063522338867, + "num_tokens": 141335854.0, + "step": 3634 + }, + { + "epoch": 0.4624093626765043, + "grad_norm": 2.6187191009521484, + "learning_rate": 1e-06, + "loss": 0.4288, + "mean_token_accuracy": 0.8544144630432129, + "num_tokens": 141377428.0, + "step": 3635 + }, + { + "epoch": 0.46253657295509476, + "grad_norm": 2.620574474334717, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8630955219268799, + "num_tokens": 141421224.0, + "step": 3636 + }, + { + "epoch": 0.4626637832336853, + "grad_norm": 3.0764260292053223, + "learning_rate": 1e-06, + "loss": 0.4443, + "mean_token_accuracy": 0.8451385498046875, + "num_tokens": 141453838.0, + "step": 3637 + }, + { + "epoch": 0.4627909935122758, + "grad_norm": 2.7624335289001465, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8816997408866882, + "num_tokens": 141489032.0, + "step": 3638 + }, + { + "epoch": 0.4629182037908663, + "grad_norm": 2.839155673980713, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.8652584552764893, + "num_tokens": 141525040.0, + "step": 3639 + }, + { + "epoch": 0.4630454140694568, + "grad_norm": 2.6530027389526367, + "learning_rate": 1e-06, + "loss": 0.395, + "mean_token_accuracy": 0.8596344590187073, + "num_tokens": 141565379.0, + "step": 3640 + }, + { + "epoch": 0.46317262434804735, + "grad_norm": 2.632235288619995, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8802821636199951, + "num_tokens": 141603821.0, + "step": 3641 + }, + { + "epoch": 0.4632998346266378, + "grad_norm": 2.602299690246582, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8649439215660095, + "num_tokens": 141647092.0, + "step": 3642 + }, + { + "epoch": 0.46342704490522835, + "grad_norm": 2.7740819454193115, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8857684135437012, + "num_tokens": 141686569.0, + "step": 3643 + }, + { + "epoch": 0.4635542551838189, + "grad_norm": 2.830341339111328, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8841127157211304, + "num_tokens": 141718153.0, + "step": 3644 + }, + { + "epoch": 0.46368146546240935, + "grad_norm": 2.5733802318573, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8796091079711914, + "num_tokens": 141760414.0, + "step": 3645 + }, + { + "epoch": 0.4638086757409999, + "grad_norm": 2.465075969696045, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8701204061508179, + "num_tokens": 141804387.0, + "step": 3646 + }, + { + "epoch": 0.4639358860195904, + "grad_norm": 2.594473123550415, + "learning_rate": 1e-06, + "loss": 0.4203, + "mean_token_accuracy": 0.8567736148834229, + "num_tokens": 141850743.0, + "step": 3647 + }, + { + "epoch": 0.4640630962981809, + "grad_norm": 2.6355223655700684, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8700956106185913, + "num_tokens": 141891352.0, + "step": 3648 + }, + { + "epoch": 0.4641903065767714, + "grad_norm": 2.487360954284668, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.895645022392273, + "num_tokens": 141930648.0, + "step": 3649 + }, + { + "epoch": 0.46431751685536193, + "grad_norm": 2.7259693145751953, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8735935688018799, + "num_tokens": 141968596.0, + "step": 3650 + }, + { + "epoch": 0.4644447271339524, + "grad_norm": 2.6060070991516113, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8779149055480957, + "num_tokens": 142006712.0, + "step": 3651 + }, + { + "epoch": 0.46457193741254293, + "grad_norm": 2.734321117401123, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.865696132183075, + "num_tokens": 142045258.0, + "step": 3652 + }, + { + "epoch": 0.46469914769113346, + "grad_norm": 2.652243137359619, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.8737409114837646, + "num_tokens": 142086893.0, + "step": 3653 + }, + { + "epoch": 0.46482635796972394, + "grad_norm": 2.4615981578826904, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8906741738319397, + "num_tokens": 142126962.0, + "step": 3654 + }, + { + "epoch": 0.46495356824831446, + "grad_norm": 2.727105140686035, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8852180242538452, + "num_tokens": 142161657.0, + "step": 3655 + }, + { + "epoch": 0.465080778526905, + "grad_norm": 2.6394925117492676, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.873755931854248, + "num_tokens": 142201545.0, + "step": 3656 + }, + { + "epoch": 0.46520798880549546, + "grad_norm": 2.514190435409546, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.866992712020874, + "num_tokens": 142248424.0, + "step": 3657 + }, + { + "epoch": 0.465335199084086, + "grad_norm": 2.8039402961730957, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.886306643486023, + "num_tokens": 142283804.0, + "step": 3658 + }, + { + "epoch": 0.4654624093626765, + "grad_norm": 2.6999998092651367, + "learning_rate": 1e-06, + "loss": 0.361, + "mean_token_accuracy": 0.8757822513580322, + "num_tokens": 142321446.0, + "step": 3659 + }, + { + "epoch": 0.465589619641267, + "grad_norm": 2.592094898223877, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8624378442764282, + "num_tokens": 142364461.0, + "step": 3660 + }, + { + "epoch": 0.4657168299198575, + "grad_norm": 2.656791925430298, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8828498125076294, + "num_tokens": 142401233.0, + "step": 3661 + }, + { + "epoch": 0.46584404019844805, + "grad_norm": 2.685018539428711, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8695393800735474, + "num_tokens": 142441183.0, + "step": 3662 + }, + { + "epoch": 0.4659712504770385, + "grad_norm": 2.7767300605773926, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8758971691131592, + "num_tokens": 142479222.0, + "step": 3663 + }, + { + "epoch": 0.46609846075562905, + "grad_norm": 2.860016345977783, + "learning_rate": 1e-06, + "loss": 0.418, + "mean_token_accuracy": 0.8594541549682617, + "num_tokens": 142515305.0, + "step": 3664 + }, + { + "epoch": 0.4662256710342196, + "grad_norm": 2.8441457748413086, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8809608817100525, + "num_tokens": 142546012.0, + "step": 3665 + }, + { + "epoch": 0.46635288131281005, + "grad_norm": 2.6407124996185303, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8734991550445557, + "num_tokens": 142588074.0, + "step": 3666 + }, + { + "epoch": 0.4664800915914006, + "grad_norm": 2.5577239990234375, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8806906342506409, + "num_tokens": 142628288.0, + "step": 3667 + }, + { + "epoch": 0.4666073018699911, + "grad_norm": 2.5860002040863037, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8780204653739929, + "num_tokens": 142670174.0, + "step": 3668 + }, + { + "epoch": 0.4667345121485816, + "grad_norm": 2.5063905715942383, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8777004480361938, + "num_tokens": 142713978.0, + "step": 3669 + }, + { + "epoch": 0.4668617224271721, + "grad_norm": 2.3256590366363525, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8918979167938232, + "num_tokens": 142757261.0, + "step": 3670 + }, + { + "epoch": 0.46698893270576264, + "grad_norm": 2.714588165283203, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.8692418336868286, + "num_tokens": 142795840.0, + "step": 3671 + }, + { + "epoch": 0.4671161429843531, + "grad_norm": 2.345367908477783, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8839020729064941, + "num_tokens": 142842014.0, + "step": 3672 + }, + { + "epoch": 0.46724335326294364, + "grad_norm": 2.8502485752105713, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.8609781265258789, + "num_tokens": 142881746.0, + "step": 3673 + }, + { + "epoch": 0.46737056354153417, + "grad_norm": 2.9479737281799316, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8673055768013, + "num_tokens": 142920191.0, + "step": 3674 + }, + { + "epoch": 0.46749777382012464, + "grad_norm": 2.8141939640045166, + "learning_rate": 1e-06, + "loss": 0.4187, + "mean_token_accuracy": 0.8542107343673706, + "num_tokens": 142960779.0, + "step": 3675 + }, + { + "epoch": 0.46762498409871517, + "grad_norm": 2.688122510910034, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8750835657119751, + "num_tokens": 142999875.0, + "step": 3676 + }, + { + "epoch": 0.4677521943773057, + "grad_norm": 2.934004068374634, + "learning_rate": 1e-06, + "loss": 0.426, + "mean_token_accuracy": 0.8509983420372009, + "num_tokens": 143034691.0, + "step": 3677 + }, + { + "epoch": 0.46787940465589617, + "grad_norm": 2.8047256469726562, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8742300868034363, + "num_tokens": 143069777.0, + "step": 3678 + }, + { + "epoch": 0.4680066149344867, + "grad_norm": 2.6308348178863525, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8738990426063538, + "num_tokens": 143110827.0, + "step": 3679 + }, + { + "epoch": 0.4681338252130772, + "grad_norm": 2.422273635864258, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8937228918075562, + "num_tokens": 143153040.0, + "step": 3680 + }, + { + "epoch": 0.46826103549166775, + "grad_norm": 2.7286007404327393, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.866692841053009, + "num_tokens": 143193856.0, + "step": 3681 + }, + { + "epoch": 0.4683882457702582, + "grad_norm": 2.5300049781799316, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.883507490158081, + "num_tokens": 143233091.0, + "step": 3682 + }, + { + "epoch": 0.46851545604884876, + "grad_norm": 2.614598274230957, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8839250802993774, + "num_tokens": 143270727.0, + "step": 3683 + }, + { + "epoch": 0.4686426663274393, + "grad_norm": 2.6858017444610596, + "learning_rate": 1e-06, + "loss": 0.4068, + "mean_token_accuracy": 0.8572098016738892, + "num_tokens": 143312539.0, + "step": 3684 + }, + { + "epoch": 0.46876987660602976, + "grad_norm": 2.4863932132720947, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8844574689865112, + "num_tokens": 143354071.0, + "step": 3685 + }, + { + "epoch": 0.4688970868846203, + "grad_norm": 2.579373359680176, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8790583610534668, + "num_tokens": 143393759.0, + "step": 3686 + }, + { + "epoch": 0.4690242971632108, + "grad_norm": 2.7285468578338623, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8683484792709351, + "num_tokens": 143430501.0, + "step": 3687 + }, + { + "epoch": 0.4691515074418013, + "grad_norm": 2.8065762519836426, + "learning_rate": 1e-06, + "loss": 0.3971, + "mean_token_accuracy": 0.8626511693000793, + "num_tokens": 143465498.0, + "step": 3688 + }, + { + "epoch": 0.4692787177203918, + "grad_norm": 2.530416488647461, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8752808570861816, + "num_tokens": 143506869.0, + "step": 3689 + }, + { + "epoch": 0.46940592799898234, + "grad_norm": 3.201627731323242, + "learning_rate": 1e-06, + "loss": 0.3902, + "mean_token_accuracy": 0.8666157722473145, + "num_tokens": 143538273.0, + "step": 3690 + }, + { + "epoch": 0.4695331382775728, + "grad_norm": 2.601249933242798, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8899455070495605, + "num_tokens": 143576626.0, + "step": 3691 + }, + { + "epoch": 0.46966034855616334, + "grad_norm": 2.86841082572937, + "learning_rate": 1e-06, + "loss": 0.4042, + "mean_token_accuracy": 0.860262930393219, + "num_tokens": 143615260.0, + "step": 3692 + }, + { + "epoch": 0.46978755883475387, + "grad_norm": 2.6259210109710693, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8822733163833618, + "num_tokens": 143651476.0, + "step": 3693 + }, + { + "epoch": 0.46991476911334434, + "grad_norm": 2.363283634185791, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8785213232040405, + "num_tokens": 143699241.0, + "step": 3694 + }, + { + "epoch": 0.47004197939193487, + "grad_norm": 2.5261924266815186, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8801290988922119, + "num_tokens": 143742059.0, + "step": 3695 + }, + { + "epoch": 0.4701691896705254, + "grad_norm": 2.7510290145874023, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8813507556915283, + "num_tokens": 143776212.0, + "step": 3696 + }, + { + "epoch": 0.4702963999491159, + "grad_norm": 2.7000622749328613, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8829227685928345, + "num_tokens": 143811499.0, + "step": 3697 + }, + { + "epoch": 0.4704236102277064, + "grad_norm": 2.5868921279907227, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8842145204544067, + "num_tokens": 143850904.0, + "step": 3698 + }, + { + "epoch": 0.47055082050629693, + "grad_norm": 2.7744405269622803, + "learning_rate": 1e-06, + "loss": 0.3975, + "mean_token_accuracy": 0.8646166324615479, + "num_tokens": 143893255.0, + "step": 3699 + }, + { + "epoch": 0.4706780307848874, + "grad_norm": 2.6902382373809814, + "learning_rate": 1e-06, + "loss": 0.4033, + "mean_token_accuracy": 0.8607321381568909, + "num_tokens": 143931951.0, + "step": 3700 + }, + { + "epoch": 0.47080524106347793, + "grad_norm": 2.6246023178100586, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8896700143814087, + "num_tokens": 143969810.0, + "step": 3701 + }, + { + "epoch": 0.47093245134206846, + "grad_norm": 2.708526849746704, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.878071665763855, + "num_tokens": 144009435.0, + "step": 3702 + }, + { + "epoch": 0.47105966162065893, + "grad_norm": 2.436817169189453, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8711670637130737, + "num_tokens": 144054106.0, + "step": 3703 + }, + { + "epoch": 0.47118687189924946, + "grad_norm": 2.5055596828460693, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8786863088607788, + "num_tokens": 144095290.0, + "step": 3704 + }, + { + "epoch": 0.47131408217784, + "grad_norm": 2.9076004028320312, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8684862852096558, + "num_tokens": 144129553.0, + "step": 3705 + }, + { + "epoch": 0.47144129245643046, + "grad_norm": 2.841764450073242, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8798201084136963, + "num_tokens": 144163857.0, + "step": 3706 + }, + { + "epoch": 0.471568502735021, + "grad_norm": 2.711385726928711, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8735637664794922, + "num_tokens": 144202008.0, + "step": 3707 + }, + { + "epoch": 0.4716957130136115, + "grad_norm": 2.6906323432922363, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8821666240692139, + "num_tokens": 144237345.0, + "step": 3708 + }, + { + "epoch": 0.471822923292202, + "grad_norm": 2.7266438007354736, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8799976706504822, + "num_tokens": 144272039.0, + "step": 3709 + }, + { + "epoch": 0.4719501335707925, + "grad_norm": 2.394383192062378, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8761058449745178, + "num_tokens": 144316814.0, + "step": 3710 + }, + { + "epoch": 0.47207734384938305, + "grad_norm": 2.6369969844818115, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8814687728881836, + "num_tokens": 144355409.0, + "step": 3711 + }, + { + "epoch": 0.4722045541279735, + "grad_norm": 2.910339117050171, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8690156936645508, + "num_tokens": 144391127.0, + "step": 3712 + }, + { + "epoch": 0.47233176440656405, + "grad_norm": 2.5944879055023193, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8695544004440308, + "num_tokens": 144435289.0, + "step": 3713 + }, + { + "epoch": 0.4724589746851546, + "grad_norm": 2.8266139030456543, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8699826002120972, + "num_tokens": 144473127.0, + "step": 3714 + }, + { + "epoch": 0.47258618496374505, + "grad_norm": 2.8405063152313232, + "learning_rate": 1e-06, + "loss": 0.3892, + "mean_token_accuracy": 0.8644523620605469, + "num_tokens": 144507141.0, + "step": 3715 + }, + { + "epoch": 0.4727133952423356, + "grad_norm": 2.7543632984161377, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8704642057418823, + "num_tokens": 144543163.0, + "step": 3716 + }, + { + "epoch": 0.4728406055209261, + "grad_norm": 2.565838575363159, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8824754357337952, + "num_tokens": 144585257.0, + "step": 3717 + }, + { + "epoch": 0.4729678157995166, + "grad_norm": 2.8393263816833496, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8892592191696167, + "num_tokens": 144625933.0, + "step": 3718 + }, + { + "epoch": 0.4730950260781071, + "grad_norm": 2.6753785610198975, + "learning_rate": 1e-06, + "loss": 0.3815, + "mean_token_accuracy": 0.8672260642051697, + "num_tokens": 144665096.0, + "step": 3719 + }, + { + "epoch": 0.47322223635669763, + "grad_norm": 2.76576828956604, + "learning_rate": 1e-06, + "loss": 0.4141, + "mean_token_accuracy": 0.8565329313278198, + "num_tokens": 144702674.0, + "step": 3720 + }, + { + "epoch": 0.4733494466352881, + "grad_norm": 2.802665948867798, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8696814775466919, + "num_tokens": 144738453.0, + "step": 3721 + }, + { + "epoch": 0.47347665691387864, + "grad_norm": 2.674945592880249, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8701866865158081, + "num_tokens": 144780423.0, + "step": 3722 + }, + { + "epoch": 0.47360386719246916, + "grad_norm": 2.8355977535247803, + "learning_rate": 1e-06, + "loss": 0.3877, + "mean_token_accuracy": 0.8636517524719238, + "num_tokens": 144816696.0, + "step": 3723 + }, + { + "epoch": 0.47373107747105964, + "grad_norm": 2.275456190109253, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8839219212532043, + "num_tokens": 144864828.0, + "step": 3724 + }, + { + "epoch": 0.47385828774965016, + "grad_norm": 2.7312309741973877, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8831655979156494, + "num_tokens": 144900018.0, + "step": 3725 + }, + { + "epoch": 0.4739854980282407, + "grad_norm": 2.636725425720215, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8750160932540894, + "num_tokens": 144941958.0, + "step": 3726 + }, + { + "epoch": 0.47411270830683117, + "grad_norm": 2.808384418487549, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8730570077896118, + "num_tokens": 144976662.0, + "step": 3727 + }, + { + "epoch": 0.4742399185854217, + "grad_norm": 2.778362274169922, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8636295199394226, + "num_tokens": 145016978.0, + "step": 3728 + }, + { + "epoch": 0.4743671288640122, + "grad_norm": 2.685359477996826, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8730052709579468, + "num_tokens": 145055156.0, + "step": 3729 + }, + { + "epoch": 0.4744943391426027, + "grad_norm": 2.726696252822876, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8798090219497681, + "num_tokens": 145088229.0, + "step": 3730 + }, + { + "epoch": 0.4746215494211932, + "grad_norm": 2.53407883644104, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8811295032501221, + "num_tokens": 145131808.0, + "step": 3731 + }, + { + "epoch": 0.47474875969978375, + "grad_norm": 2.4613418579101562, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8702842593193054, + "num_tokens": 145176219.0, + "step": 3732 + }, + { + "epoch": 0.4748759699783743, + "grad_norm": 2.8458266258239746, + "learning_rate": 1e-06, + "loss": 0.4224, + "mean_token_accuracy": 0.8577020764350891, + "num_tokens": 145214567.0, + "step": 3733 + }, + { + "epoch": 0.47500318025696475, + "grad_norm": 2.962217330932617, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.8690077066421509, + "num_tokens": 145249075.0, + "step": 3734 + }, + { + "epoch": 0.4751303905355553, + "grad_norm": 2.802926778793335, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8743458986282349, + "num_tokens": 145285331.0, + "step": 3735 + }, + { + "epoch": 0.4752576008141458, + "grad_norm": 2.908797025680542, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8859010934829712, + "num_tokens": 145316462.0, + "step": 3736 + }, + { + "epoch": 0.4753848110927363, + "grad_norm": 2.5245201587677, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8863208889961243, + "num_tokens": 145358273.0, + "step": 3737 + }, + { + "epoch": 0.4755120213713268, + "grad_norm": 2.6868298053741455, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8693056106567383, + "num_tokens": 145396869.0, + "step": 3738 + }, + { + "epoch": 0.47563923164991734, + "grad_norm": 2.813488721847534, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8730651140213013, + "num_tokens": 145434209.0, + "step": 3739 + }, + { + "epoch": 0.4757664419285078, + "grad_norm": 2.805943012237549, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8669923543930054, + "num_tokens": 145472233.0, + "step": 3740 + }, + { + "epoch": 0.47589365220709834, + "grad_norm": 2.709763288497925, + "learning_rate": 1e-06, + "loss": 0.4199, + "mean_token_accuracy": 0.8549802303314209, + "num_tokens": 145517365.0, + "step": 3741 + }, + { + "epoch": 0.47602086248568887, + "grad_norm": 2.724517822265625, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.872173547744751, + "num_tokens": 145555772.0, + "step": 3742 + }, + { + "epoch": 0.47614807276427934, + "grad_norm": 2.6436824798583984, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8757086992263794, + "num_tokens": 145594656.0, + "step": 3743 + }, + { + "epoch": 0.47627528304286987, + "grad_norm": 2.772963285446167, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8727039098739624, + "num_tokens": 145635649.0, + "step": 3744 + }, + { + "epoch": 0.4764024933214604, + "grad_norm": 2.6420798301696777, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8828808665275574, + "num_tokens": 145674646.0, + "step": 3745 + }, + { + "epoch": 0.47652970360005087, + "grad_norm": 2.5430941581726074, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8834478855133057, + "num_tokens": 145715111.0, + "step": 3746 + }, + { + "epoch": 0.4766569138786414, + "grad_norm": 2.649021625518799, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.884777843952179, + "num_tokens": 145749182.0, + "step": 3747 + }, + { + "epoch": 0.4767841241572319, + "grad_norm": 2.4167754650115967, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8681840896606445, + "num_tokens": 145797301.0, + "step": 3748 + }, + { + "epoch": 0.4769113344358224, + "grad_norm": 2.7170825004577637, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8651577830314636, + "num_tokens": 145837158.0, + "step": 3749 + }, + { + "epoch": 0.4770385447144129, + "grad_norm": 2.415167808532715, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8764272928237915, + "num_tokens": 145881876.0, + "step": 3750 + }, + { + "epoch": 0.47716575499300345, + "grad_norm": 2.6572699546813965, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8895158767700195, + "num_tokens": 145915424.0, + "step": 3751 + }, + { + "epoch": 0.4772929652715939, + "grad_norm": 2.7900166511535645, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8826154470443726, + "num_tokens": 145948156.0, + "step": 3752 + }, + { + "epoch": 0.47742017555018446, + "grad_norm": 2.50726318359375, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8802480101585388, + "num_tokens": 145990755.0, + "step": 3753 + }, + { + "epoch": 0.477547385828775, + "grad_norm": 2.826627492904663, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8778113126754761, + "num_tokens": 146029488.0, + "step": 3754 + }, + { + "epoch": 0.47767459610736546, + "grad_norm": 2.7919106483459473, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8712184429168701, + "num_tokens": 146067042.0, + "step": 3755 + }, + { + "epoch": 0.477801806385956, + "grad_norm": 2.4937760829925537, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8896243572235107, + "num_tokens": 146106428.0, + "step": 3756 + }, + { + "epoch": 0.4779290166645465, + "grad_norm": 2.72019100189209, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8734362125396729, + "num_tokens": 146144720.0, + "step": 3757 + }, + { + "epoch": 0.478056226943137, + "grad_norm": 2.702383041381836, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8693876266479492, + "num_tokens": 146181393.0, + "step": 3758 + }, + { + "epoch": 0.4781834372217275, + "grad_norm": 2.673703670501709, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.873681902885437, + "num_tokens": 146222339.0, + "step": 3759 + }, + { + "epoch": 0.47831064750031804, + "grad_norm": 2.5425703525543213, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8806768655776978, + "num_tokens": 146262306.0, + "step": 3760 + }, + { + "epoch": 0.4784378577789085, + "grad_norm": 2.7095789909362793, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8724464178085327, + "num_tokens": 146300434.0, + "step": 3761 + }, + { + "epoch": 0.47856506805749904, + "grad_norm": 3.043466806411743, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.874636709690094, + "num_tokens": 146330591.0, + "step": 3762 + }, + { + "epoch": 0.47869227833608957, + "grad_norm": 2.8764073848724365, + "learning_rate": 1e-06, + "loss": 0.4092, + "mean_token_accuracy": 0.8581998944282532, + "num_tokens": 146372206.0, + "step": 3763 + }, + { + "epoch": 0.47881948861468004, + "grad_norm": 2.8100759983062744, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8827064037322998, + "num_tokens": 146407006.0, + "step": 3764 + }, + { + "epoch": 0.4789466988932706, + "grad_norm": 2.881382465362549, + "learning_rate": 1e-06, + "loss": 0.4002, + "mean_token_accuracy": 0.8653564453125, + "num_tokens": 146444100.0, + "step": 3765 + }, + { + "epoch": 0.4790739091718611, + "grad_norm": 2.6131412982940674, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8865340948104858, + "num_tokens": 146481000.0, + "step": 3766 + }, + { + "epoch": 0.4792011194504516, + "grad_norm": 2.7032651901245117, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8760936260223389, + "num_tokens": 146520407.0, + "step": 3767 + }, + { + "epoch": 0.4793283297290421, + "grad_norm": 2.664315700531006, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8773079514503479, + "num_tokens": 146555873.0, + "step": 3768 + }, + { + "epoch": 0.47945554000763263, + "grad_norm": 2.6092724800109863, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8735507726669312, + "num_tokens": 146593545.0, + "step": 3769 + }, + { + "epoch": 0.4795827502862231, + "grad_norm": 2.786677837371826, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8660601377487183, + "num_tokens": 146633278.0, + "step": 3770 + }, + { + "epoch": 0.47970996056481363, + "grad_norm": 2.766630172729492, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8734298944473267, + "num_tokens": 146668259.0, + "step": 3771 + }, + { + "epoch": 0.47983717084340416, + "grad_norm": 2.3408243656158447, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8784024119377136, + "num_tokens": 146716148.0, + "step": 3772 + }, + { + "epoch": 0.47996438112199463, + "grad_norm": 2.5780162811279297, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8878916501998901, + "num_tokens": 146757031.0, + "step": 3773 + }, + { + "epoch": 0.48009159140058516, + "grad_norm": 2.7654268741607666, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8848646879196167, + "num_tokens": 146790965.0, + "step": 3774 + }, + { + "epoch": 0.4802188016791757, + "grad_norm": 2.51664400100708, + "learning_rate": 1e-06, + "loss": 0.3976, + "mean_token_accuracy": 0.8634111881256104, + "num_tokens": 146837089.0, + "step": 3775 + }, + { + "epoch": 0.48034601195776616, + "grad_norm": 2.7587063312530518, + "learning_rate": 1e-06, + "loss": 0.404, + "mean_token_accuracy": 0.8605777621269226, + "num_tokens": 146876849.0, + "step": 3776 + }, + { + "epoch": 0.4804732222363567, + "grad_norm": 2.563115119934082, + "learning_rate": 1e-06, + "loss": 0.4262, + "mean_token_accuracy": 0.8570117950439453, + "num_tokens": 146924942.0, + "step": 3777 + }, + { + "epoch": 0.4806004325149472, + "grad_norm": 2.5551187992095947, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8680423498153687, + "num_tokens": 146968401.0, + "step": 3778 + }, + { + "epoch": 0.4807276427935377, + "grad_norm": 2.693220376968384, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8765220046043396, + "num_tokens": 147005080.0, + "step": 3779 + }, + { + "epoch": 0.4808548530721282, + "grad_norm": 2.6322522163391113, + "learning_rate": 1e-06, + "loss": 0.4058, + "mean_token_accuracy": 0.859545111656189, + "num_tokens": 147050129.0, + "step": 3780 + }, + { + "epoch": 0.48098206335071875, + "grad_norm": 2.791006565093994, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8813814520835876, + "num_tokens": 147086290.0, + "step": 3781 + }, + { + "epoch": 0.4811092736293093, + "grad_norm": 2.694582939147949, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8830367922782898, + "num_tokens": 147121946.0, + "step": 3782 + }, + { + "epoch": 0.48123648390789975, + "grad_norm": 2.8578145503997803, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8588429689407349, + "num_tokens": 147158718.0, + "step": 3783 + }, + { + "epoch": 0.4813636941864903, + "grad_norm": 2.5732364654541016, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8781522512435913, + "num_tokens": 147199269.0, + "step": 3784 + }, + { + "epoch": 0.4814909044650808, + "grad_norm": 2.7741799354553223, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8739910125732422, + "num_tokens": 147237140.0, + "step": 3785 + }, + { + "epoch": 0.4816181147436713, + "grad_norm": 2.672088623046875, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8714586496353149, + "num_tokens": 147275448.0, + "step": 3786 + }, + { + "epoch": 0.4817453250222618, + "grad_norm": 2.875502347946167, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8764981031417847, + "num_tokens": 147310787.0, + "step": 3787 + }, + { + "epoch": 0.48187253530085233, + "grad_norm": 2.839083671569824, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8687810301780701, + "num_tokens": 147344438.0, + "step": 3788 + }, + { + "epoch": 0.4819997455794428, + "grad_norm": 2.5450687408447266, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8762381076812744, + "num_tokens": 147383076.0, + "step": 3789 + }, + { + "epoch": 0.48212695585803333, + "grad_norm": 2.6427948474884033, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8891181945800781, + "num_tokens": 147416961.0, + "step": 3790 + }, + { + "epoch": 0.48225416613662386, + "grad_norm": 2.751535654067993, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8734332919120789, + "num_tokens": 147453383.0, + "step": 3791 + }, + { + "epoch": 0.48238137641521434, + "grad_norm": 2.5871217250823975, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8815773725509644, + "num_tokens": 147489397.0, + "step": 3792 + }, + { + "epoch": 0.48250858669380486, + "grad_norm": 2.7616889476776123, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8746404647827148, + "num_tokens": 147524937.0, + "step": 3793 + }, + { + "epoch": 0.4826357969723954, + "grad_norm": 2.843444347381592, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8742122650146484, + "num_tokens": 147558728.0, + "step": 3794 + }, + { + "epoch": 0.48276300725098586, + "grad_norm": 2.6187946796417236, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8850583434104919, + "num_tokens": 147595923.0, + "step": 3795 + }, + { + "epoch": 0.4828902175295764, + "grad_norm": 2.6622767448425293, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.876579999923706, + "num_tokens": 147631342.0, + "step": 3796 + }, + { + "epoch": 0.4830174278081669, + "grad_norm": 2.873302698135376, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8701941967010498, + "num_tokens": 147664485.0, + "step": 3797 + }, + { + "epoch": 0.4831446380867574, + "grad_norm": 2.4569199085235596, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.887220025062561, + "num_tokens": 147704112.0, + "step": 3798 + }, + { + "epoch": 0.4832718483653479, + "grad_norm": 2.5857913494110107, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8825802803039551, + "num_tokens": 147743217.0, + "step": 3799 + }, + { + "epoch": 0.48339905864393845, + "grad_norm": 2.7572529315948486, + "learning_rate": 1e-06, + "loss": 0.404, + "mean_token_accuracy": 0.8650330901145935, + "num_tokens": 147781000.0, + "step": 3800 + }, + { + "epoch": 0.4835262689225289, + "grad_norm": 2.7318389415740967, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8695787787437439, + "num_tokens": 147817786.0, + "step": 3801 + }, + { + "epoch": 0.48365347920111945, + "grad_norm": 2.903642177581787, + "learning_rate": 1e-06, + "loss": 0.3981, + "mean_token_accuracy": 0.8630099296569824, + "num_tokens": 147853976.0, + "step": 3802 + }, + { + "epoch": 0.48378068947971, + "grad_norm": 2.776542901992798, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8727399110794067, + "num_tokens": 147889425.0, + "step": 3803 + }, + { + "epoch": 0.48390789975830045, + "grad_norm": 2.383514881134033, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8811021447181702, + "num_tokens": 147935456.0, + "step": 3804 + }, + { + "epoch": 0.484035110036891, + "grad_norm": 2.3731303215026855, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8913042545318604, + "num_tokens": 147980379.0, + "step": 3805 + }, + { + "epoch": 0.4841623203154815, + "grad_norm": 2.450352907180786, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8874200582504272, + "num_tokens": 148023494.0, + "step": 3806 + }, + { + "epoch": 0.484289530594072, + "grad_norm": 2.711162805557251, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8749563097953796, + "num_tokens": 148061548.0, + "step": 3807 + }, + { + "epoch": 0.4844167408726625, + "grad_norm": 2.7535018920898438, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8778209686279297, + "num_tokens": 148097544.0, + "step": 3808 + }, + { + "epoch": 0.48454395115125304, + "grad_norm": 2.5375101566314697, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8725384473800659, + "num_tokens": 148139734.0, + "step": 3809 + }, + { + "epoch": 0.4846711614298435, + "grad_norm": 2.6119000911712646, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.8632692694664001, + "num_tokens": 148182390.0, + "step": 3810 + }, + { + "epoch": 0.48479837170843404, + "grad_norm": 2.6825437545776367, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8715285062789917, + "num_tokens": 148221688.0, + "step": 3811 + }, + { + "epoch": 0.48492558198702457, + "grad_norm": 2.8355565071105957, + "learning_rate": 1e-06, + "loss": 0.4174, + "mean_token_accuracy": 0.8594009876251221, + "num_tokens": 148263419.0, + "step": 3812 + }, + { + "epoch": 0.48505279226561504, + "grad_norm": 2.9697906970977783, + "learning_rate": 1e-06, + "loss": 0.3963, + "mean_token_accuracy": 0.8652724623680115, + "num_tokens": 148300644.0, + "step": 3813 + }, + { + "epoch": 0.48518000254420557, + "grad_norm": 2.771721363067627, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.8608726859092712, + "num_tokens": 148337562.0, + "step": 3814 + }, + { + "epoch": 0.4853072128227961, + "grad_norm": 2.8433547019958496, + "learning_rate": 1e-06, + "loss": 0.4123, + "mean_token_accuracy": 0.8575426340103149, + "num_tokens": 148374663.0, + "step": 3815 + }, + { + "epoch": 0.48543442310138657, + "grad_norm": 3.0086026191711426, + "learning_rate": 1e-06, + "loss": 0.384, + "mean_token_accuracy": 0.8688364028930664, + "num_tokens": 148407980.0, + "step": 3816 + }, + { + "epoch": 0.4855616333799771, + "grad_norm": 2.6566720008850098, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8776885867118835, + "num_tokens": 148446495.0, + "step": 3817 + }, + { + "epoch": 0.4856888436585676, + "grad_norm": 2.9864723682403564, + "learning_rate": 1e-06, + "loss": 0.4075, + "mean_token_accuracy": 0.8590255379676819, + "num_tokens": 148481331.0, + "step": 3818 + }, + { + "epoch": 0.4858160539371581, + "grad_norm": 2.632770538330078, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8779453635215759, + "num_tokens": 148519471.0, + "step": 3819 + }, + { + "epoch": 0.4859432642157486, + "grad_norm": 2.8287930488586426, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8688410520553589, + "num_tokens": 148554309.0, + "step": 3820 + }, + { + "epoch": 0.48607047449433916, + "grad_norm": 2.346372127532959, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.888529360294342, + "num_tokens": 148601327.0, + "step": 3821 + }, + { + "epoch": 0.48619768477292963, + "grad_norm": 2.44820499420166, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8901669979095459, + "num_tokens": 148647063.0, + "step": 3822 + }, + { + "epoch": 0.48632489505152016, + "grad_norm": 2.6989920139312744, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8794673681259155, + "num_tokens": 148681886.0, + "step": 3823 + }, + { + "epoch": 0.4864521053301107, + "grad_norm": 2.4490127563476562, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8911229968070984, + "num_tokens": 148723107.0, + "step": 3824 + }, + { + "epoch": 0.48657931560870116, + "grad_norm": 2.466479778289795, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8772306442260742, + "num_tokens": 148766400.0, + "step": 3825 + }, + { + "epoch": 0.4867065258872917, + "grad_norm": 2.657748222351074, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8617997169494629, + "num_tokens": 148808525.0, + "step": 3826 + }, + { + "epoch": 0.4868337361658822, + "grad_norm": 2.9342637062072754, + "learning_rate": 1e-06, + "loss": 0.384, + "mean_token_accuracy": 0.8701878786087036, + "num_tokens": 148845907.0, + "step": 3827 + }, + { + "epoch": 0.4869609464444727, + "grad_norm": 2.4526829719543457, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8861835598945618, + "num_tokens": 148884851.0, + "step": 3828 + }, + { + "epoch": 0.4870881567230632, + "grad_norm": 2.3953890800476074, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8827636241912842, + "num_tokens": 148930190.0, + "step": 3829 + }, + { + "epoch": 0.48721536700165374, + "grad_norm": 2.9566807746887207, + "learning_rate": 1e-06, + "loss": 0.4112, + "mean_token_accuracy": 0.863740086555481, + "num_tokens": 148964815.0, + "step": 3830 + }, + { + "epoch": 0.48734257728024427, + "grad_norm": 2.7661430835723877, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8720834851264954, + "num_tokens": 148999217.0, + "step": 3831 + }, + { + "epoch": 0.48746978755883474, + "grad_norm": 2.5627739429473877, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8825887441635132, + "num_tokens": 149037482.0, + "step": 3832 + }, + { + "epoch": 0.48759699783742527, + "grad_norm": 2.5398247241973877, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8720726370811462, + "num_tokens": 149079339.0, + "step": 3833 + }, + { + "epoch": 0.4877242081160158, + "grad_norm": 2.513655662536621, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8801878690719604, + "num_tokens": 149120126.0, + "step": 3834 + }, + { + "epoch": 0.4878514183946063, + "grad_norm": 2.9483604431152344, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8650891780853271, + "num_tokens": 149153869.0, + "step": 3835 + }, + { + "epoch": 0.4879786286731968, + "grad_norm": 2.632112503051758, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8713082671165466, + "num_tokens": 149191633.0, + "step": 3836 + }, + { + "epoch": 0.48810583895178733, + "grad_norm": 2.576395273208618, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8823356032371521, + "num_tokens": 149231438.0, + "step": 3837 + }, + { + "epoch": 0.4882330492303778, + "grad_norm": 2.705889940261841, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8744024038314819, + "num_tokens": 149269552.0, + "step": 3838 + }, + { + "epoch": 0.48836025950896833, + "grad_norm": 2.334271192550659, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8858395218849182, + "num_tokens": 149312283.0, + "step": 3839 + }, + { + "epoch": 0.48848746978755886, + "grad_norm": 2.7145955562591553, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.86939537525177, + "num_tokens": 149350832.0, + "step": 3840 + }, + { + "epoch": 0.48861468006614933, + "grad_norm": 2.8224782943725586, + "learning_rate": 1e-06, + "loss": 0.4141, + "mean_token_accuracy": 0.8591995239257812, + "num_tokens": 149391060.0, + "step": 3841 + }, + { + "epoch": 0.48874189034473986, + "grad_norm": 2.9726169109344482, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8805552124977112, + "num_tokens": 149423415.0, + "step": 3842 + }, + { + "epoch": 0.4888691006233304, + "grad_norm": 2.6834356784820557, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8709880709648132, + "num_tokens": 149459827.0, + "step": 3843 + }, + { + "epoch": 0.48899631090192086, + "grad_norm": 2.8800652027130127, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8685901165008545, + "num_tokens": 149496278.0, + "step": 3844 + }, + { + "epoch": 0.4891235211805114, + "grad_norm": 2.549365758895874, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8661053776741028, + "num_tokens": 149540964.0, + "step": 3845 + }, + { + "epoch": 0.4892507314591019, + "grad_norm": 2.5152111053466797, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8871442675590515, + "num_tokens": 149584795.0, + "step": 3846 + }, + { + "epoch": 0.4893779417376924, + "grad_norm": 2.7696616649627686, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8748087286949158, + "num_tokens": 149621777.0, + "step": 3847 + }, + { + "epoch": 0.4895051520162829, + "grad_norm": 2.786530017852783, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8793914318084717, + "num_tokens": 149656150.0, + "step": 3848 + }, + { + "epoch": 0.48963236229487345, + "grad_norm": 2.637563705444336, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8792175054550171, + "num_tokens": 149693454.0, + "step": 3849 + }, + { + "epoch": 0.4897595725734639, + "grad_norm": 2.6401336193084717, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.869990885257721, + "num_tokens": 149735807.0, + "step": 3850 + }, + { + "epoch": 0.48988678285205445, + "grad_norm": 2.7418880462646484, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8700069189071655, + "num_tokens": 149777030.0, + "step": 3851 + }, + { + "epoch": 0.490013993130645, + "grad_norm": 2.4823758602142334, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8837558031082153, + "num_tokens": 149816666.0, + "step": 3852 + }, + { + "epoch": 0.49014120340923545, + "grad_norm": 2.726243495941162, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8698121309280396, + "num_tokens": 149853270.0, + "step": 3853 + }, + { + "epoch": 0.490268413687826, + "grad_norm": 2.7026212215423584, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8713831901550293, + "num_tokens": 149889139.0, + "step": 3854 + }, + { + "epoch": 0.4903956239664165, + "grad_norm": 2.7061431407928467, + "learning_rate": 1e-06, + "loss": 0.4016, + "mean_token_accuracy": 0.8586747646331787, + "num_tokens": 149926941.0, + "step": 3855 + }, + { + "epoch": 0.490522834245007, + "grad_norm": 2.7662248611450195, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8817095756530762, + "num_tokens": 149961177.0, + "step": 3856 + }, + { + "epoch": 0.4906500445235975, + "grad_norm": 2.7867918014526367, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8707147836685181, + "num_tokens": 149995184.0, + "step": 3857 + }, + { + "epoch": 0.49077725480218803, + "grad_norm": 2.803299903869629, + "learning_rate": 1e-06, + "loss": 0.4334, + "mean_token_accuracy": 0.8509706258773804, + "num_tokens": 150037194.0, + "step": 3858 + }, + { + "epoch": 0.4909044650807785, + "grad_norm": 2.5726327896118164, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8783963322639465, + "num_tokens": 150079137.0, + "step": 3859 + }, + { + "epoch": 0.49103167535936904, + "grad_norm": 2.5860674381256104, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8722730875015259, + "num_tokens": 150123043.0, + "step": 3860 + }, + { + "epoch": 0.49115888563795956, + "grad_norm": 2.8210246562957764, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8888376951217651, + "num_tokens": 150158555.0, + "step": 3861 + }, + { + "epoch": 0.49128609591655004, + "grad_norm": 2.5192160606384277, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8815777897834778, + "num_tokens": 150199229.0, + "step": 3862 + }, + { + "epoch": 0.49141330619514056, + "grad_norm": 2.811727285385132, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8652794361114502, + "num_tokens": 150244693.0, + "step": 3863 + }, + { + "epoch": 0.4915405164737311, + "grad_norm": 2.6234631538391113, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8768686056137085, + "num_tokens": 150282681.0, + "step": 3864 + }, + { + "epoch": 0.49166772675232157, + "grad_norm": 2.4133331775665283, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8777286410331726, + "num_tokens": 150326648.0, + "step": 3865 + }, + { + "epoch": 0.4917949370309121, + "grad_norm": 2.5522425174713135, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8736392259597778, + "num_tokens": 150368766.0, + "step": 3866 + }, + { + "epoch": 0.4919221473095026, + "grad_norm": 2.930082082748413, + "learning_rate": 1e-06, + "loss": 0.4217, + "mean_token_accuracy": 0.8552370071411133, + "num_tokens": 150408412.0, + "step": 3867 + }, + { + "epoch": 0.4920493575880931, + "grad_norm": 2.7990915775299072, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8696088790893555, + "num_tokens": 150444773.0, + "step": 3868 + }, + { + "epoch": 0.4921765678666836, + "grad_norm": 2.503168821334839, + "learning_rate": 1e-06, + "loss": 0.3815, + "mean_token_accuracy": 0.8703272342681885, + "num_tokens": 150488867.0, + "step": 3869 + }, + { + "epoch": 0.49230377814527415, + "grad_norm": 2.759781837463379, + "learning_rate": 1e-06, + "loss": 0.4119, + "mean_token_accuracy": 0.8603881597518921, + "num_tokens": 150531581.0, + "step": 3870 + }, + { + "epoch": 0.4924309884238646, + "grad_norm": 2.8583710193634033, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8741049766540527, + "num_tokens": 150564019.0, + "step": 3871 + }, + { + "epoch": 0.49255819870245515, + "grad_norm": 2.7845377922058105, + "learning_rate": 1e-06, + "loss": 0.4457, + "mean_token_accuracy": 0.8497000932693481, + "num_tokens": 150603774.0, + "step": 3872 + }, + { + "epoch": 0.4926854089810457, + "grad_norm": 2.6760551929473877, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8767398595809937, + "num_tokens": 150638901.0, + "step": 3873 + }, + { + "epoch": 0.49281261925963615, + "grad_norm": 2.5991978645324707, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8681219816207886, + "num_tokens": 150678683.0, + "step": 3874 + }, + { + "epoch": 0.4929398295382267, + "grad_norm": 2.574007272720337, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8713439106941223, + "num_tokens": 150719539.0, + "step": 3875 + }, + { + "epoch": 0.4930670398168172, + "grad_norm": 2.6462745666503906, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8811582326889038, + "num_tokens": 150762639.0, + "step": 3876 + }, + { + "epoch": 0.4931942500954077, + "grad_norm": 2.917947292327881, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.8677439093589783, + "num_tokens": 150795650.0, + "step": 3877 + }, + { + "epoch": 0.4933214603739982, + "grad_norm": 2.751323699951172, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8690327405929565, + "num_tokens": 150832339.0, + "step": 3878 + }, + { + "epoch": 0.49344867065258874, + "grad_norm": 2.630525588989258, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8723844885826111, + "num_tokens": 150868510.0, + "step": 3879 + }, + { + "epoch": 0.4935758809311792, + "grad_norm": 3.071676015853882, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8679918050765991, + "num_tokens": 150899803.0, + "step": 3880 + }, + { + "epoch": 0.49370309120976974, + "grad_norm": 2.7117929458618164, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8826829195022583, + "num_tokens": 150935182.0, + "step": 3881 + }, + { + "epoch": 0.49383030148836027, + "grad_norm": 2.674531936645508, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8760994672775269, + "num_tokens": 150976097.0, + "step": 3882 + }, + { + "epoch": 0.4939575117669508, + "grad_norm": 2.8956124782562256, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8659880757331848, + "num_tokens": 151011812.0, + "step": 3883 + }, + { + "epoch": 0.49408472204554127, + "grad_norm": 2.732815742492676, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8682544827461243, + "num_tokens": 151047652.0, + "step": 3884 + }, + { + "epoch": 0.4942119323241318, + "grad_norm": 2.6259841918945312, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.8664312362670898, + "num_tokens": 151089710.0, + "step": 3885 + }, + { + "epoch": 0.4943391426027223, + "grad_norm": 2.642303228378296, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8827220797538757, + "num_tokens": 151127642.0, + "step": 3886 + }, + { + "epoch": 0.4944663528813128, + "grad_norm": 2.677061080932617, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8777413368225098, + "num_tokens": 151166886.0, + "step": 3887 + }, + { + "epoch": 0.4945935631599033, + "grad_norm": 2.7582905292510986, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8748739957809448, + "num_tokens": 151203831.0, + "step": 3888 + }, + { + "epoch": 0.49472077343849385, + "grad_norm": 2.5475027561187744, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8785654902458191, + "num_tokens": 151243190.0, + "step": 3889 + }, + { + "epoch": 0.4948479837170843, + "grad_norm": 2.6158220767974854, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8954162001609802, + "num_tokens": 151275289.0, + "step": 3890 + }, + { + "epoch": 0.49497519399567486, + "grad_norm": 2.573169708251953, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8777392506599426, + "num_tokens": 151320091.0, + "step": 3891 + }, + { + "epoch": 0.4951024042742654, + "grad_norm": 2.7230887413024902, + "learning_rate": 1e-06, + "loss": 0.4315, + "mean_token_accuracy": 0.8532875180244446, + "num_tokens": 151361398.0, + "step": 3892 + }, + { + "epoch": 0.49522961455285586, + "grad_norm": 2.9150893688201904, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8700434565544128, + "num_tokens": 151394528.0, + "step": 3893 + }, + { + "epoch": 0.4953568248314464, + "grad_norm": 2.7498035430908203, + "learning_rate": 1e-06, + "loss": 0.4054, + "mean_token_accuracy": 0.8634021282196045, + "num_tokens": 151434203.0, + "step": 3894 + }, + { + "epoch": 0.4954840351100369, + "grad_norm": 2.6668827533721924, + "learning_rate": 1e-06, + "loss": 0.3946, + "mean_token_accuracy": 0.8638472557067871, + "num_tokens": 151474727.0, + "step": 3895 + }, + { + "epoch": 0.4956112453886274, + "grad_norm": 2.6304843425750732, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8709688186645508, + "num_tokens": 151514995.0, + "step": 3896 + }, + { + "epoch": 0.4957384556672179, + "grad_norm": 2.6317148208618164, + "learning_rate": 1e-06, + "loss": 0.3939, + "mean_token_accuracy": 0.8637324571609497, + "num_tokens": 151558138.0, + "step": 3897 + }, + { + "epoch": 0.49586566594580844, + "grad_norm": 2.6844522953033447, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8725107908248901, + "num_tokens": 151600041.0, + "step": 3898 + }, + { + "epoch": 0.4959928762243989, + "grad_norm": 2.8257944583892822, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8678591251373291, + "num_tokens": 151636530.0, + "step": 3899 + }, + { + "epoch": 0.49612008650298944, + "grad_norm": 3.0201165676116943, + "learning_rate": 1e-06, + "loss": 0.3881, + "mean_token_accuracy": 0.8639272451400757, + "num_tokens": 151670646.0, + "step": 3900 + }, + { + "epoch": 0.49624729678157997, + "grad_norm": 2.5808122158050537, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8696207404136658, + "num_tokens": 151711582.0, + "step": 3901 + }, + { + "epoch": 0.49637450706017044, + "grad_norm": 2.7335426807403564, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8726460933685303, + "num_tokens": 151750652.0, + "step": 3902 + }, + { + "epoch": 0.496501717338761, + "grad_norm": 2.6974265575408936, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.868472695350647, + "num_tokens": 151789724.0, + "step": 3903 + }, + { + "epoch": 0.4966289276173515, + "grad_norm": 2.929313898086548, + "learning_rate": 1e-06, + "loss": 0.4441, + "mean_token_accuracy": 0.8505873680114746, + "num_tokens": 151829106.0, + "step": 3904 + }, + { + "epoch": 0.496756137895942, + "grad_norm": 2.6605353355407715, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8750549554824829, + "num_tokens": 151868324.0, + "step": 3905 + }, + { + "epoch": 0.4968833481745325, + "grad_norm": 2.8114047050476074, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8737029433250427, + "num_tokens": 151901958.0, + "step": 3906 + }, + { + "epoch": 0.49701055845312303, + "grad_norm": 2.879204750061035, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8801941871643066, + "num_tokens": 151933136.0, + "step": 3907 + }, + { + "epoch": 0.4971377687317135, + "grad_norm": 2.7235138416290283, + "learning_rate": 1e-06, + "loss": 0.4127, + "mean_token_accuracy": 0.8561923503875732, + "num_tokens": 151971314.0, + "step": 3908 + }, + { + "epoch": 0.49726497901030403, + "grad_norm": 2.6336140632629395, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8692588210105896, + "num_tokens": 152012351.0, + "step": 3909 + }, + { + "epoch": 0.49739218928889456, + "grad_norm": 2.8802602291107178, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8729134202003479, + "num_tokens": 152047252.0, + "step": 3910 + }, + { + "epoch": 0.49751939956748503, + "grad_norm": 2.696622133255005, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8685839772224426, + "num_tokens": 152090253.0, + "step": 3911 + }, + { + "epoch": 0.49764660984607556, + "grad_norm": 2.620960235595703, + "learning_rate": 1e-06, + "loss": 0.4021, + "mean_token_accuracy": 0.8618053197860718, + "num_tokens": 152133999.0, + "step": 3912 + }, + { + "epoch": 0.4977738201246661, + "grad_norm": 2.704357385635376, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8697818517684937, + "num_tokens": 152171144.0, + "step": 3913 + }, + { + "epoch": 0.49790103040325656, + "grad_norm": 3.0154125690460205, + "learning_rate": 1e-06, + "loss": 0.4167, + "mean_token_accuracy": 0.8596088290214539, + "num_tokens": 152206600.0, + "step": 3914 + }, + { + "epoch": 0.4980282406818471, + "grad_norm": 2.7061550617218018, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8804795742034912, + "num_tokens": 152243468.0, + "step": 3915 + }, + { + "epoch": 0.4981554509604376, + "grad_norm": 2.9126667976379395, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8864094018936157, + "num_tokens": 152277194.0, + "step": 3916 + }, + { + "epoch": 0.4982826612390281, + "grad_norm": 2.866015911102295, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8699617385864258, + "num_tokens": 152314561.0, + "step": 3917 + }, + { + "epoch": 0.4984098715176186, + "grad_norm": 2.73817777633667, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8658013343811035, + "num_tokens": 152351377.0, + "step": 3918 + }, + { + "epoch": 0.49853708179620915, + "grad_norm": 2.386669874191284, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8789463043212891, + "num_tokens": 152395396.0, + "step": 3919 + }, + { + "epoch": 0.4986642920747996, + "grad_norm": 2.7544105052948, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8852517604827881, + "num_tokens": 152429628.0, + "step": 3920 + }, + { + "epoch": 0.49879150235339015, + "grad_norm": 2.6242237091064453, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8719801902770996, + "num_tokens": 152469885.0, + "step": 3921 + }, + { + "epoch": 0.4989187126319807, + "grad_norm": 2.8579113483428955, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.86998450756073, + "num_tokens": 152504196.0, + "step": 3922 + }, + { + "epoch": 0.49904592291057115, + "grad_norm": 2.6303415298461914, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8822013735771179, + "num_tokens": 152539074.0, + "step": 3923 + }, + { + "epoch": 0.4991731331891617, + "grad_norm": 2.6497836112976074, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.875063419342041, + "num_tokens": 152577641.0, + "step": 3924 + }, + { + "epoch": 0.4993003434677522, + "grad_norm": 2.6598503589630127, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8756383657455444, + "num_tokens": 152616992.0, + "step": 3925 + }, + { + "epoch": 0.4994275537463427, + "grad_norm": 2.7337515354156494, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8761440515518188, + "num_tokens": 152653907.0, + "step": 3926 + }, + { + "epoch": 0.4995547640249332, + "grad_norm": 2.674917221069336, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8713164925575256, + "num_tokens": 152696614.0, + "step": 3927 + }, + { + "epoch": 0.49968197430352373, + "grad_norm": 2.991765022277832, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.87127685546875, + "num_tokens": 152728311.0, + "step": 3928 + }, + { + "epoch": 0.4998091845821142, + "grad_norm": 2.6609294414520264, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8815293312072754, + "num_tokens": 152763512.0, + "step": 3929 + }, + { + "epoch": 0.49993639486070474, + "grad_norm": 2.820060968399048, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8712717890739441, + "num_tokens": 152806125.0, + "step": 3930 + }, + { + "epoch": 0.5000636051392953, + "grad_norm": 2.362513303756714, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8851134181022644, + "num_tokens": 152851854.0, + "step": 3931 + }, + { + "epoch": 0.5001908154178858, + "grad_norm": 2.7171790599823, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8719651699066162, + "num_tokens": 152891668.0, + "step": 3932 + }, + { + "epoch": 0.5003180256964763, + "grad_norm": 2.401949405670166, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8791829347610474, + "num_tokens": 152936288.0, + "step": 3933 + }, + { + "epoch": 0.5004452359750667, + "grad_norm": 2.7325565814971924, + "learning_rate": 1e-06, + "loss": 0.3937, + "mean_token_accuracy": 0.8694538474082947, + "num_tokens": 152977903.0, + "step": 3934 + }, + { + "epoch": 0.5005724462536573, + "grad_norm": 2.4114415645599365, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8823509216308594, + "num_tokens": 153019298.0, + "step": 3935 + }, + { + "epoch": 0.5006996565322478, + "grad_norm": 2.7755391597747803, + "learning_rate": 1e-06, + "loss": 0.3974, + "mean_token_accuracy": 0.859843909740448, + "num_tokens": 153059227.0, + "step": 3936 + }, + { + "epoch": 0.5008268668108383, + "grad_norm": 2.6116368770599365, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8888563513755798, + "num_tokens": 153094614.0, + "step": 3937 + }, + { + "epoch": 0.5009540770894289, + "grad_norm": 2.7548537254333496, + "learning_rate": 1e-06, + "loss": 0.4633, + "mean_token_accuracy": 0.8468079566955566, + "num_tokens": 153140123.0, + "step": 3938 + }, + { + "epoch": 0.5010812873680194, + "grad_norm": 2.4173295497894287, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8800313472747803, + "num_tokens": 153183875.0, + "step": 3939 + }, + { + "epoch": 0.5012084976466098, + "grad_norm": 2.735570192337036, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8756225109100342, + "num_tokens": 153218301.0, + "step": 3940 + }, + { + "epoch": 0.5013357079252003, + "grad_norm": 2.5518269538879395, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8824582695960999, + "num_tokens": 153258930.0, + "step": 3941 + }, + { + "epoch": 0.5014629182037909, + "grad_norm": 2.5807912349700928, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.8662071228027344, + "num_tokens": 153301197.0, + "step": 3942 + }, + { + "epoch": 0.5015901284823814, + "grad_norm": 2.534024238586426, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8685685396194458, + "num_tokens": 153340056.0, + "step": 3943 + }, + { + "epoch": 0.5017173387609719, + "grad_norm": 2.616286277770996, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8806060552597046, + "num_tokens": 153375160.0, + "step": 3944 + }, + { + "epoch": 0.5018445490395624, + "grad_norm": 2.6643259525299072, + "learning_rate": 1e-06, + "loss": 0.4067, + "mean_token_accuracy": 0.8565634489059448, + "num_tokens": 153416126.0, + "step": 3945 + }, + { + "epoch": 0.5019717593181529, + "grad_norm": 2.6195459365844727, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8789089918136597, + "num_tokens": 153455249.0, + "step": 3946 + }, + { + "epoch": 0.5020989695967434, + "grad_norm": 2.4138777256011963, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8767957091331482, + "num_tokens": 153499771.0, + "step": 3947 + }, + { + "epoch": 0.5022261798753339, + "grad_norm": 2.395641326904297, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8878164291381836, + "num_tokens": 153542595.0, + "step": 3948 + }, + { + "epoch": 0.5023533901539244, + "grad_norm": 2.687675952911377, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.884520411491394, + "num_tokens": 153579151.0, + "step": 3949 + }, + { + "epoch": 0.502480600432515, + "grad_norm": 2.951107978820801, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8682302236557007, + "num_tokens": 153616645.0, + "step": 3950 + }, + { + "epoch": 0.5026078107111055, + "grad_norm": 2.62034010887146, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8823475241661072, + "num_tokens": 153649878.0, + "step": 3951 + }, + { + "epoch": 0.5027350209896959, + "grad_norm": 2.9832327365875244, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8629640340805054, + "num_tokens": 153685302.0, + "step": 3952 + }, + { + "epoch": 0.5028622312682864, + "grad_norm": 2.8138978481292725, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8693188428878784, + "num_tokens": 153720181.0, + "step": 3953 + }, + { + "epoch": 0.502989441546877, + "grad_norm": 2.5938332080841064, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8819442391395569, + "num_tokens": 153757402.0, + "step": 3954 + }, + { + "epoch": 0.5031166518254675, + "grad_norm": 2.731139898300171, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8718419075012207, + "num_tokens": 153797322.0, + "step": 3955 + }, + { + "epoch": 0.503243862104058, + "grad_norm": 2.662212610244751, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8748024702072144, + "num_tokens": 153838182.0, + "step": 3956 + }, + { + "epoch": 0.5033710723826486, + "grad_norm": 2.4855611324310303, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8714790940284729, + "num_tokens": 153885155.0, + "step": 3957 + }, + { + "epoch": 0.5034982826612391, + "grad_norm": 2.4938106536865234, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8807272911071777, + "num_tokens": 153924015.0, + "step": 3958 + }, + { + "epoch": 0.5036254929398295, + "grad_norm": 2.7782585620880127, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.8673832416534424, + "num_tokens": 153963672.0, + "step": 3959 + }, + { + "epoch": 0.50375270321842, + "grad_norm": 2.653799057006836, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8759487867355347, + "num_tokens": 154002365.0, + "step": 3960 + }, + { + "epoch": 0.5038799134970106, + "grad_norm": 2.73087215423584, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8734660148620605, + "num_tokens": 154038528.0, + "step": 3961 + }, + { + "epoch": 0.5040071237756011, + "grad_norm": 2.610891580581665, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8871063590049744, + "num_tokens": 154078635.0, + "step": 3962 + }, + { + "epoch": 0.5041343340541916, + "grad_norm": 2.58453106880188, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8712807893753052, + "num_tokens": 154120320.0, + "step": 3963 + }, + { + "epoch": 0.5042615443327821, + "grad_norm": 2.5902204513549805, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8781623840332031, + "num_tokens": 154162030.0, + "step": 3964 + }, + { + "epoch": 0.5043887546113726, + "grad_norm": 2.856247663497925, + "learning_rate": 1e-06, + "loss": 0.3993, + "mean_token_accuracy": 0.8639406561851501, + "num_tokens": 154202650.0, + "step": 3965 + }, + { + "epoch": 0.5045159648899631, + "grad_norm": 2.5071702003479004, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8694765567779541, + "num_tokens": 154245474.0, + "step": 3966 + }, + { + "epoch": 0.5046431751685536, + "grad_norm": 2.580479860305786, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8779594898223877, + "num_tokens": 154283486.0, + "step": 3967 + }, + { + "epoch": 0.5047703854471441, + "grad_norm": 2.823930501937866, + "learning_rate": 1e-06, + "loss": 0.3976, + "mean_token_accuracy": 0.8643462657928467, + "num_tokens": 154320719.0, + "step": 3968 + }, + { + "epoch": 0.5048975957257347, + "grad_norm": 2.75154185295105, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8724793195724487, + "num_tokens": 154356145.0, + "step": 3969 + }, + { + "epoch": 0.5050248060043252, + "grad_norm": 2.5057291984558105, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8822156190872192, + "num_tokens": 154395966.0, + "step": 3970 + }, + { + "epoch": 0.5051520162829156, + "grad_norm": 2.7230465412139893, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8713699579238892, + "num_tokens": 154437225.0, + "step": 3971 + }, + { + "epoch": 0.5052792265615061, + "grad_norm": 2.9357404708862305, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.868552565574646, + "num_tokens": 154472474.0, + "step": 3972 + }, + { + "epoch": 0.5054064368400967, + "grad_norm": 2.7265870571136475, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8710654973983765, + "num_tokens": 154507095.0, + "step": 3973 + }, + { + "epoch": 0.5055336471186872, + "grad_norm": 2.9675328731536865, + "learning_rate": 1e-06, + "loss": 0.4229, + "mean_token_accuracy": 0.8551082015037537, + "num_tokens": 154542639.0, + "step": 3974 + }, + { + "epoch": 0.5056608573972777, + "grad_norm": 2.5958411693573, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8824656009674072, + "num_tokens": 154580764.0, + "step": 3975 + }, + { + "epoch": 0.5057880676758683, + "grad_norm": 2.9245314598083496, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8707233667373657, + "num_tokens": 154612236.0, + "step": 3976 + }, + { + "epoch": 0.5059152779544587, + "grad_norm": 2.6842381954193115, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8795371651649475, + "num_tokens": 154649540.0, + "step": 3977 + }, + { + "epoch": 0.5060424882330492, + "grad_norm": 2.769775390625, + "learning_rate": 1e-06, + "loss": 0.4186, + "mean_token_accuracy": 0.8548080921173096, + "num_tokens": 154690771.0, + "step": 3978 + }, + { + "epoch": 0.5061696985116397, + "grad_norm": 2.6017792224884033, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8849282264709473, + "num_tokens": 154726710.0, + "step": 3979 + }, + { + "epoch": 0.5062969087902303, + "grad_norm": 2.513343572616577, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8822035789489746, + "num_tokens": 154765155.0, + "step": 3980 + }, + { + "epoch": 0.5064241190688208, + "grad_norm": 2.7229297161102295, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.876677930355072, + "num_tokens": 154798050.0, + "step": 3981 + }, + { + "epoch": 0.5065513293474113, + "grad_norm": 2.7405741214752197, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.8636717796325684, + "num_tokens": 154839937.0, + "step": 3982 + }, + { + "epoch": 0.5066785396260017, + "grad_norm": 2.5794458389282227, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8719126582145691, + "num_tokens": 154878974.0, + "step": 3983 + }, + { + "epoch": 0.5068057499045923, + "grad_norm": 2.4717600345611572, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.875385046005249, + "num_tokens": 154923488.0, + "step": 3984 + }, + { + "epoch": 0.5069329601831828, + "grad_norm": 2.6068623065948486, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8682374358177185, + "num_tokens": 154964712.0, + "step": 3985 + }, + { + "epoch": 0.5070601704617733, + "grad_norm": 2.5121328830718994, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8816303610801697, + "num_tokens": 155006947.0, + "step": 3986 + }, + { + "epoch": 0.5071873807403638, + "grad_norm": 2.6590094566345215, + "learning_rate": 1e-06, + "loss": 0.4139, + "mean_token_accuracy": 0.8567380905151367, + "num_tokens": 155052002.0, + "step": 3987 + }, + { + "epoch": 0.5073145910189544, + "grad_norm": 2.5223352909088135, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8774313926696777, + "num_tokens": 155094120.0, + "step": 3988 + }, + { + "epoch": 0.5074418012975448, + "grad_norm": 2.6226813793182373, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8748507499694824, + "num_tokens": 155133179.0, + "step": 3989 + }, + { + "epoch": 0.5075690115761353, + "grad_norm": 2.6206214427948, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8745035529136658, + "num_tokens": 155174600.0, + "step": 3990 + }, + { + "epoch": 0.5076962218547258, + "grad_norm": 2.93524169921875, + "learning_rate": 1e-06, + "loss": 0.4194, + "mean_token_accuracy": 0.8534601926803589, + "num_tokens": 155210722.0, + "step": 3991 + }, + { + "epoch": 0.5078234321333164, + "grad_norm": 2.8434298038482666, + "learning_rate": 1e-06, + "loss": 0.392, + "mean_token_accuracy": 0.8637650012969971, + "num_tokens": 155247058.0, + "step": 3992 + }, + { + "epoch": 0.5079506424119069, + "grad_norm": 2.6754488945007324, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8782421350479126, + "num_tokens": 155284786.0, + "step": 3993 + }, + { + "epoch": 0.5080778526904974, + "grad_norm": 2.7084579467773438, + "learning_rate": 1e-06, + "loss": 0.3934, + "mean_token_accuracy": 0.8650248050689697, + "num_tokens": 155325497.0, + "step": 3994 + }, + { + "epoch": 0.5082050629690879, + "grad_norm": 2.5601048469543457, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8701918125152588, + "num_tokens": 155367190.0, + "step": 3995 + }, + { + "epoch": 0.5083322732476784, + "grad_norm": 2.699526309967041, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8694614171981812, + "num_tokens": 155407358.0, + "step": 3996 + }, + { + "epoch": 0.5084594835262689, + "grad_norm": 2.8778443336486816, + "learning_rate": 1e-06, + "loss": 0.4129, + "mean_token_accuracy": 0.8631705641746521, + "num_tokens": 155443912.0, + "step": 3997 + }, + { + "epoch": 0.5085866938048594, + "grad_norm": 2.646885633468628, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8724620342254639, + "num_tokens": 155482641.0, + "step": 3998 + }, + { + "epoch": 0.50871390408345, + "grad_norm": 2.931802749633789, + "learning_rate": 1e-06, + "loss": 0.4074, + "mean_token_accuracy": 0.8574167490005493, + "num_tokens": 155518251.0, + "step": 3999 + }, + { + "epoch": 0.5088411143620405, + "grad_norm": 2.6810522079467773, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8848253488540649, + "num_tokens": 155552930.0, + "step": 4000 + }, + { + "epoch": 0.5089683246406309, + "grad_norm": 2.7856605052948, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8754244446754456, + "num_tokens": 155585363.0, + "step": 4001 + }, + { + "epoch": 0.5090955349192214, + "grad_norm": 2.6038174629211426, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8795759677886963, + "num_tokens": 155624091.0, + "step": 4002 + }, + { + "epoch": 0.509222745197812, + "grad_norm": 2.651028633117676, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8706812858581543, + "num_tokens": 155664175.0, + "step": 4003 + }, + { + "epoch": 0.5093499554764025, + "grad_norm": 2.65861439704895, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8867921829223633, + "num_tokens": 155698588.0, + "step": 4004 + }, + { + "epoch": 0.509477165754993, + "grad_norm": 2.524012565612793, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8816628456115723, + "num_tokens": 155738626.0, + "step": 4005 + }, + { + "epoch": 0.5096043760335836, + "grad_norm": 2.582735061645508, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.887688398361206, + "num_tokens": 155776059.0, + "step": 4006 + }, + { + "epoch": 0.5097315863121741, + "grad_norm": 2.5933403968811035, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8623472452163696, + "num_tokens": 155817140.0, + "step": 4007 + }, + { + "epoch": 0.5098587965907645, + "grad_norm": 2.6029839515686035, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8856199383735657, + "num_tokens": 155853910.0, + "step": 4008 + }, + { + "epoch": 0.509986006869355, + "grad_norm": 3.0407445430755615, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8742455244064331, + "num_tokens": 155882629.0, + "step": 4009 + }, + { + "epoch": 0.5101132171479456, + "grad_norm": 2.6328072547912598, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.861186146736145, + "num_tokens": 155923547.0, + "step": 4010 + }, + { + "epoch": 0.5102404274265361, + "grad_norm": 2.638131856918335, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8708339333534241, + "num_tokens": 155967981.0, + "step": 4011 + }, + { + "epoch": 0.5103676377051266, + "grad_norm": 3.1178548336029053, + "learning_rate": 1e-06, + "loss": 0.4475, + "mean_token_accuracy": 0.8477570414543152, + "num_tokens": 156003893.0, + "step": 4012 + }, + { + "epoch": 0.5104948479837171, + "grad_norm": 2.6618940830230713, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8751676678657532, + "num_tokens": 156044801.0, + "step": 4013 + }, + { + "epoch": 0.5106220582623076, + "grad_norm": 2.7935543060302734, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8795107007026672, + "num_tokens": 156080074.0, + "step": 4014 + }, + { + "epoch": 0.5107492685408981, + "grad_norm": 2.754167079925537, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8795554041862488, + "num_tokens": 156116298.0, + "step": 4015 + }, + { + "epoch": 0.5108764788194886, + "grad_norm": 2.7372355461120605, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8841036558151245, + "num_tokens": 156154087.0, + "step": 4016 + }, + { + "epoch": 0.5110036890980791, + "grad_norm": 2.59594464302063, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8784412145614624, + "num_tokens": 156193349.0, + "step": 4017 + }, + { + "epoch": 0.5111308993766697, + "grad_norm": 2.5869221687316895, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8859109878540039, + "num_tokens": 156231503.0, + "step": 4018 + }, + { + "epoch": 0.5112581096552602, + "grad_norm": 2.6807501316070557, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8656067848205566, + "num_tokens": 156272310.0, + "step": 4019 + }, + { + "epoch": 0.5113853199338506, + "grad_norm": 2.9047272205352783, + "learning_rate": 1e-06, + "loss": 0.3963, + "mean_token_accuracy": 0.861473798751831, + "num_tokens": 156307200.0, + "step": 4020 + }, + { + "epoch": 0.5115125302124411, + "grad_norm": 2.4795446395874023, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8929533362388611, + "num_tokens": 156343646.0, + "step": 4021 + }, + { + "epoch": 0.5116397404910317, + "grad_norm": 2.527276039123535, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8792628645896912, + "num_tokens": 156385496.0, + "step": 4022 + }, + { + "epoch": 0.5117669507696222, + "grad_norm": 2.5727217197418213, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.8646658658981323, + "num_tokens": 156429842.0, + "step": 4023 + }, + { + "epoch": 0.5118941610482127, + "grad_norm": 2.5671608448028564, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8686000108718872, + "num_tokens": 156472882.0, + "step": 4024 + }, + { + "epoch": 0.5120213713268033, + "grad_norm": 2.5553879737854004, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8843443393707275, + "num_tokens": 156514901.0, + "step": 4025 + }, + { + "epoch": 0.5121485816053937, + "grad_norm": 2.6685168743133545, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8770884871482849, + "num_tokens": 156550713.0, + "step": 4026 + }, + { + "epoch": 0.5122757918839842, + "grad_norm": 2.6372220516204834, + "learning_rate": 1e-06, + "loss": 0.4028, + "mean_token_accuracy": 0.8604593276977539, + "num_tokens": 156593334.0, + "step": 4027 + }, + { + "epoch": 0.5124030021625747, + "grad_norm": 2.595968246459961, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8781689405441284, + "num_tokens": 156632001.0, + "step": 4028 + }, + { + "epoch": 0.5125302124411653, + "grad_norm": 2.789106607437134, + "learning_rate": 1e-06, + "loss": 0.4189, + "mean_token_accuracy": 0.8601428866386414, + "num_tokens": 156672537.0, + "step": 4029 + }, + { + "epoch": 0.5126574227197558, + "grad_norm": 2.6693801879882812, + "learning_rate": 1e-06, + "loss": 0.4123, + "mean_token_accuracy": 0.8582517504692078, + "num_tokens": 156717260.0, + "step": 4030 + }, + { + "epoch": 0.5127846329983463, + "grad_norm": 2.6899125576019287, + "learning_rate": 1e-06, + "loss": 0.397, + "mean_token_accuracy": 0.8610528707504272, + "num_tokens": 156760062.0, + "step": 4031 + }, + { + "epoch": 0.5129118432769367, + "grad_norm": 2.8778469562530518, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8725796341896057, + "num_tokens": 156793327.0, + "step": 4032 + }, + { + "epoch": 0.5130390535555273, + "grad_norm": 2.8257651329040527, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8904392123222351, + "num_tokens": 156827187.0, + "step": 4033 + }, + { + "epoch": 0.5131662638341178, + "grad_norm": 2.569546937942505, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8705838918685913, + "num_tokens": 156872791.0, + "step": 4034 + }, + { + "epoch": 0.5132934741127083, + "grad_norm": 2.661938428878784, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8712145090103149, + "num_tokens": 156914978.0, + "step": 4035 + }, + { + "epoch": 0.5134206843912988, + "grad_norm": 2.8652000427246094, + "learning_rate": 1e-06, + "loss": 0.3863, + "mean_token_accuracy": 0.8646490573883057, + "num_tokens": 156950518.0, + "step": 4036 + }, + { + "epoch": 0.5135478946698894, + "grad_norm": 2.7389297485351562, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.877671480178833, + "num_tokens": 156989534.0, + "step": 4037 + }, + { + "epoch": 0.5136751049484798, + "grad_norm": 2.7343266010284424, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8740609884262085, + "num_tokens": 157026850.0, + "step": 4038 + }, + { + "epoch": 0.5138023152270703, + "grad_norm": 2.730689287185669, + "learning_rate": 1e-06, + "loss": 0.4058, + "mean_token_accuracy": 0.857883095741272, + "num_tokens": 157067494.0, + "step": 4039 + }, + { + "epoch": 0.5139295255056608, + "grad_norm": 2.649665117263794, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8760815858840942, + "num_tokens": 157106805.0, + "step": 4040 + }, + { + "epoch": 0.5140567357842514, + "grad_norm": 2.394718647003174, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8860861659049988, + "num_tokens": 157147508.0, + "step": 4041 + }, + { + "epoch": 0.5141839460628419, + "grad_norm": 2.7046306133270264, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8730543255805969, + "num_tokens": 157185271.0, + "step": 4042 + }, + { + "epoch": 0.5143111563414324, + "grad_norm": 2.587402582168579, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8713529109954834, + "num_tokens": 157225732.0, + "step": 4043 + }, + { + "epoch": 0.5144383666200228, + "grad_norm": 2.4593825340270996, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8756946325302124, + "num_tokens": 157267229.0, + "step": 4044 + }, + { + "epoch": 0.5145655768986134, + "grad_norm": 2.968533515930176, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8712016344070435, + "num_tokens": 157309364.0, + "step": 4045 + }, + { + "epoch": 0.5146927871772039, + "grad_norm": 2.7243666648864746, + "learning_rate": 1e-06, + "loss": 0.3676, + "mean_token_accuracy": 0.8733711242675781, + "num_tokens": 157348459.0, + "step": 4046 + }, + { + "epoch": 0.5148199974557944, + "grad_norm": 2.6176340579986572, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8855636119842529, + "num_tokens": 157383636.0, + "step": 4047 + }, + { + "epoch": 0.514947207734385, + "grad_norm": 2.624671697616577, + "learning_rate": 1e-06, + "loss": 0.3968, + "mean_token_accuracy": 0.8646761178970337, + "num_tokens": 157426750.0, + "step": 4048 + }, + { + "epoch": 0.5150744180129755, + "grad_norm": 2.5792508125305176, + "learning_rate": 1e-06, + "loss": 0.4155, + "mean_token_accuracy": 0.8580970764160156, + "num_tokens": 157470249.0, + "step": 4049 + }, + { + "epoch": 0.5152016282915659, + "grad_norm": 2.888094186782837, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.8656578660011292, + "num_tokens": 157508065.0, + "step": 4050 + }, + { + "epoch": 0.5153288385701564, + "grad_norm": 2.709259271621704, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8771701455116272, + "num_tokens": 157547575.0, + "step": 4051 + }, + { + "epoch": 0.515456048848747, + "grad_norm": 2.582456588745117, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8837506771087646, + "num_tokens": 157586277.0, + "step": 4052 + }, + { + "epoch": 0.5155832591273375, + "grad_norm": 2.8130881786346436, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8711866140365601, + "num_tokens": 157625552.0, + "step": 4053 + }, + { + "epoch": 0.515710469405928, + "grad_norm": 2.4257683753967285, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8773654699325562, + "num_tokens": 157668457.0, + "step": 4054 + }, + { + "epoch": 0.5158376796845185, + "grad_norm": 2.879568338394165, + "learning_rate": 1e-06, + "loss": 0.3927, + "mean_token_accuracy": 0.8624253273010254, + "num_tokens": 157706024.0, + "step": 4055 + }, + { + "epoch": 0.5159648899631091, + "grad_norm": 2.5898404121398926, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8835660219192505, + "num_tokens": 157743828.0, + "step": 4056 + }, + { + "epoch": 0.5160921002416995, + "grad_norm": 2.4077768325805664, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8884283304214478, + "num_tokens": 157786295.0, + "step": 4057 + }, + { + "epoch": 0.51621931052029, + "grad_norm": 2.681441307067871, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8789699077606201, + "num_tokens": 157825232.0, + "step": 4058 + }, + { + "epoch": 0.5163465207988805, + "grad_norm": 2.4947023391723633, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8727201819419861, + "num_tokens": 157868826.0, + "step": 4059 + }, + { + "epoch": 0.5164737310774711, + "grad_norm": 2.5618703365325928, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8733101487159729, + "num_tokens": 157910562.0, + "step": 4060 + }, + { + "epoch": 0.5166009413560616, + "grad_norm": 2.885333776473999, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8790560364723206, + "num_tokens": 157940440.0, + "step": 4061 + }, + { + "epoch": 0.5167281516346521, + "grad_norm": 2.616421937942505, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8762803077697754, + "num_tokens": 157980321.0, + "step": 4062 + }, + { + "epoch": 0.5168553619132426, + "grad_norm": 2.41371750831604, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8819496035575867, + "num_tokens": 158024065.0, + "step": 4063 + }, + { + "epoch": 0.5169825721918331, + "grad_norm": 3.099442720413208, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8704630136489868, + "num_tokens": 158053787.0, + "step": 4064 + }, + { + "epoch": 0.5171097824704236, + "grad_norm": 2.528424024581909, + "learning_rate": 1e-06, + "loss": 0.4146, + "mean_token_accuracy": 0.8575503826141357, + "num_tokens": 158102200.0, + "step": 4065 + }, + { + "epoch": 0.5172369927490141, + "grad_norm": 2.531564950942993, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8747910857200623, + "num_tokens": 158147933.0, + "step": 4066 + }, + { + "epoch": 0.5173642030276047, + "grad_norm": 2.6220595836639404, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8741843700408936, + "num_tokens": 158186619.0, + "step": 4067 + }, + { + "epoch": 0.5174914133061952, + "grad_norm": 5.9391279220581055, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8893546462059021, + "num_tokens": 158221472.0, + "step": 4068 + }, + { + "epoch": 0.5176186235847856, + "grad_norm": 2.7011191844940186, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8757630586624146, + "num_tokens": 158259921.0, + "step": 4069 + }, + { + "epoch": 0.5177458338633761, + "grad_norm": 2.616171360015869, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8814576864242554, + "num_tokens": 158295346.0, + "step": 4070 + }, + { + "epoch": 0.5178730441419667, + "grad_norm": 2.5988943576812744, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8798019886016846, + "num_tokens": 158336459.0, + "step": 4071 + }, + { + "epoch": 0.5180002544205572, + "grad_norm": 2.693615674972534, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8920514583587646, + "num_tokens": 158369769.0, + "step": 4072 + }, + { + "epoch": 0.5181274646991477, + "grad_norm": 2.519181251525879, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8735220432281494, + "num_tokens": 158412173.0, + "step": 4073 + }, + { + "epoch": 0.5182546749777382, + "grad_norm": 2.4894306659698486, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8768237233161926, + "num_tokens": 158455101.0, + "step": 4074 + }, + { + "epoch": 0.5183818852563287, + "grad_norm": 2.56707501411438, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8759453296661377, + "num_tokens": 158496896.0, + "step": 4075 + }, + { + "epoch": 0.5185090955349192, + "grad_norm": 2.741283893585205, + "learning_rate": 1e-06, + "loss": 0.4316, + "mean_token_accuracy": 0.8519501686096191, + "num_tokens": 158538594.0, + "step": 4076 + }, + { + "epoch": 0.5186363058135097, + "grad_norm": 3.1214046478271484, + "learning_rate": 1e-06, + "loss": 0.4123, + "mean_token_accuracy": 0.8604283332824707, + "num_tokens": 158570262.0, + "step": 4077 + }, + { + "epoch": 0.5187635160921003, + "grad_norm": 2.3379509449005127, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8837840557098389, + "num_tokens": 158615719.0, + "step": 4078 + }, + { + "epoch": 0.5188907263706908, + "grad_norm": 2.7190473079681396, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8716241717338562, + "num_tokens": 158653310.0, + "step": 4079 + }, + { + "epoch": 0.5190179366492813, + "grad_norm": 2.667490243911743, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.8624464273452759, + "num_tokens": 158692184.0, + "step": 4080 + }, + { + "epoch": 0.5191451469278717, + "grad_norm": 2.578303575515747, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8872755765914917, + "num_tokens": 158728419.0, + "step": 4081 + }, + { + "epoch": 0.5192723572064623, + "grad_norm": 2.5936715602874756, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8823418617248535, + "num_tokens": 158765581.0, + "step": 4082 + }, + { + "epoch": 0.5193995674850528, + "grad_norm": 2.9473202228546143, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8814762234687805, + "num_tokens": 158797917.0, + "step": 4083 + }, + { + "epoch": 0.5195267777636433, + "grad_norm": 2.839804172515869, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8622814416885376, + "num_tokens": 158833379.0, + "step": 4084 + }, + { + "epoch": 0.5196539880422338, + "grad_norm": 2.6566035747528076, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8763750195503235, + "num_tokens": 158871156.0, + "step": 4085 + }, + { + "epoch": 0.5197811983208244, + "grad_norm": 2.5370452404022217, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8828858137130737, + "num_tokens": 158911140.0, + "step": 4086 + }, + { + "epoch": 0.5199084085994148, + "grad_norm": 2.532608985900879, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8819118738174438, + "num_tokens": 158952094.0, + "step": 4087 + }, + { + "epoch": 0.5200356188780053, + "grad_norm": 2.5491232872009277, + "learning_rate": 1e-06, + "loss": 0.4025, + "mean_token_accuracy": 0.8610556125640869, + "num_tokens": 158997407.0, + "step": 4088 + }, + { + "epoch": 0.5201628291565958, + "grad_norm": 2.665207862854004, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.888064444065094, + "num_tokens": 159033795.0, + "step": 4089 + }, + { + "epoch": 0.5202900394351864, + "grad_norm": 2.719128370285034, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8777811527252197, + "num_tokens": 159069986.0, + "step": 4090 + }, + { + "epoch": 0.5204172497137769, + "grad_norm": 2.6239569187164307, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.8666613101959229, + "num_tokens": 159113329.0, + "step": 4091 + }, + { + "epoch": 0.5205444599923674, + "grad_norm": 2.9935142993927, + "learning_rate": 1e-06, + "loss": 0.4132, + "mean_token_accuracy": 0.8587523102760315, + "num_tokens": 159148890.0, + "step": 4092 + }, + { + "epoch": 0.5206716702709578, + "grad_norm": 2.684587240219116, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.8645334839820862, + "num_tokens": 159193085.0, + "step": 4093 + }, + { + "epoch": 0.5207988805495484, + "grad_norm": 2.5390443801879883, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8800539970397949, + "num_tokens": 159235564.0, + "step": 4094 + }, + { + "epoch": 0.5209260908281389, + "grad_norm": 2.5671586990356445, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8792445659637451, + "num_tokens": 159276944.0, + "step": 4095 + }, + { + "epoch": 0.5210533011067294, + "grad_norm": 2.853940486907959, + "learning_rate": 1e-06, + "loss": 0.4171, + "mean_token_accuracy": 0.8584047555923462, + "num_tokens": 159317708.0, + "step": 4096 + }, + { + "epoch": 0.52118051138532, + "grad_norm": 2.5651869773864746, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8679199814796448, + "num_tokens": 159360748.0, + "step": 4097 + }, + { + "epoch": 0.5213077216639105, + "grad_norm": 3.0546982288360596, + "learning_rate": 1e-06, + "loss": 0.3937, + "mean_token_accuracy": 0.8702394962310791, + "num_tokens": 159393076.0, + "step": 4098 + }, + { + "epoch": 0.5214349319425009, + "grad_norm": 2.5445730686187744, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8821049928665161, + "num_tokens": 159431507.0, + "step": 4099 + }, + { + "epoch": 0.5215621422210914, + "grad_norm": 2.6863300800323486, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8723015785217285, + "num_tokens": 159470195.0, + "step": 4100 + }, + { + "epoch": 0.521689352499682, + "grad_norm": 2.638580083847046, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8892669081687927, + "num_tokens": 159503396.0, + "step": 4101 + }, + { + "epoch": 0.5218165627782725, + "grad_norm": 2.5989768505096436, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.881307065486908, + "num_tokens": 159542409.0, + "step": 4102 + }, + { + "epoch": 0.521943773056863, + "grad_norm": 2.562856674194336, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8857343196868896, + "num_tokens": 159582781.0, + "step": 4103 + }, + { + "epoch": 0.5220709833354535, + "grad_norm": 2.6202139854431152, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8848932981491089, + "num_tokens": 159618584.0, + "step": 4104 + }, + { + "epoch": 0.522198193614044, + "grad_norm": 2.7572286128997803, + "learning_rate": 1e-06, + "loss": 0.4315, + "mean_token_accuracy": 0.8550176620483398, + "num_tokens": 159664658.0, + "step": 4105 + }, + { + "epoch": 0.5223254038926345, + "grad_norm": 2.5292000770568848, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8649352788925171, + "num_tokens": 159707079.0, + "step": 4106 + }, + { + "epoch": 0.522452614171225, + "grad_norm": 2.671837091445923, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8682123422622681, + "num_tokens": 159747347.0, + "step": 4107 + }, + { + "epoch": 0.5225798244498155, + "grad_norm": 2.878091812133789, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8769413232803345, + "num_tokens": 159780716.0, + "step": 4108 + }, + { + "epoch": 0.5227070347284061, + "grad_norm": 2.8843507766723633, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8660334348678589, + "num_tokens": 159814809.0, + "step": 4109 + }, + { + "epoch": 0.5228342450069966, + "grad_norm": 2.5652027130126953, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8757724761962891, + "num_tokens": 159853606.0, + "step": 4110 + }, + { + "epoch": 0.5229614552855871, + "grad_norm": 2.4948222637176514, + "learning_rate": 1e-06, + "loss": 0.3979, + "mean_token_accuracy": 0.8637337684631348, + "num_tokens": 159898598.0, + "step": 4111 + }, + { + "epoch": 0.5230886655641775, + "grad_norm": 2.557340621948242, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8780862092971802, + "num_tokens": 159938160.0, + "step": 4112 + }, + { + "epoch": 0.5232158758427681, + "grad_norm": 2.9253594875335693, + "learning_rate": 1e-06, + "loss": 0.3835, + "mean_token_accuracy": 0.866351306438446, + "num_tokens": 159971563.0, + "step": 4113 + }, + { + "epoch": 0.5233430861213586, + "grad_norm": 2.7624664306640625, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8630829453468323, + "num_tokens": 160010518.0, + "step": 4114 + }, + { + "epoch": 0.5234702963999491, + "grad_norm": 2.6327788829803467, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8782448172569275, + "num_tokens": 160051063.0, + "step": 4115 + }, + { + "epoch": 0.5235975066785397, + "grad_norm": 3.093623399734497, + "learning_rate": 1e-06, + "loss": 0.4356, + "mean_token_accuracy": 0.8511170744895935, + "num_tokens": 160085150.0, + "step": 4116 + }, + { + "epoch": 0.5237247169571302, + "grad_norm": 2.7947874069213867, + "learning_rate": 1e-06, + "loss": 0.391, + "mean_token_accuracy": 0.8662214279174805, + "num_tokens": 160120078.0, + "step": 4117 + }, + { + "epoch": 0.5238519272357206, + "grad_norm": 2.7628931999206543, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.874146580696106, + "num_tokens": 160155899.0, + "step": 4118 + }, + { + "epoch": 0.5239791375143111, + "grad_norm": 2.483311176300049, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8796991109848022, + "num_tokens": 160197069.0, + "step": 4119 + }, + { + "epoch": 0.5241063477929017, + "grad_norm": 2.6137685775756836, + "learning_rate": 1e-06, + "loss": 0.4441, + "mean_token_accuracy": 0.8476381301879883, + "num_tokens": 160243696.0, + "step": 4120 + }, + { + "epoch": 0.5242335580714922, + "grad_norm": 2.5627682209014893, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8852832913398743, + "num_tokens": 160282157.0, + "step": 4121 + }, + { + "epoch": 0.5243607683500827, + "grad_norm": 2.582689046859741, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.8642896413803101, + "num_tokens": 160322259.0, + "step": 4122 + }, + { + "epoch": 0.5244879786286732, + "grad_norm": 2.7867038249969482, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8776235580444336, + "num_tokens": 160359049.0, + "step": 4123 + }, + { + "epoch": 0.5246151889072637, + "grad_norm": 2.588562250137329, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8638172745704651, + "num_tokens": 160404403.0, + "step": 4124 + }, + { + "epoch": 0.5247423991858542, + "grad_norm": 2.7717580795288086, + "learning_rate": 1e-06, + "loss": 0.398, + "mean_token_accuracy": 0.8636617660522461, + "num_tokens": 160443037.0, + "step": 4125 + }, + { + "epoch": 0.5248696094644447, + "grad_norm": 2.555544376373291, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8781401515007019, + "num_tokens": 160486850.0, + "step": 4126 + }, + { + "epoch": 0.5249968197430352, + "grad_norm": 3.0305416584014893, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8721404671669006, + "num_tokens": 160519103.0, + "step": 4127 + }, + { + "epoch": 0.5251240300216258, + "grad_norm": 2.6735496520996094, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8766418695449829, + "num_tokens": 160560854.0, + "step": 4128 + }, + { + "epoch": 0.5252512403002163, + "grad_norm": 2.5796828269958496, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8893722891807556, + "num_tokens": 160598678.0, + "step": 4129 + }, + { + "epoch": 0.5253784505788067, + "grad_norm": 2.662630319595337, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8701944351196289, + "num_tokens": 160636682.0, + "step": 4130 + }, + { + "epoch": 0.5255056608573972, + "grad_norm": 2.5501952171325684, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8844304084777832, + "num_tokens": 160674489.0, + "step": 4131 + }, + { + "epoch": 0.5256328711359878, + "grad_norm": 2.8160691261291504, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8781812191009521, + "num_tokens": 160708101.0, + "step": 4132 + }, + { + "epoch": 0.5257600814145783, + "grad_norm": 2.828850507736206, + "learning_rate": 1e-06, + "loss": 0.4422, + "mean_token_accuracy": 0.8525873422622681, + "num_tokens": 160751682.0, + "step": 4133 + }, + { + "epoch": 0.5258872916931688, + "grad_norm": 2.5820071697235107, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8867878913879395, + "num_tokens": 160791645.0, + "step": 4134 + }, + { + "epoch": 0.5260145019717594, + "grad_norm": 2.7765114307403564, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8727505803108215, + "num_tokens": 160826991.0, + "step": 4135 + }, + { + "epoch": 0.5261417122503498, + "grad_norm": 2.8824803829193115, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8730836510658264, + "num_tokens": 160867228.0, + "step": 4136 + }, + { + "epoch": 0.5262689225289403, + "grad_norm": 2.9583427906036377, + "learning_rate": 1e-06, + "loss": 0.4108, + "mean_token_accuracy": 0.8625869154930115, + "num_tokens": 160905716.0, + "step": 4137 + }, + { + "epoch": 0.5263961328075308, + "grad_norm": 2.6819710731506348, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8722123503684998, + "num_tokens": 160944918.0, + "step": 4138 + }, + { + "epoch": 0.5265233430861214, + "grad_norm": 2.742936134338379, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8805861473083496, + "num_tokens": 160986685.0, + "step": 4139 + }, + { + "epoch": 0.5266505533647119, + "grad_norm": 2.7231714725494385, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8701241612434387, + "num_tokens": 161025624.0, + "step": 4140 + }, + { + "epoch": 0.5267777636433024, + "grad_norm": 2.8081412315368652, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.8657196760177612, + "num_tokens": 161066431.0, + "step": 4141 + }, + { + "epoch": 0.5269049739218928, + "grad_norm": 2.610313892364502, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8748796582221985, + "num_tokens": 161103089.0, + "step": 4142 + }, + { + "epoch": 0.5270321842004834, + "grad_norm": 2.447695732116699, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8800332546234131, + "num_tokens": 161143653.0, + "step": 4143 + }, + { + "epoch": 0.5271593944790739, + "grad_norm": 2.8941867351531982, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8596112728118896, + "num_tokens": 161178423.0, + "step": 4144 + }, + { + "epoch": 0.5272866047576644, + "grad_norm": 2.527722120285034, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8838070631027222, + "num_tokens": 161220357.0, + "step": 4145 + }, + { + "epoch": 0.527413815036255, + "grad_norm": 2.6037261486053467, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.875569760799408, + "num_tokens": 161261597.0, + "step": 4146 + }, + { + "epoch": 0.5275410253148455, + "grad_norm": 3.4159841537475586, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8779814839363098, + "num_tokens": 161298843.0, + "step": 4147 + }, + { + "epoch": 0.5276682355934359, + "grad_norm": 2.6509084701538086, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8847453594207764, + "num_tokens": 161340199.0, + "step": 4148 + }, + { + "epoch": 0.5277954458720264, + "grad_norm": 2.605426788330078, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.876976490020752, + "num_tokens": 161377413.0, + "step": 4149 + }, + { + "epoch": 0.527922656150617, + "grad_norm": 2.8415186405181885, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8724349737167358, + "num_tokens": 161411490.0, + "step": 4150 + }, + { + "epoch": 0.5280498664292075, + "grad_norm": 2.655672788619995, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.8602328896522522, + "num_tokens": 161456697.0, + "step": 4151 + }, + { + "epoch": 0.528177076707798, + "grad_norm": 2.714836359024048, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.87195885181427, + "num_tokens": 161494386.0, + "step": 4152 + }, + { + "epoch": 0.5283042869863885, + "grad_norm": 2.6457912921905518, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8767121434211731, + "num_tokens": 161536424.0, + "step": 4153 + }, + { + "epoch": 0.528431497264979, + "grad_norm": 2.6104300022125244, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8840926885604858, + "num_tokens": 161574557.0, + "step": 4154 + }, + { + "epoch": 0.5285587075435695, + "grad_norm": 2.779616355895996, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.8716766834259033, + "num_tokens": 161611642.0, + "step": 4155 + }, + { + "epoch": 0.52868591782216, + "grad_norm": 2.700673818588257, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8770351409912109, + "num_tokens": 161652176.0, + "step": 4156 + }, + { + "epoch": 0.5288131281007505, + "grad_norm": 2.520764112472534, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8828913569450378, + "num_tokens": 161689808.0, + "step": 4157 + }, + { + "epoch": 0.5289403383793411, + "grad_norm": 2.4283876419067383, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8877213001251221, + "num_tokens": 161730607.0, + "step": 4158 + }, + { + "epoch": 0.5290675486579316, + "grad_norm": 2.8176584243774414, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.87596595287323, + "num_tokens": 161763617.0, + "step": 4159 + }, + { + "epoch": 0.5291947589365221, + "grad_norm": 2.7699103355407715, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8745249509811401, + "num_tokens": 161802493.0, + "step": 4160 + }, + { + "epoch": 0.5293219692151125, + "grad_norm": 3.0711331367492676, + "learning_rate": 1e-06, + "loss": 0.4265, + "mean_token_accuracy": 0.8537236452102661, + "num_tokens": 161837531.0, + "step": 4161 + }, + { + "epoch": 0.5294491794937031, + "grad_norm": 2.930664539337158, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8646659851074219, + "num_tokens": 161870921.0, + "step": 4162 + }, + { + "epoch": 0.5295763897722936, + "grad_norm": 2.642639636993408, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.885549783706665, + "num_tokens": 161907131.0, + "step": 4163 + }, + { + "epoch": 0.5297036000508841, + "grad_norm": 2.596121311187744, + "learning_rate": 1e-06, + "loss": 0.3668, + "mean_token_accuracy": 0.872888445854187, + "num_tokens": 161947376.0, + "step": 4164 + }, + { + "epoch": 0.5298308103294747, + "grad_norm": 2.561713457107544, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8710887432098389, + "num_tokens": 161990291.0, + "step": 4165 + }, + { + "epoch": 0.5299580206080652, + "grad_norm": 2.4855237007141113, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8844199776649475, + "num_tokens": 162032213.0, + "step": 4166 + }, + { + "epoch": 0.5300852308866556, + "grad_norm": 2.5439488887786865, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8607274293899536, + "num_tokens": 162073579.0, + "step": 4167 + }, + { + "epoch": 0.5302124411652461, + "grad_norm": 2.6454575061798096, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8807881474494934, + "num_tokens": 162111301.0, + "step": 4168 + }, + { + "epoch": 0.5303396514438367, + "grad_norm": 2.8067467212677, + "learning_rate": 1e-06, + "loss": 0.4165, + "mean_token_accuracy": 0.857672929763794, + "num_tokens": 162151636.0, + "step": 4169 + }, + { + "epoch": 0.5304668617224272, + "grad_norm": 2.547192096710205, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8880957365036011, + "num_tokens": 162190304.0, + "step": 4170 + }, + { + "epoch": 0.5305940720010177, + "grad_norm": 2.7121739387512207, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8783227205276489, + "num_tokens": 162225372.0, + "step": 4171 + }, + { + "epoch": 0.5307212822796082, + "grad_norm": 2.7207891941070557, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8820129036903381, + "num_tokens": 162260538.0, + "step": 4172 + }, + { + "epoch": 0.5308484925581987, + "grad_norm": 3.0051534175872803, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8726641535758972, + "num_tokens": 162291668.0, + "step": 4173 + }, + { + "epoch": 0.5309757028367892, + "grad_norm": 2.4745967388153076, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8853311538696289, + "num_tokens": 162332491.0, + "step": 4174 + }, + { + "epoch": 0.5311029131153797, + "grad_norm": 2.7408125400543213, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.865234911441803, + "num_tokens": 162373407.0, + "step": 4175 + }, + { + "epoch": 0.5312301233939702, + "grad_norm": 2.7037923336029053, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8849952220916748, + "num_tokens": 162407273.0, + "step": 4176 + }, + { + "epoch": 0.5313573336725608, + "grad_norm": 3.091764450073242, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8654292821884155, + "num_tokens": 162440245.0, + "step": 4177 + }, + { + "epoch": 0.5314845439511513, + "grad_norm": 2.6921989917755127, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8809515237808228, + "num_tokens": 162476094.0, + "step": 4178 + }, + { + "epoch": 0.5316117542297417, + "grad_norm": 2.4722495079040527, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8697822690010071, + "num_tokens": 162520749.0, + "step": 4179 + }, + { + "epoch": 0.5317389645083322, + "grad_norm": 2.640500545501709, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.87102872133255, + "num_tokens": 162559447.0, + "step": 4180 + }, + { + "epoch": 0.5318661747869228, + "grad_norm": 2.6513378620147705, + "learning_rate": 1e-06, + "loss": 0.4072, + "mean_token_accuracy": 0.860148549079895, + "num_tokens": 162599915.0, + "step": 4181 + }, + { + "epoch": 0.5319933850655133, + "grad_norm": 2.6233596801757812, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8721911907196045, + "num_tokens": 162638735.0, + "step": 4182 + }, + { + "epoch": 0.5321205953441038, + "grad_norm": 2.672696590423584, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8812878131866455, + "num_tokens": 162676337.0, + "step": 4183 + }, + { + "epoch": 0.5322478056226944, + "grad_norm": 2.7858946323394775, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8805433511734009, + "num_tokens": 162707840.0, + "step": 4184 + }, + { + "epoch": 0.5323750159012848, + "grad_norm": 2.6421902179718018, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8709456920623779, + "num_tokens": 162748534.0, + "step": 4185 + }, + { + "epoch": 0.5325022261798753, + "grad_norm": 2.6904077529907227, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8843258619308472, + "num_tokens": 162784494.0, + "step": 4186 + }, + { + "epoch": 0.5326294364584658, + "grad_norm": 2.6304614543914795, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.8649663925170898, + "num_tokens": 162825361.0, + "step": 4187 + }, + { + "epoch": 0.5327566467370564, + "grad_norm": 2.5988237857818604, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.874455988407135, + "num_tokens": 162867724.0, + "step": 4188 + }, + { + "epoch": 0.5328838570156469, + "grad_norm": 2.473888874053955, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8767163753509521, + "num_tokens": 162909839.0, + "step": 4189 + }, + { + "epoch": 0.5330110672942374, + "grad_norm": 2.7380313873291016, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8711506724357605, + "num_tokens": 162944464.0, + "step": 4190 + }, + { + "epoch": 0.5331382775728278, + "grad_norm": 2.412008762359619, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8899170160293579, + "num_tokens": 162987242.0, + "step": 4191 + }, + { + "epoch": 0.5332654878514184, + "grad_norm": 2.924523115158081, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8836729526519775, + "num_tokens": 163016851.0, + "step": 4192 + }, + { + "epoch": 0.5333926981300089, + "grad_norm": 2.7990736961364746, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8761972188949585, + "num_tokens": 163052635.0, + "step": 4193 + }, + { + "epoch": 0.5335199084085994, + "grad_norm": 2.8982768058776855, + "learning_rate": 1e-06, + "loss": 0.4064, + "mean_token_accuracy": 0.8600729703903198, + "num_tokens": 163089270.0, + "step": 4194 + }, + { + "epoch": 0.53364711868719, + "grad_norm": 2.573608160018921, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8752206563949585, + "num_tokens": 163133860.0, + "step": 4195 + }, + { + "epoch": 0.5337743289657805, + "grad_norm": 2.6742770671844482, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8670780062675476, + "num_tokens": 163176613.0, + "step": 4196 + }, + { + "epoch": 0.5339015392443709, + "grad_norm": 2.828995943069458, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8723893165588379, + "num_tokens": 163213865.0, + "step": 4197 + }, + { + "epoch": 0.5340287495229614, + "grad_norm": 2.7224929332733154, + "learning_rate": 1e-06, + "loss": 0.4147, + "mean_token_accuracy": 0.8559982776641846, + "num_tokens": 163257040.0, + "step": 4198 + }, + { + "epoch": 0.534155959801552, + "grad_norm": 2.3276402950286865, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8901566863059998, + "num_tokens": 163300889.0, + "step": 4199 + }, + { + "epoch": 0.5342831700801425, + "grad_norm": 2.869652509689331, + "learning_rate": 1e-06, + "loss": 0.4061, + "mean_token_accuracy": 0.8599057197570801, + "num_tokens": 163338120.0, + "step": 4200 + }, + { + "epoch": 0.534410380358733, + "grad_norm": 2.871582269668579, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8643114566802979, + "num_tokens": 163375520.0, + "step": 4201 + }, + { + "epoch": 0.5345375906373235, + "grad_norm": 2.512927532196045, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8909169435501099, + "num_tokens": 163412371.0, + "step": 4202 + }, + { + "epoch": 0.534664800915914, + "grad_norm": 2.8758814334869385, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8848087787628174, + "num_tokens": 163445430.0, + "step": 4203 + }, + { + "epoch": 0.5347920111945045, + "grad_norm": 2.9112091064453125, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8627568483352661, + "num_tokens": 163482320.0, + "step": 4204 + }, + { + "epoch": 0.534919221473095, + "grad_norm": 2.566730260848999, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8733474016189575, + "num_tokens": 163526388.0, + "step": 4205 + }, + { + "epoch": 0.5350464317516855, + "grad_norm": 2.6794216632843018, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.869033932685852, + "num_tokens": 163568068.0, + "step": 4206 + }, + { + "epoch": 0.5351736420302761, + "grad_norm": 2.7530345916748047, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8792392015457153, + "num_tokens": 163603813.0, + "step": 4207 + }, + { + "epoch": 0.5353008523088666, + "grad_norm": 2.563065767288208, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8875815868377686, + "num_tokens": 163642993.0, + "step": 4208 + }, + { + "epoch": 0.5354280625874571, + "grad_norm": 2.391932964324951, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8932901620864868, + "num_tokens": 163683717.0, + "step": 4209 + }, + { + "epoch": 0.5355552728660475, + "grad_norm": 2.8452749252319336, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8759538531303406, + "num_tokens": 163719061.0, + "step": 4210 + }, + { + "epoch": 0.5356824831446381, + "grad_norm": 2.8503684997558594, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8798494935035706, + "num_tokens": 163753243.0, + "step": 4211 + }, + { + "epoch": 0.5358096934232286, + "grad_norm": 2.951596736907959, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8890786170959473, + "num_tokens": 163786668.0, + "step": 4212 + }, + { + "epoch": 0.5359369037018191, + "grad_norm": 2.7937681674957275, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8703979849815369, + "num_tokens": 163824696.0, + "step": 4213 + }, + { + "epoch": 0.5360641139804097, + "grad_norm": 2.5577239990234375, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8775651454925537, + "num_tokens": 163864470.0, + "step": 4214 + }, + { + "epoch": 0.5361913242590002, + "grad_norm": 2.649008274078369, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8821618556976318, + "num_tokens": 163900346.0, + "step": 4215 + }, + { + "epoch": 0.5363185345375906, + "grad_norm": 2.5481507778167725, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8678494691848755, + "num_tokens": 163941188.0, + "step": 4216 + }, + { + "epoch": 0.5364457448161811, + "grad_norm": 2.9035184383392334, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8724289536476135, + "num_tokens": 163977245.0, + "step": 4217 + }, + { + "epoch": 0.5365729550947717, + "grad_norm": 2.7289087772369385, + "learning_rate": 1e-06, + "loss": 0.4126, + "mean_token_accuracy": 0.858396053314209, + "num_tokens": 164019118.0, + "step": 4218 + }, + { + "epoch": 0.5367001653733622, + "grad_norm": 2.7880916595458984, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8697389960289001, + "num_tokens": 164055929.0, + "step": 4219 + }, + { + "epoch": 0.5368273756519527, + "grad_norm": 2.719780921936035, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8719846606254578, + "num_tokens": 164091995.0, + "step": 4220 + }, + { + "epoch": 0.5369545859305432, + "grad_norm": 2.526853084564209, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8740777373313904, + "num_tokens": 164135080.0, + "step": 4221 + }, + { + "epoch": 0.5370817962091337, + "grad_norm": 2.815953254699707, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8664575815200806, + "num_tokens": 164172472.0, + "step": 4222 + }, + { + "epoch": 0.5372090064877242, + "grad_norm": 2.518073081970215, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8920917510986328, + "num_tokens": 164208440.0, + "step": 4223 + }, + { + "epoch": 0.5373362167663147, + "grad_norm": 2.9956276416778564, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.861182689666748, + "num_tokens": 164242547.0, + "step": 4224 + }, + { + "epoch": 0.5374634270449052, + "grad_norm": 2.5341956615448, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8849011659622192, + "num_tokens": 164283096.0, + "step": 4225 + }, + { + "epoch": 0.5375906373234958, + "grad_norm": 2.58074688911438, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8739351630210876, + "num_tokens": 164324283.0, + "step": 4226 + }, + { + "epoch": 0.5377178476020863, + "grad_norm": 2.501513957977295, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8780826330184937, + "num_tokens": 164367742.0, + "step": 4227 + }, + { + "epoch": 0.5378450578806767, + "grad_norm": 2.778932571411133, + "learning_rate": 1e-06, + "loss": 0.3953, + "mean_token_accuracy": 0.8622108697891235, + "num_tokens": 164403361.0, + "step": 4228 + }, + { + "epoch": 0.5379722681592672, + "grad_norm": 2.560333013534546, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8729320764541626, + "num_tokens": 164442204.0, + "step": 4229 + }, + { + "epoch": 0.5380994784378578, + "grad_norm": 2.686891794204712, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8794558644294739, + "num_tokens": 164479522.0, + "step": 4230 + }, + { + "epoch": 0.5382266887164483, + "grad_norm": 2.7352168560028076, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.866466760635376, + "num_tokens": 164520004.0, + "step": 4231 + }, + { + "epoch": 0.5383538989950388, + "grad_norm": 2.554483413696289, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8765715956687927, + "num_tokens": 164559968.0, + "step": 4232 + }, + { + "epoch": 0.5384811092736294, + "grad_norm": 3.0589795112609863, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.859384298324585, + "num_tokens": 164591234.0, + "step": 4233 + }, + { + "epoch": 0.5386083195522198, + "grad_norm": 2.706157922744751, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8683272004127502, + "num_tokens": 164628309.0, + "step": 4234 + }, + { + "epoch": 0.5387355298308103, + "grad_norm": 2.3706583976745605, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8786516189575195, + "num_tokens": 164675542.0, + "step": 4235 + }, + { + "epoch": 0.5388627401094008, + "grad_norm": 2.708249568939209, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.877193033695221, + "num_tokens": 164712010.0, + "step": 4236 + }, + { + "epoch": 0.5389899503879914, + "grad_norm": 2.8329927921295166, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8667847514152527, + "num_tokens": 164746938.0, + "step": 4237 + }, + { + "epoch": 0.5391171606665819, + "grad_norm": 2.5365095138549805, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8857380151748657, + "num_tokens": 164783961.0, + "step": 4238 + }, + { + "epoch": 0.5392443709451724, + "grad_norm": 2.5661332607269287, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8882406949996948, + "num_tokens": 164819233.0, + "step": 4239 + }, + { + "epoch": 0.5393715812237628, + "grad_norm": 2.505722761154175, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8819190263748169, + "num_tokens": 164856696.0, + "step": 4240 + }, + { + "epoch": 0.5394987915023534, + "grad_norm": 2.9427108764648438, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.872843325138092, + "num_tokens": 164893741.0, + "step": 4241 + }, + { + "epoch": 0.5396260017809439, + "grad_norm": 2.5965845584869385, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8837127089500427, + "num_tokens": 164933513.0, + "step": 4242 + }, + { + "epoch": 0.5397532120595344, + "grad_norm": 2.471470594406128, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8697138428688049, + "num_tokens": 164980463.0, + "step": 4243 + }, + { + "epoch": 0.5398804223381249, + "grad_norm": 2.6712334156036377, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8719074726104736, + "num_tokens": 165022890.0, + "step": 4244 + }, + { + "epoch": 0.5400076326167155, + "grad_norm": 2.3903682231903076, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8936216235160828, + "num_tokens": 165061135.0, + "step": 4245 + }, + { + "epoch": 0.5401348428953059, + "grad_norm": 2.9853079319000244, + "learning_rate": 1e-06, + "loss": 0.3923, + "mean_token_accuracy": 0.8631347417831421, + "num_tokens": 165092480.0, + "step": 4246 + }, + { + "epoch": 0.5402620531738964, + "grad_norm": 2.976318597793579, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.869855523109436, + "num_tokens": 165126703.0, + "step": 4247 + }, + { + "epoch": 0.540389263452487, + "grad_norm": 2.960679531097412, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8792870044708252, + "num_tokens": 165160322.0, + "step": 4248 + }, + { + "epoch": 0.5405164737310775, + "grad_norm": 2.585970640182495, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8749254941940308, + "num_tokens": 165203339.0, + "step": 4249 + }, + { + "epoch": 0.540643684009668, + "grad_norm": 2.7744140625, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8762853741645813, + "num_tokens": 165239171.0, + "step": 4250 + }, + { + "epoch": 0.5407708942882585, + "grad_norm": 2.4675912857055664, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8767168521881104, + "num_tokens": 165282182.0, + "step": 4251 + }, + { + "epoch": 0.540898104566849, + "grad_norm": 2.600156784057617, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8793506026268005, + "num_tokens": 165322467.0, + "step": 4252 + }, + { + "epoch": 0.5410253148454395, + "grad_norm": 2.7818119525909424, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8717827796936035, + "num_tokens": 165357749.0, + "step": 4253 + }, + { + "epoch": 0.54115252512403, + "grad_norm": 2.9080002307891846, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8700717687606812, + "num_tokens": 165388667.0, + "step": 4254 + }, + { + "epoch": 0.5412797354026205, + "grad_norm": 2.7381672859191895, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8769960403442383, + "num_tokens": 165425262.0, + "step": 4255 + }, + { + "epoch": 0.5414069456812111, + "grad_norm": 2.8217341899871826, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8708502054214478, + "num_tokens": 165462717.0, + "step": 4256 + }, + { + "epoch": 0.5415341559598016, + "grad_norm": 2.674954652786255, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.87667316198349, + "num_tokens": 165498091.0, + "step": 4257 + }, + { + "epoch": 0.5416613662383921, + "grad_norm": 3.0402071475982666, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8788295984268188, + "num_tokens": 165530434.0, + "step": 4258 + }, + { + "epoch": 0.5417885765169825, + "grad_norm": 2.5621557235717773, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8817776441574097, + "num_tokens": 165568168.0, + "step": 4259 + }, + { + "epoch": 0.5419157867955731, + "grad_norm": 2.3789422512054443, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8767158389091492, + "num_tokens": 165613323.0, + "step": 4260 + }, + { + "epoch": 0.5420429970741636, + "grad_norm": 2.971881866455078, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8708564043045044, + "num_tokens": 165647768.0, + "step": 4261 + }, + { + "epoch": 0.5421702073527541, + "grad_norm": 2.6298742294311523, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8728523254394531, + "num_tokens": 165689436.0, + "step": 4262 + }, + { + "epoch": 0.5422974176313446, + "grad_norm": 2.546710252761841, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8740955591201782, + "num_tokens": 165729419.0, + "step": 4263 + }, + { + "epoch": 0.5424246279099352, + "grad_norm": 2.6274430751800537, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8856839537620544, + "num_tokens": 165768216.0, + "step": 4264 + }, + { + "epoch": 0.5425518381885256, + "grad_norm": 2.567800521850586, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8810979127883911, + "num_tokens": 165807257.0, + "step": 4265 + }, + { + "epoch": 0.5426790484671161, + "grad_norm": 2.693525791168213, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8731536269187927, + "num_tokens": 165846993.0, + "step": 4266 + }, + { + "epoch": 0.5428062587457066, + "grad_norm": 2.7982568740844727, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8817546367645264, + "num_tokens": 165882538.0, + "step": 4267 + }, + { + "epoch": 0.5429334690242972, + "grad_norm": 2.7144601345062256, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.8656903505325317, + "num_tokens": 165921107.0, + "step": 4268 + }, + { + "epoch": 0.5430606793028877, + "grad_norm": 2.8144736289978027, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8637388944625854, + "num_tokens": 165960044.0, + "step": 4269 + }, + { + "epoch": 0.5431878895814782, + "grad_norm": 2.6742441654205322, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.88169264793396, + "num_tokens": 165993983.0, + "step": 4270 + }, + { + "epoch": 0.5433150998600687, + "grad_norm": 2.9071340560913086, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8860933780670166, + "num_tokens": 166025042.0, + "step": 4271 + }, + { + "epoch": 0.5434423101386592, + "grad_norm": 2.6272425651550293, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8812763690948486, + "num_tokens": 166064677.0, + "step": 4272 + }, + { + "epoch": 0.5435695204172497, + "grad_norm": 2.798940420150757, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.872832179069519, + "num_tokens": 166102100.0, + "step": 4273 + }, + { + "epoch": 0.5436967306958402, + "grad_norm": 2.969726324081421, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8710341453552246, + "num_tokens": 166132800.0, + "step": 4274 + }, + { + "epoch": 0.5438239409744308, + "grad_norm": 2.675727128982544, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8697104454040527, + "num_tokens": 166174411.0, + "step": 4275 + }, + { + "epoch": 0.5439511512530213, + "grad_norm": 2.597606897354126, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.869705855846405, + "num_tokens": 166217165.0, + "step": 4276 + }, + { + "epoch": 0.5440783615316117, + "grad_norm": 2.447629928588867, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8763127326965332, + "num_tokens": 166260941.0, + "step": 4277 + }, + { + "epoch": 0.5442055718102022, + "grad_norm": 2.5877468585968018, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.874306321144104, + "num_tokens": 166302474.0, + "step": 4278 + }, + { + "epoch": 0.5443327820887928, + "grad_norm": 2.745689630508423, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8711984157562256, + "num_tokens": 166341727.0, + "step": 4279 + }, + { + "epoch": 0.5444599923673833, + "grad_norm": 2.6533539295196533, + "learning_rate": 1e-06, + "loss": 0.3826, + "mean_token_accuracy": 0.865609884262085, + "num_tokens": 166380898.0, + "step": 4280 + }, + { + "epoch": 0.5445872026459738, + "grad_norm": 2.7281785011291504, + "learning_rate": 1e-06, + "loss": 0.4065, + "mean_token_accuracy": 0.8583367466926575, + "num_tokens": 166423070.0, + "step": 4281 + }, + { + "epoch": 0.5447144129245644, + "grad_norm": 3.221092462539673, + "learning_rate": 1e-06, + "loss": 0.4367, + "mean_token_accuracy": 0.8515137434005737, + "num_tokens": 166457704.0, + "step": 4282 + }, + { + "epoch": 0.5448416232031548, + "grad_norm": 2.584603786468506, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8650591373443604, + "num_tokens": 166502861.0, + "step": 4283 + }, + { + "epoch": 0.5449688334817453, + "grad_norm": 2.7883927822113037, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8764129877090454, + "num_tokens": 166537271.0, + "step": 4284 + }, + { + "epoch": 0.5450960437603358, + "grad_norm": 2.7638673782348633, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8783624172210693, + "num_tokens": 166571784.0, + "step": 4285 + }, + { + "epoch": 0.5452232540389264, + "grad_norm": 2.80627703666687, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8835349082946777, + "num_tokens": 166603902.0, + "step": 4286 + }, + { + "epoch": 0.5453504643175169, + "grad_norm": 2.610907554626465, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8884202241897583, + "num_tokens": 166643088.0, + "step": 4287 + }, + { + "epoch": 0.5454776745961074, + "grad_norm": 2.797612190246582, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8690086603164673, + "num_tokens": 166679514.0, + "step": 4288 + }, + { + "epoch": 0.5456048848746978, + "grad_norm": 2.6134510040283203, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8815481066703796, + "num_tokens": 166720700.0, + "step": 4289 + }, + { + "epoch": 0.5457320951532884, + "grad_norm": 2.8227179050445557, + "learning_rate": 1e-06, + "loss": 0.3942, + "mean_token_accuracy": 0.8623859882354736, + "num_tokens": 166758584.0, + "step": 4290 + }, + { + "epoch": 0.5458593054318789, + "grad_norm": 2.801731824874878, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8762415051460266, + "num_tokens": 166793408.0, + "step": 4291 + }, + { + "epoch": 0.5459865157104694, + "grad_norm": 2.5804102420806885, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8739584684371948, + "num_tokens": 166832756.0, + "step": 4292 + }, + { + "epoch": 0.5461137259890599, + "grad_norm": 2.7658820152282715, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8888969421386719, + "num_tokens": 166866720.0, + "step": 4293 + }, + { + "epoch": 0.5462409362676505, + "grad_norm": 2.882026195526123, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8607078194618225, + "num_tokens": 166900834.0, + "step": 4294 + }, + { + "epoch": 0.5463681465462409, + "grad_norm": 2.801931142807007, + "learning_rate": 1e-06, + "loss": 0.4014, + "mean_token_accuracy": 0.8613536953926086, + "num_tokens": 166940781.0, + "step": 4295 + }, + { + "epoch": 0.5464953568248314, + "grad_norm": 2.5723211765289307, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8768743276596069, + "num_tokens": 166977635.0, + "step": 4296 + }, + { + "epoch": 0.5466225671034219, + "grad_norm": 2.751336097717285, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8709719181060791, + "num_tokens": 167013733.0, + "step": 4297 + }, + { + "epoch": 0.5467497773820125, + "grad_norm": 2.7752654552459717, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8710883259773254, + "num_tokens": 167051709.0, + "step": 4298 + }, + { + "epoch": 0.546876987660603, + "grad_norm": 2.7597312927246094, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8712749481201172, + "num_tokens": 167088256.0, + "step": 4299 + }, + { + "epoch": 0.5470041979391935, + "grad_norm": 2.5577406883239746, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8722293972969055, + "num_tokens": 167131101.0, + "step": 4300 + }, + { + "epoch": 0.5471314082177839, + "grad_norm": 2.683467388153076, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.87548828125, + "num_tokens": 167167297.0, + "step": 4301 + }, + { + "epoch": 0.5472586184963745, + "grad_norm": 2.5811941623687744, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8791775703430176, + "num_tokens": 167204998.0, + "step": 4302 + }, + { + "epoch": 0.547385828774965, + "grad_norm": 2.4422221183776855, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8846023082733154, + "num_tokens": 167243577.0, + "step": 4303 + }, + { + "epoch": 0.5475130390535555, + "grad_norm": 2.87292218208313, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8719140887260437, + "num_tokens": 167275746.0, + "step": 4304 + }, + { + "epoch": 0.5476402493321461, + "grad_norm": 2.72745680809021, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.874354362487793, + "num_tokens": 167312063.0, + "step": 4305 + }, + { + "epoch": 0.5477674596107366, + "grad_norm": 2.5538489818573, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8709074854850769, + "num_tokens": 167351786.0, + "step": 4306 + }, + { + "epoch": 0.5478946698893271, + "grad_norm": 2.614445686340332, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8672848343849182, + "num_tokens": 167392906.0, + "step": 4307 + }, + { + "epoch": 0.5480218801679175, + "grad_norm": 2.7517879009246826, + "learning_rate": 1e-06, + "loss": 0.3812, + "mean_token_accuracy": 0.8679002523422241, + "num_tokens": 167429417.0, + "step": 4308 + }, + { + "epoch": 0.5481490904465081, + "grad_norm": 2.617992401123047, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8805952668190002, + "num_tokens": 167467845.0, + "step": 4309 + }, + { + "epoch": 0.5482763007250986, + "grad_norm": 2.2655136585235596, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8940823674201965, + "num_tokens": 167514436.0, + "step": 4310 + }, + { + "epoch": 0.5484035110036891, + "grad_norm": 2.7033779621124268, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8739994168281555, + "num_tokens": 167554121.0, + "step": 4311 + }, + { + "epoch": 0.5485307212822796, + "grad_norm": 2.6565680503845215, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.8696348071098328, + "num_tokens": 167594048.0, + "step": 4312 + }, + { + "epoch": 0.5486579315608702, + "grad_norm": 2.5868852138519287, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8781328201293945, + "num_tokens": 167634727.0, + "step": 4313 + }, + { + "epoch": 0.5487851418394606, + "grad_norm": 2.653341054916382, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8683299422264099, + "num_tokens": 167673554.0, + "step": 4314 + }, + { + "epoch": 0.5489123521180511, + "grad_norm": 2.894420862197876, + "learning_rate": 1e-06, + "loss": 0.3959, + "mean_token_accuracy": 0.8619147539138794, + "num_tokens": 167709610.0, + "step": 4315 + }, + { + "epoch": 0.5490395623966416, + "grad_norm": 2.7869203090667725, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8740401268005371, + "num_tokens": 167748502.0, + "step": 4316 + }, + { + "epoch": 0.5491667726752322, + "grad_norm": 2.665288209915161, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8827922344207764, + "num_tokens": 167783717.0, + "step": 4317 + }, + { + "epoch": 0.5492939829538227, + "grad_norm": 2.501032590866089, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8790469169616699, + "num_tokens": 167830760.0, + "step": 4318 + }, + { + "epoch": 0.5494211932324132, + "grad_norm": 2.468733787536621, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8774514198303223, + "num_tokens": 167873220.0, + "step": 4319 + }, + { + "epoch": 0.5495484035110036, + "grad_norm": 2.5242321491241455, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8684351444244385, + "num_tokens": 167916609.0, + "step": 4320 + }, + { + "epoch": 0.5496756137895942, + "grad_norm": 2.7683064937591553, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8726203441619873, + "num_tokens": 167953060.0, + "step": 4321 + }, + { + "epoch": 0.5498028240681847, + "grad_norm": 2.593822479248047, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8685663938522339, + "num_tokens": 167994984.0, + "step": 4322 + }, + { + "epoch": 0.5499300343467752, + "grad_norm": 2.650127649307251, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8739701509475708, + "num_tokens": 168034040.0, + "step": 4323 + }, + { + "epoch": 0.5500572446253658, + "grad_norm": 2.589998245239258, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8753688335418701, + "num_tokens": 168076532.0, + "step": 4324 + }, + { + "epoch": 0.5501844549039563, + "grad_norm": 2.7753498554229736, + "learning_rate": 1e-06, + "loss": 0.4689, + "mean_token_accuracy": 0.8413863182067871, + "num_tokens": 168117317.0, + "step": 4325 + }, + { + "epoch": 0.5503116651825467, + "grad_norm": 2.476961851119995, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8707413673400879, + "num_tokens": 168160166.0, + "step": 4326 + }, + { + "epoch": 0.5504388754611372, + "grad_norm": 2.8111467361450195, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8739866018295288, + "num_tokens": 168192985.0, + "step": 4327 + }, + { + "epoch": 0.5505660857397278, + "grad_norm": 2.8293497562408447, + "learning_rate": 1e-06, + "loss": 0.4262, + "mean_token_accuracy": 0.854229211807251, + "num_tokens": 168235231.0, + "step": 4328 + }, + { + "epoch": 0.5506932960183183, + "grad_norm": 2.9041242599487305, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8819147348403931, + "num_tokens": 168265986.0, + "step": 4329 + }, + { + "epoch": 0.5508205062969088, + "grad_norm": 3.265939474105835, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8680312037467957, + "num_tokens": 168296834.0, + "step": 4330 + }, + { + "epoch": 0.5509477165754993, + "grad_norm": 2.696094274520874, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8743302822113037, + "num_tokens": 168336421.0, + "step": 4331 + }, + { + "epoch": 0.5510749268540898, + "grad_norm": 2.66811203956604, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8739035129547119, + "num_tokens": 168376037.0, + "step": 4332 + }, + { + "epoch": 0.5512021371326803, + "grad_norm": 2.4537885189056396, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8837093114852905, + "num_tokens": 168419191.0, + "step": 4333 + }, + { + "epoch": 0.5513293474112708, + "grad_norm": 2.6332156658172607, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8776804208755493, + "num_tokens": 168457816.0, + "step": 4334 + }, + { + "epoch": 0.5514565576898613, + "grad_norm": 2.7348482608795166, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8715519905090332, + "num_tokens": 168495780.0, + "step": 4335 + }, + { + "epoch": 0.5515837679684519, + "grad_norm": 2.434194803237915, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8898727297782898, + "num_tokens": 168537532.0, + "step": 4336 + }, + { + "epoch": 0.5517109782470424, + "grad_norm": 2.9057881832122803, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.875104546546936, + "num_tokens": 168570103.0, + "step": 4337 + }, + { + "epoch": 0.5518381885256328, + "grad_norm": 2.8538997173309326, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8750320672988892, + "num_tokens": 168600781.0, + "step": 4338 + }, + { + "epoch": 0.5519653988042234, + "grad_norm": 2.557372570037842, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8784447908401489, + "num_tokens": 168639873.0, + "step": 4339 + }, + { + "epoch": 0.5520926090828139, + "grad_norm": 2.6638123989105225, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8765925765037537, + "num_tokens": 168678936.0, + "step": 4340 + }, + { + "epoch": 0.5522198193614044, + "grad_norm": 2.485147714614868, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8729991912841797, + "num_tokens": 168723435.0, + "step": 4341 + }, + { + "epoch": 0.5523470296399949, + "grad_norm": 2.7427282333374023, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8747411966323853, + "num_tokens": 168761552.0, + "step": 4342 + }, + { + "epoch": 0.5524742399185855, + "grad_norm": 2.4535911083221436, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8817542791366577, + "num_tokens": 168804084.0, + "step": 4343 + }, + { + "epoch": 0.5526014501971759, + "grad_norm": 2.601284980773926, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8732752799987793, + "num_tokens": 168846043.0, + "step": 4344 + }, + { + "epoch": 0.5527286604757664, + "grad_norm": 2.866014242172241, + "learning_rate": 1e-06, + "loss": 0.3988, + "mean_token_accuracy": 0.8609637022018433, + "num_tokens": 168882793.0, + "step": 4345 + }, + { + "epoch": 0.5528558707543569, + "grad_norm": 2.491666555404663, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8733847141265869, + "num_tokens": 168928600.0, + "step": 4346 + }, + { + "epoch": 0.5529830810329475, + "grad_norm": 2.4955337047576904, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8901054859161377, + "num_tokens": 168969925.0, + "step": 4347 + }, + { + "epoch": 0.553110291311538, + "grad_norm": 2.5269834995269775, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8808547258377075, + "num_tokens": 169008525.0, + "step": 4348 + }, + { + "epoch": 0.5532375015901285, + "grad_norm": 2.92829966545105, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8744131326675415, + "num_tokens": 169043007.0, + "step": 4349 + }, + { + "epoch": 0.5533647118687189, + "grad_norm": 2.595273733139038, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8807123899459839, + "num_tokens": 169081860.0, + "step": 4350 + }, + { + "epoch": 0.5534919221473095, + "grad_norm": 2.4677677154541016, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8891430497169495, + "num_tokens": 169119895.0, + "step": 4351 + }, + { + "epoch": 0.5536191324259, + "grad_norm": 2.5746467113494873, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8851958513259888, + "num_tokens": 169161474.0, + "step": 4352 + }, + { + "epoch": 0.5537463427044905, + "grad_norm": 2.8065590858459473, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8728610277175903, + "num_tokens": 169197152.0, + "step": 4353 + }, + { + "epoch": 0.553873552983081, + "grad_norm": 2.640207052230835, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.8718400001525879, + "num_tokens": 169241326.0, + "step": 4354 + }, + { + "epoch": 0.5540007632616716, + "grad_norm": 2.7448713779449463, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8729705810546875, + "num_tokens": 169276982.0, + "step": 4355 + }, + { + "epoch": 0.554127973540262, + "grad_norm": 2.759845733642578, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8636044859886169, + "num_tokens": 169313828.0, + "step": 4356 + }, + { + "epoch": 0.5542551838188525, + "grad_norm": 2.6605265140533447, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8696211576461792, + "num_tokens": 169351928.0, + "step": 4357 + }, + { + "epoch": 0.554382394097443, + "grad_norm": 2.5660033226013184, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.889849066734314, + "num_tokens": 169387958.0, + "step": 4358 + }, + { + "epoch": 0.5545096043760336, + "grad_norm": 3.046875238418579, + "learning_rate": 1e-06, + "loss": 0.4238, + "mean_token_accuracy": 0.8543691635131836, + "num_tokens": 169422285.0, + "step": 4359 + }, + { + "epoch": 0.5546368146546241, + "grad_norm": 2.61089825630188, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8721213936805725, + "num_tokens": 169459751.0, + "step": 4360 + }, + { + "epoch": 0.5547640249332146, + "grad_norm": 2.511742353439331, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8671876192092896, + "num_tokens": 169503318.0, + "step": 4361 + }, + { + "epoch": 0.5548912352118052, + "grad_norm": 2.4973888397216797, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8802527189254761, + "num_tokens": 169541971.0, + "step": 4362 + }, + { + "epoch": 0.5550184454903956, + "grad_norm": 2.54193115234375, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8695945143699646, + "num_tokens": 169585831.0, + "step": 4363 + }, + { + "epoch": 0.5551456557689861, + "grad_norm": 2.579838514328003, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8888134956359863, + "num_tokens": 169624443.0, + "step": 4364 + }, + { + "epoch": 0.5552728660475766, + "grad_norm": 2.515767812728882, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8717659711837769, + "num_tokens": 169666212.0, + "step": 4365 + }, + { + "epoch": 0.5554000763261672, + "grad_norm": 2.3931446075439453, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8851296901702881, + "num_tokens": 169708562.0, + "step": 4366 + }, + { + "epoch": 0.5555272866047577, + "grad_norm": 2.8896894454956055, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8780068755149841, + "num_tokens": 169740572.0, + "step": 4367 + }, + { + "epoch": 0.5556544968833482, + "grad_norm": 2.608736753463745, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8705132007598877, + "num_tokens": 169781058.0, + "step": 4368 + }, + { + "epoch": 0.5557817071619386, + "grad_norm": 2.461709499359131, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8834013938903809, + "num_tokens": 169822536.0, + "step": 4369 + }, + { + "epoch": 0.5559089174405292, + "grad_norm": 2.889133930206299, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.8655576705932617, + "num_tokens": 169860635.0, + "step": 4370 + }, + { + "epoch": 0.5560361277191197, + "grad_norm": 2.623208522796631, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8750796318054199, + "num_tokens": 169898423.0, + "step": 4371 + }, + { + "epoch": 0.5561633379977102, + "grad_norm": 2.771423578262329, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8728485703468323, + "num_tokens": 169934379.0, + "step": 4372 + }, + { + "epoch": 0.5562905482763008, + "grad_norm": 2.871396780014038, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8790310621261597, + "num_tokens": 169966952.0, + "step": 4373 + }, + { + "epoch": 0.5564177585548913, + "grad_norm": 2.5553455352783203, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8865174055099487, + "num_tokens": 170004342.0, + "step": 4374 + }, + { + "epoch": 0.5565449688334817, + "grad_norm": 2.459536075592041, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8845469355583191, + "num_tokens": 170047143.0, + "step": 4375 + }, + { + "epoch": 0.5566721791120722, + "grad_norm": 2.774885654449463, + "learning_rate": 1e-06, + "loss": 0.4021, + "mean_token_accuracy": 0.8651156425476074, + "num_tokens": 170086838.0, + "step": 4376 + }, + { + "epoch": 0.5567993893906628, + "grad_norm": 2.644357681274414, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8872609734535217, + "num_tokens": 170121972.0, + "step": 4377 + }, + { + "epoch": 0.5569265996692533, + "grad_norm": 3.0157923698425293, + "learning_rate": 1e-06, + "loss": 0.425, + "mean_token_accuracy": 0.8582946062088013, + "num_tokens": 170158665.0, + "step": 4378 + }, + { + "epoch": 0.5570538099478438, + "grad_norm": 2.8956081867218018, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.867830753326416, + "num_tokens": 170192277.0, + "step": 4379 + }, + { + "epoch": 0.5571810202264343, + "grad_norm": 2.6495141983032227, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8715822696685791, + "num_tokens": 170229229.0, + "step": 4380 + }, + { + "epoch": 0.5573082305050248, + "grad_norm": 2.535463571548462, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8703214526176453, + "num_tokens": 170271850.0, + "step": 4381 + }, + { + "epoch": 0.5574354407836153, + "grad_norm": 2.7357091903686523, + "learning_rate": 1e-06, + "loss": 0.4201, + "mean_token_accuracy": 0.859503984451294, + "num_tokens": 170314574.0, + "step": 4382 + }, + { + "epoch": 0.5575626510622058, + "grad_norm": 2.8273749351501465, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8643786311149597, + "num_tokens": 170349376.0, + "step": 4383 + }, + { + "epoch": 0.5576898613407963, + "grad_norm": 2.7915875911712646, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8703693151473999, + "num_tokens": 170384953.0, + "step": 4384 + }, + { + "epoch": 0.5578170716193869, + "grad_norm": 2.7409474849700928, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8722779750823975, + "num_tokens": 170421300.0, + "step": 4385 + }, + { + "epoch": 0.5579442818979774, + "grad_norm": 2.8584187030792236, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8836811780929565, + "num_tokens": 170455870.0, + "step": 4386 + }, + { + "epoch": 0.5580714921765678, + "grad_norm": 2.735036611557007, + "learning_rate": 1e-06, + "loss": 0.3988, + "mean_token_accuracy": 0.8624069690704346, + "num_tokens": 170492854.0, + "step": 4387 + }, + { + "epoch": 0.5581987024551583, + "grad_norm": 2.403826951980591, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8893882036209106, + "num_tokens": 170534511.0, + "step": 4388 + }, + { + "epoch": 0.5583259127337489, + "grad_norm": 2.6335644721984863, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8711926341056824, + "num_tokens": 170575647.0, + "step": 4389 + }, + { + "epoch": 0.5584531230123394, + "grad_norm": 3.6308021545410156, + "learning_rate": 1e-06, + "loss": 0.4233, + "mean_token_accuracy": 0.8552249670028687, + "num_tokens": 170606805.0, + "step": 4390 + }, + { + "epoch": 0.5585803332909299, + "grad_norm": 2.639878988265991, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8766407370567322, + "num_tokens": 170646371.0, + "step": 4391 + }, + { + "epoch": 0.5587075435695205, + "grad_norm": 2.7198100090026855, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8685219287872314, + "num_tokens": 170685307.0, + "step": 4392 + }, + { + "epoch": 0.5588347538481109, + "grad_norm": 2.576195478439331, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8673374652862549, + "num_tokens": 170727104.0, + "step": 4393 + }, + { + "epoch": 0.5589619641267014, + "grad_norm": 2.732279062271118, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8740918040275574, + "num_tokens": 170762673.0, + "step": 4394 + }, + { + "epoch": 0.5590891744052919, + "grad_norm": 2.661580801010132, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.874272882938385, + "num_tokens": 170800010.0, + "step": 4395 + }, + { + "epoch": 0.5592163846838825, + "grad_norm": 2.4517598152160645, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8786730766296387, + "num_tokens": 170840674.0, + "step": 4396 + }, + { + "epoch": 0.559343594962473, + "grad_norm": 2.5799756050109863, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8764382004737854, + "num_tokens": 170878276.0, + "step": 4397 + }, + { + "epoch": 0.5594708052410635, + "grad_norm": 2.7956321239471436, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8698680400848389, + "num_tokens": 170917198.0, + "step": 4398 + }, + { + "epoch": 0.5595980155196539, + "grad_norm": 2.681666851043701, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8760878443717957, + "num_tokens": 170953724.0, + "step": 4399 + }, + { + "epoch": 0.5597252257982445, + "grad_norm": 2.6492061614990234, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.880364716053009, + "num_tokens": 170993688.0, + "step": 4400 + }, + { + "epoch": 0.559852436076835, + "grad_norm": 2.48520565032959, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8875318765640259, + "num_tokens": 171034224.0, + "step": 4401 + }, + { + "epoch": 0.5599796463554255, + "grad_norm": 2.6511523723602295, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8864156007766724, + "num_tokens": 171069697.0, + "step": 4402 + }, + { + "epoch": 0.560106856634016, + "grad_norm": 2.676384687423706, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.8643093109130859, + "num_tokens": 171113042.0, + "step": 4403 + }, + { + "epoch": 0.5602340669126066, + "grad_norm": 2.6228187084198, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8734344840049744, + "num_tokens": 171152335.0, + "step": 4404 + }, + { + "epoch": 0.560361277191197, + "grad_norm": 2.544454336166382, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8787945508956909, + "num_tokens": 171193765.0, + "step": 4405 + }, + { + "epoch": 0.5604884874697875, + "grad_norm": 2.692842960357666, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8853601217269897, + "num_tokens": 171228126.0, + "step": 4406 + }, + { + "epoch": 0.560615697748378, + "grad_norm": 2.632402181625366, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8848205804824829, + "num_tokens": 171261078.0, + "step": 4407 + }, + { + "epoch": 0.5607429080269686, + "grad_norm": 2.9509153366088867, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8676367998123169, + "num_tokens": 171295840.0, + "step": 4408 + }, + { + "epoch": 0.5608701183055591, + "grad_norm": 2.7012720108032227, + "learning_rate": 1e-06, + "loss": 0.3923, + "mean_token_accuracy": 0.8670039772987366, + "num_tokens": 171338532.0, + "step": 4409 + }, + { + "epoch": 0.5609973285841496, + "grad_norm": 2.6309890747070312, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8812788128852844, + "num_tokens": 171373647.0, + "step": 4410 + }, + { + "epoch": 0.5611245388627402, + "grad_norm": 2.553525686264038, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8826684951782227, + "num_tokens": 171412489.0, + "step": 4411 + }, + { + "epoch": 0.5612517491413306, + "grad_norm": 2.4261746406555176, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8907972574234009, + "num_tokens": 171454290.0, + "step": 4412 + }, + { + "epoch": 0.5613789594199211, + "grad_norm": 2.761054754257202, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8706101775169373, + "num_tokens": 171496622.0, + "step": 4413 + }, + { + "epoch": 0.5615061696985116, + "grad_norm": 2.7092039585113525, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8800599575042725, + "num_tokens": 171531497.0, + "step": 4414 + }, + { + "epoch": 0.5616333799771022, + "grad_norm": 2.7997190952301025, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8699926137924194, + "num_tokens": 171566232.0, + "step": 4415 + }, + { + "epoch": 0.5617605902556927, + "grad_norm": 2.8543200492858887, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8707352876663208, + "num_tokens": 171600713.0, + "step": 4416 + }, + { + "epoch": 0.5618878005342832, + "grad_norm": 2.448580265045166, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8905117511749268, + "num_tokens": 171641921.0, + "step": 4417 + }, + { + "epoch": 0.5620150108128736, + "grad_norm": 2.7951838970184326, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8870631456375122, + "num_tokens": 171674227.0, + "step": 4418 + }, + { + "epoch": 0.5621422210914642, + "grad_norm": 2.6055080890655518, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8898596167564392, + "num_tokens": 171708332.0, + "step": 4419 + }, + { + "epoch": 0.5622694313700547, + "grad_norm": 2.7399110794067383, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8804054260253906, + "num_tokens": 171744014.0, + "step": 4420 + }, + { + "epoch": 0.5623966416486452, + "grad_norm": 2.8613405227661133, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8782139420509338, + "num_tokens": 171776606.0, + "step": 4421 + }, + { + "epoch": 0.5625238519272358, + "grad_norm": 2.3484296798706055, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8892015218734741, + "num_tokens": 171819768.0, + "step": 4422 + }, + { + "epoch": 0.5626510622058263, + "grad_norm": 2.718299627304077, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8751513361930847, + "num_tokens": 171857031.0, + "step": 4423 + }, + { + "epoch": 0.5627782724844167, + "grad_norm": 2.5656144618988037, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.8716626763343811, + "num_tokens": 171899532.0, + "step": 4424 + }, + { + "epoch": 0.5629054827630072, + "grad_norm": 2.5617690086364746, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8781336545944214, + "num_tokens": 171940751.0, + "step": 4425 + }, + { + "epoch": 0.5630326930415978, + "grad_norm": 2.8553617000579834, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8699450492858887, + "num_tokens": 171973824.0, + "step": 4426 + }, + { + "epoch": 0.5631599033201883, + "grad_norm": 2.6836912631988525, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8672232627868652, + "num_tokens": 172013307.0, + "step": 4427 + }, + { + "epoch": 0.5632871135987788, + "grad_norm": 2.8470375537872314, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8646016716957092, + "num_tokens": 172047609.0, + "step": 4428 + }, + { + "epoch": 0.5634143238773693, + "grad_norm": 2.55702543258667, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8771664500236511, + "num_tokens": 172088160.0, + "step": 4429 + }, + { + "epoch": 0.5635415341559598, + "grad_norm": 2.659921407699585, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8761775493621826, + "num_tokens": 172127169.0, + "step": 4430 + }, + { + "epoch": 0.5636687444345503, + "grad_norm": 2.5438590049743652, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.879836916923523, + "num_tokens": 172171050.0, + "step": 4431 + }, + { + "epoch": 0.5637959547131408, + "grad_norm": 2.3430161476135254, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8887510299682617, + "num_tokens": 172212570.0, + "step": 4432 + }, + { + "epoch": 0.5639231649917313, + "grad_norm": 2.515594720840454, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8843735456466675, + "num_tokens": 172256746.0, + "step": 4433 + }, + { + "epoch": 0.5640503752703219, + "grad_norm": 2.4672300815582275, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8871755599975586, + "num_tokens": 172296942.0, + "step": 4434 + }, + { + "epoch": 0.5641775855489124, + "grad_norm": 2.6831343173980713, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8734792470932007, + "num_tokens": 172335323.0, + "step": 4435 + }, + { + "epoch": 0.5643047958275028, + "grad_norm": 2.5614166259765625, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8723506927490234, + "num_tokens": 172381095.0, + "step": 4436 + }, + { + "epoch": 0.5644320061060933, + "grad_norm": 2.7388737201690674, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8781626224517822, + "num_tokens": 172420997.0, + "step": 4437 + }, + { + "epoch": 0.5645592163846839, + "grad_norm": 2.8432047367095947, + "learning_rate": 1e-06, + "loss": 0.4203, + "mean_token_accuracy": 0.8545160293579102, + "num_tokens": 172458945.0, + "step": 4438 + }, + { + "epoch": 0.5646864266632744, + "grad_norm": 2.411259174346924, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8884230852127075, + "num_tokens": 172499579.0, + "step": 4439 + }, + { + "epoch": 0.5648136369418649, + "grad_norm": 2.9593114852905273, + "learning_rate": 1e-06, + "loss": 0.4161, + "mean_token_accuracy": 0.8544729948043823, + "num_tokens": 172539837.0, + "step": 4440 + }, + { + "epoch": 0.5649408472204555, + "grad_norm": 2.4672646522521973, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8773981928825378, + "num_tokens": 172582868.0, + "step": 4441 + }, + { + "epoch": 0.5650680574990459, + "grad_norm": 2.5804007053375244, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8869954943656921, + "num_tokens": 172618394.0, + "step": 4442 + }, + { + "epoch": 0.5651952677776364, + "grad_norm": 2.541208028793335, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8850614428520203, + "num_tokens": 172660179.0, + "step": 4443 + }, + { + "epoch": 0.5653224780562269, + "grad_norm": 2.804344892501831, + "learning_rate": 1e-06, + "loss": 0.3961, + "mean_token_accuracy": 0.8611189126968384, + "num_tokens": 172698843.0, + "step": 4444 + }, + { + "epoch": 0.5654496883348175, + "grad_norm": 2.820444107055664, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8826338052749634, + "num_tokens": 172730159.0, + "step": 4445 + }, + { + "epoch": 0.565576898613408, + "grad_norm": 2.661525011062622, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8725025653839111, + "num_tokens": 172767638.0, + "step": 4446 + }, + { + "epoch": 0.5657041088919985, + "grad_norm": 2.419759750366211, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8702616691589355, + "num_tokens": 172813846.0, + "step": 4447 + }, + { + "epoch": 0.5658313191705889, + "grad_norm": 2.4446985721588135, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8874379992485046, + "num_tokens": 172855439.0, + "step": 4448 + }, + { + "epoch": 0.5659585294491795, + "grad_norm": 2.707308292388916, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.876096248626709, + "num_tokens": 172892393.0, + "step": 4449 + }, + { + "epoch": 0.56608573972777, + "grad_norm": 2.7819812297821045, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8699476718902588, + "num_tokens": 172928959.0, + "step": 4450 + }, + { + "epoch": 0.5662129500063605, + "grad_norm": 2.542480230331421, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8807374238967896, + "num_tokens": 172969768.0, + "step": 4451 + }, + { + "epoch": 0.566340160284951, + "grad_norm": 2.5828826427459717, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8810718059539795, + "num_tokens": 173005956.0, + "step": 4452 + }, + { + "epoch": 0.5664673705635416, + "grad_norm": 2.5869832038879395, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8734840154647827, + "num_tokens": 173046017.0, + "step": 4453 + }, + { + "epoch": 0.566594580842132, + "grad_norm": 2.6452367305755615, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8819577097892761, + "num_tokens": 173082445.0, + "step": 4454 + }, + { + "epoch": 0.5667217911207225, + "grad_norm": 2.703848123550415, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8732175827026367, + "num_tokens": 173125720.0, + "step": 4455 + }, + { + "epoch": 0.566849001399313, + "grad_norm": 2.783081293106079, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8667569160461426, + "num_tokens": 173163551.0, + "step": 4456 + }, + { + "epoch": 0.5669762116779036, + "grad_norm": 2.6542305946350098, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8744591474533081, + "num_tokens": 173204294.0, + "step": 4457 + }, + { + "epoch": 0.5671034219564941, + "grad_norm": 2.735516309738159, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8768466114997864, + "num_tokens": 173240183.0, + "step": 4458 + }, + { + "epoch": 0.5672306322350846, + "grad_norm": 2.7836990356445312, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8791223764419556, + "num_tokens": 173273618.0, + "step": 4459 + }, + { + "epoch": 0.5673578425136752, + "grad_norm": 2.3860924243927, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8830699324607849, + "num_tokens": 173315215.0, + "step": 4460 + }, + { + "epoch": 0.5674850527922656, + "grad_norm": 2.760103225708008, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8710207343101501, + "num_tokens": 173355354.0, + "step": 4461 + }, + { + "epoch": 0.5676122630708561, + "grad_norm": 2.8341152667999268, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8714813590049744, + "num_tokens": 173388396.0, + "step": 4462 + }, + { + "epoch": 0.5677394733494466, + "grad_norm": 2.5792765617370605, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8713140487670898, + "num_tokens": 173432036.0, + "step": 4463 + }, + { + "epoch": 0.5678666836280372, + "grad_norm": 2.4313621520996094, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8920591473579407, + "num_tokens": 173474443.0, + "step": 4464 + }, + { + "epoch": 0.5679938939066277, + "grad_norm": 2.827288866043091, + "learning_rate": 1e-06, + "loss": 0.3998, + "mean_token_accuracy": 0.8631830215454102, + "num_tokens": 173513017.0, + "step": 4465 + }, + { + "epoch": 0.5681211041852182, + "grad_norm": 2.7960596084594727, + "learning_rate": 1e-06, + "loss": 0.3847, + "mean_token_accuracy": 0.8641172647476196, + "num_tokens": 173551532.0, + "step": 4466 + }, + { + "epoch": 0.5682483144638086, + "grad_norm": 2.8936290740966797, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8678649067878723, + "num_tokens": 173585921.0, + "step": 4467 + }, + { + "epoch": 0.5683755247423992, + "grad_norm": 2.4614179134368896, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8794403076171875, + "num_tokens": 173628452.0, + "step": 4468 + }, + { + "epoch": 0.5685027350209897, + "grad_norm": 2.533447265625, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8812413215637207, + "num_tokens": 173670269.0, + "step": 4469 + }, + { + "epoch": 0.5686299452995802, + "grad_norm": 2.493419885635376, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8739005327224731, + "num_tokens": 173711507.0, + "step": 4470 + }, + { + "epoch": 0.5687571555781707, + "grad_norm": 2.908565044403076, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8759404420852661, + "num_tokens": 173742398.0, + "step": 4471 + }, + { + "epoch": 0.5688843658567613, + "grad_norm": 2.526073694229126, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.870697021484375, + "num_tokens": 173783615.0, + "step": 4472 + }, + { + "epoch": 0.5690115761353517, + "grad_norm": 2.5649116039276123, + "learning_rate": 1e-06, + "loss": 0.3957, + "mean_token_accuracy": 0.8649863600730896, + "num_tokens": 173829817.0, + "step": 4473 + }, + { + "epoch": 0.5691387864139422, + "grad_norm": 2.475271463394165, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8821281790733337, + "num_tokens": 173871451.0, + "step": 4474 + }, + { + "epoch": 0.5692659966925327, + "grad_norm": 2.5972812175750732, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.868278980255127, + "num_tokens": 173910923.0, + "step": 4475 + }, + { + "epoch": 0.5693932069711233, + "grad_norm": 2.389178514480591, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8892285823822021, + "num_tokens": 173952175.0, + "step": 4476 + }, + { + "epoch": 0.5695204172497138, + "grad_norm": 2.739995241165161, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8735940456390381, + "num_tokens": 173987688.0, + "step": 4477 + }, + { + "epoch": 0.5696476275283043, + "grad_norm": 2.6664299964904785, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.872026801109314, + "num_tokens": 174027123.0, + "step": 4478 + }, + { + "epoch": 0.5697748378068948, + "grad_norm": 2.809772253036499, + "learning_rate": 1e-06, + "loss": 0.4532, + "mean_token_accuracy": 0.8483943939208984, + "num_tokens": 174068728.0, + "step": 4479 + }, + { + "epoch": 0.5699020480854853, + "grad_norm": 2.628330707550049, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8674739599227905, + "num_tokens": 174106500.0, + "step": 4480 + }, + { + "epoch": 0.5700292583640758, + "grad_norm": 2.757744073867798, + "learning_rate": 1e-06, + "loss": 0.4332, + "mean_token_accuracy": 0.8554710149765015, + "num_tokens": 174144944.0, + "step": 4481 + }, + { + "epoch": 0.5701564686426663, + "grad_norm": 2.9101402759552, + "learning_rate": 1e-06, + "loss": 0.4037, + "mean_token_accuracy": 0.8613088726997375, + "num_tokens": 174180264.0, + "step": 4482 + }, + { + "epoch": 0.5702836789212569, + "grad_norm": 2.473330020904541, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8921740651130676, + "num_tokens": 174218097.0, + "step": 4483 + }, + { + "epoch": 0.5704108891998474, + "grad_norm": 2.858569860458374, + "learning_rate": 1e-06, + "loss": 0.3998, + "mean_token_accuracy": 0.8615834712982178, + "num_tokens": 174256866.0, + "step": 4484 + }, + { + "epoch": 0.5705380994784378, + "grad_norm": 2.753901958465576, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8819447755813599, + "num_tokens": 174293645.0, + "step": 4485 + }, + { + "epoch": 0.5706653097570283, + "grad_norm": 2.8687407970428467, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8821815252304077, + "num_tokens": 174324141.0, + "step": 4486 + }, + { + "epoch": 0.5707925200356189, + "grad_norm": 2.413102626800537, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8780601024627686, + "num_tokens": 174367289.0, + "step": 4487 + }, + { + "epoch": 0.5709197303142094, + "grad_norm": 2.540877342224121, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8782558441162109, + "num_tokens": 174406781.0, + "step": 4488 + }, + { + "epoch": 0.5710469405927999, + "grad_norm": 2.698986530303955, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8754976987838745, + "num_tokens": 174445037.0, + "step": 4489 + }, + { + "epoch": 0.5711741508713905, + "grad_norm": 2.676283597946167, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8693411350250244, + "num_tokens": 174485590.0, + "step": 4490 + }, + { + "epoch": 0.5713013611499809, + "grad_norm": 2.880566120147705, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8783106803894043, + "num_tokens": 174516715.0, + "step": 4491 + }, + { + "epoch": 0.5714285714285714, + "grad_norm": 2.64430570602417, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8782786726951599, + "num_tokens": 174553309.0, + "step": 4492 + }, + { + "epoch": 0.5715557817071619, + "grad_norm": 2.6275222301483154, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8729107975959778, + "num_tokens": 174588501.0, + "step": 4493 + }, + { + "epoch": 0.5716829919857525, + "grad_norm": 2.8957536220550537, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8661588430404663, + "num_tokens": 174622499.0, + "step": 4494 + }, + { + "epoch": 0.571810202264343, + "grad_norm": 2.574922800064087, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8830341100692749, + "num_tokens": 174662385.0, + "step": 4495 + }, + { + "epoch": 0.5719374125429335, + "grad_norm": 2.829793930053711, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8694717884063721, + "num_tokens": 174698902.0, + "step": 4496 + }, + { + "epoch": 0.5720646228215239, + "grad_norm": 2.560659646987915, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.872929036617279, + "num_tokens": 174740698.0, + "step": 4497 + }, + { + "epoch": 0.5721918331001145, + "grad_norm": 2.6767871379852295, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8945835828781128, + "num_tokens": 174775697.0, + "step": 4498 + }, + { + "epoch": 0.572319043378705, + "grad_norm": 2.641354560852051, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8710024356842041, + "num_tokens": 174815744.0, + "step": 4499 + }, + { + "epoch": 0.5724462536572955, + "grad_norm": 2.5680501461029053, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8801767826080322, + "num_tokens": 174855613.0, + "step": 4500 + }, + { + "epoch": 0.572573463935886, + "grad_norm": 2.7219133377075195, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8818352818489075, + "num_tokens": 174893213.0, + "step": 4501 + }, + { + "epoch": 0.5727006742144766, + "grad_norm": 2.7588727474212646, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8744964003562927, + "num_tokens": 174930209.0, + "step": 4502 + }, + { + "epoch": 0.572827884493067, + "grad_norm": 2.664670705795288, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8711337447166443, + "num_tokens": 174972125.0, + "step": 4503 + }, + { + "epoch": 0.5729550947716575, + "grad_norm": 2.903701066970825, + "learning_rate": 1e-06, + "loss": 0.402, + "mean_token_accuracy": 0.8562213182449341, + "num_tokens": 175005383.0, + "step": 4504 + }, + { + "epoch": 0.573082305050248, + "grad_norm": 2.527501106262207, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8735320568084717, + "num_tokens": 175050091.0, + "step": 4505 + }, + { + "epoch": 0.5732095153288386, + "grad_norm": 2.751739263534546, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8670568466186523, + "num_tokens": 175087384.0, + "step": 4506 + }, + { + "epoch": 0.5733367256074291, + "grad_norm": 2.8195242881774902, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8770893812179565, + "num_tokens": 175123387.0, + "step": 4507 + }, + { + "epoch": 0.5734639358860196, + "grad_norm": 2.683513879776001, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8816171288490295, + "num_tokens": 175159264.0, + "step": 4508 + }, + { + "epoch": 0.5735911461646102, + "grad_norm": 2.920835018157959, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8783731460571289, + "num_tokens": 175191332.0, + "step": 4509 + }, + { + "epoch": 0.5737183564432006, + "grad_norm": 2.8031165599823, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8701866865158081, + "num_tokens": 175227356.0, + "step": 4510 + }, + { + "epoch": 0.5738455667217911, + "grad_norm": 2.4490671157836914, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8853250741958618, + "num_tokens": 175268514.0, + "step": 4511 + }, + { + "epoch": 0.5739727770003816, + "grad_norm": 2.689993143081665, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8801189661026001, + "num_tokens": 175304428.0, + "step": 4512 + }, + { + "epoch": 0.5740999872789722, + "grad_norm": 2.759805917739868, + "learning_rate": 1e-06, + "loss": 0.3896, + "mean_token_accuracy": 0.8652756214141846, + "num_tokens": 175343258.0, + "step": 4513 + }, + { + "epoch": 0.5742271975575627, + "grad_norm": 2.396946430206299, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8757538795471191, + "num_tokens": 175391961.0, + "step": 4514 + }, + { + "epoch": 0.5743544078361532, + "grad_norm": 2.568760633468628, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.871823251247406, + "num_tokens": 175436292.0, + "step": 4515 + }, + { + "epoch": 0.5744816181147436, + "grad_norm": 2.9537594318389893, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8753332495689392, + "num_tokens": 175470334.0, + "step": 4516 + }, + { + "epoch": 0.5746088283933342, + "grad_norm": 2.637437105178833, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8829742670059204, + "num_tokens": 175508520.0, + "step": 4517 + }, + { + "epoch": 0.5747360386719247, + "grad_norm": 2.5386576652526855, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8825066089630127, + "num_tokens": 175548041.0, + "step": 4518 + }, + { + "epoch": 0.5748632489505152, + "grad_norm": 2.5425562858581543, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8829176425933838, + "num_tokens": 175589257.0, + "step": 4519 + }, + { + "epoch": 0.5749904592291057, + "grad_norm": 2.555896759033203, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8800525665283203, + "num_tokens": 175631000.0, + "step": 4520 + }, + { + "epoch": 0.5751176695076963, + "grad_norm": 3.084965467453003, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8706127405166626, + "num_tokens": 175667626.0, + "step": 4521 + }, + { + "epoch": 0.5752448797862867, + "grad_norm": 2.6988027095794678, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8890249729156494, + "num_tokens": 175712680.0, + "step": 4522 + }, + { + "epoch": 0.5753720900648772, + "grad_norm": 2.7009270191192627, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.865166187286377, + "num_tokens": 175754157.0, + "step": 4523 + }, + { + "epoch": 0.5754993003434677, + "grad_norm": 2.4490296840667725, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8939102292060852, + "num_tokens": 175789023.0, + "step": 4524 + }, + { + "epoch": 0.5756265106220583, + "grad_norm": 2.739563465118408, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8735418319702148, + "num_tokens": 175829957.0, + "step": 4525 + }, + { + "epoch": 0.5757537209006488, + "grad_norm": 2.572740316390991, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8771125078201294, + "num_tokens": 175870520.0, + "step": 4526 + }, + { + "epoch": 0.5758809311792393, + "grad_norm": 2.628013849258423, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8864609599113464, + "num_tokens": 175906272.0, + "step": 4527 + }, + { + "epoch": 0.5760081414578297, + "grad_norm": 2.517268180847168, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8860503435134888, + "num_tokens": 175946979.0, + "step": 4528 + }, + { + "epoch": 0.5761353517364203, + "grad_norm": 2.6720402240753174, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8823161125183105, + "num_tokens": 175982471.0, + "step": 4529 + }, + { + "epoch": 0.5762625620150108, + "grad_norm": 2.6034228801727295, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.865422785282135, + "num_tokens": 176023229.0, + "step": 4530 + }, + { + "epoch": 0.5763897722936013, + "grad_norm": 2.749265193939209, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8791593909263611, + "num_tokens": 176058882.0, + "step": 4531 + }, + { + "epoch": 0.5765169825721919, + "grad_norm": 3.2479248046875, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.879359245300293, + "num_tokens": 176095652.0, + "step": 4532 + }, + { + "epoch": 0.5766441928507824, + "grad_norm": 2.5400211811065674, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8895957469940186, + "num_tokens": 176136355.0, + "step": 4533 + }, + { + "epoch": 0.5767714031293728, + "grad_norm": 2.7600131034851074, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8792150020599365, + "num_tokens": 176174230.0, + "step": 4534 + }, + { + "epoch": 0.5768986134079633, + "grad_norm": 2.8377060890197754, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8656200766563416, + "num_tokens": 176208424.0, + "step": 4535 + }, + { + "epoch": 0.5770258236865539, + "grad_norm": 2.4835057258605957, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.880422055721283, + "num_tokens": 176246830.0, + "step": 4536 + }, + { + "epoch": 0.5771530339651444, + "grad_norm": 2.4668328762054443, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9011438488960266, + "num_tokens": 176284706.0, + "step": 4537 + }, + { + "epoch": 0.5772802442437349, + "grad_norm": 2.8850998878479004, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.879425048828125, + "num_tokens": 176319951.0, + "step": 4538 + }, + { + "epoch": 0.5774074545223254, + "grad_norm": 2.873411178588867, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8721973896026611, + "num_tokens": 176359565.0, + "step": 4539 + }, + { + "epoch": 0.5775346648009159, + "grad_norm": 2.8080649375915527, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.87714022397995, + "num_tokens": 176394352.0, + "step": 4540 + }, + { + "epoch": 0.5776618750795064, + "grad_norm": 2.6198134422302246, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8751225471496582, + "num_tokens": 176436256.0, + "step": 4541 + }, + { + "epoch": 0.5777890853580969, + "grad_norm": 2.6306910514831543, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8717018365859985, + "num_tokens": 176475511.0, + "step": 4542 + }, + { + "epoch": 0.5779162956366874, + "grad_norm": 2.603078842163086, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8819664716720581, + "num_tokens": 176516170.0, + "step": 4543 + }, + { + "epoch": 0.578043505915278, + "grad_norm": 2.983616828918457, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8693212270736694, + "num_tokens": 176547448.0, + "step": 4544 + }, + { + "epoch": 0.5781707161938685, + "grad_norm": 2.4515295028686523, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8869783878326416, + "num_tokens": 176589744.0, + "step": 4545 + }, + { + "epoch": 0.5782979264724589, + "grad_norm": 2.620396852493286, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8715685606002808, + "num_tokens": 176627945.0, + "step": 4546 + }, + { + "epoch": 0.5784251367510495, + "grad_norm": 2.6653456687927246, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8728290796279907, + "num_tokens": 176666080.0, + "step": 4547 + }, + { + "epoch": 0.57855234702964, + "grad_norm": 2.5875375270843506, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8788415789604187, + "num_tokens": 176705914.0, + "step": 4548 + }, + { + "epoch": 0.5786795573082305, + "grad_norm": 2.577484130859375, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8941749334335327, + "num_tokens": 176744550.0, + "step": 4549 + }, + { + "epoch": 0.578806767586821, + "grad_norm": 2.672329902648926, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8883086442947388, + "num_tokens": 176777252.0, + "step": 4550 + }, + { + "epoch": 0.5789339778654116, + "grad_norm": 2.8443851470947266, + "learning_rate": 1e-06, + "loss": 0.3881, + "mean_token_accuracy": 0.8663235902786255, + "num_tokens": 176815371.0, + "step": 4551 + }, + { + "epoch": 0.579061188144002, + "grad_norm": 2.313676595687866, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8914147615432739, + "num_tokens": 176857429.0, + "step": 4552 + }, + { + "epoch": 0.5791883984225925, + "grad_norm": 2.5723369121551514, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8760660886764526, + "num_tokens": 176898552.0, + "step": 4553 + }, + { + "epoch": 0.579315608701183, + "grad_norm": 2.493551731109619, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8790972232818604, + "num_tokens": 176941918.0, + "step": 4554 + }, + { + "epoch": 0.5794428189797736, + "grad_norm": 2.8635406494140625, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8768564462661743, + "num_tokens": 176980773.0, + "step": 4555 + }, + { + "epoch": 0.5795700292583641, + "grad_norm": 2.7026264667510986, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8722703456878662, + "num_tokens": 177016639.0, + "step": 4556 + }, + { + "epoch": 0.5796972395369546, + "grad_norm": 2.7740962505340576, + "learning_rate": 1e-06, + "loss": 0.4052, + "mean_token_accuracy": 0.8629004955291748, + "num_tokens": 177058691.0, + "step": 4557 + }, + { + "epoch": 0.5798244498155452, + "grad_norm": 2.6003966331481934, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8826956748962402, + "num_tokens": 177095906.0, + "step": 4558 + }, + { + "epoch": 0.5799516600941356, + "grad_norm": 2.591578245162964, + "learning_rate": 1e-06, + "loss": 0.3956, + "mean_token_accuracy": 0.8630107641220093, + "num_tokens": 177138831.0, + "step": 4559 + }, + { + "epoch": 0.5800788703727261, + "grad_norm": 2.5590927600860596, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8811998963356018, + "num_tokens": 177177750.0, + "step": 4560 + }, + { + "epoch": 0.5802060806513166, + "grad_norm": 2.5738143920898438, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8824642896652222, + "num_tokens": 177218883.0, + "step": 4561 + }, + { + "epoch": 0.5803332909299072, + "grad_norm": 2.468285083770752, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8870505690574646, + "num_tokens": 177258855.0, + "step": 4562 + }, + { + "epoch": 0.5804605012084977, + "grad_norm": 2.858952760696411, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8702951669692993, + "num_tokens": 177293499.0, + "step": 4563 + }, + { + "epoch": 0.5805877114870882, + "grad_norm": 2.681126594543457, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8717189431190491, + "num_tokens": 177334158.0, + "step": 4564 + }, + { + "epoch": 0.5807149217656786, + "grad_norm": 2.5305016040802, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8704291582107544, + "num_tokens": 177377729.0, + "step": 4565 + }, + { + "epoch": 0.5808421320442692, + "grad_norm": 2.6873137950897217, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.871098518371582, + "num_tokens": 177414390.0, + "step": 4566 + }, + { + "epoch": 0.5809693423228597, + "grad_norm": 2.721710681915283, + "learning_rate": 1e-06, + "loss": 0.393, + "mean_token_accuracy": 0.8636606335639954, + "num_tokens": 177453919.0, + "step": 4567 + }, + { + "epoch": 0.5810965526014502, + "grad_norm": 2.3037004470825195, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8902624845504761, + "num_tokens": 177496972.0, + "step": 4568 + }, + { + "epoch": 0.5812237628800407, + "grad_norm": 2.948065757751465, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8684068918228149, + "num_tokens": 177534191.0, + "step": 4569 + }, + { + "epoch": 0.5813509731586313, + "grad_norm": 2.4214582443237305, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8721851110458374, + "num_tokens": 177581412.0, + "step": 4570 + }, + { + "epoch": 0.5814781834372217, + "grad_norm": 2.5498056411743164, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8798012733459473, + "num_tokens": 177619370.0, + "step": 4571 + }, + { + "epoch": 0.5816053937158122, + "grad_norm": 2.5289061069488525, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8870100975036621, + "num_tokens": 177659315.0, + "step": 4572 + }, + { + "epoch": 0.5817326039944027, + "grad_norm": 2.710524797439575, + "learning_rate": 1e-06, + "loss": 0.3918, + "mean_token_accuracy": 0.8632183074951172, + "num_tokens": 177700371.0, + "step": 4573 + }, + { + "epoch": 0.5818598142729933, + "grad_norm": 2.3909859657287598, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8841360807418823, + "num_tokens": 177746015.0, + "step": 4574 + }, + { + "epoch": 0.5819870245515838, + "grad_norm": 2.475898265838623, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8821409940719604, + "num_tokens": 177785524.0, + "step": 4575 + }, + { + "epoch": 0.5821142348301743, + "grad_norm": 2.7209339141845703, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8735538721084595, + "num_tokens": 177822753.0, + "step": 4576 + }, + { + "epoch": 0.5822414451087647, + "grad_norm": 2.7165963649749756, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.8634650707244873, + "num_tokens": 177863595.0, + "step": 4577 + }, + { + "epoch": 0.5823686553873553, + "grad_norm": 2.726064682006836, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8702719211578369, + "num_tokens": 177902587.0, + "step": 4578 + }, + { + "epoch": 0.5824958656659458, + "grad_norm": 2.3046350479125977, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8976436853408813, + "num_tokens": 177945397.0, + "step": 4579 + }, + { + "epoch": 0.5826230759445363, + "grad_norm": 3.1141037940979004, + "learning_rate": 1e-06, + "loss": 0.4363, + "mean_token_accuracy": 0.8509180545806885, + "num_tokens": 177982487.0, + "step": 4580 + }, + { + "epoch": 0.5827502862231269, + "grad_norm": 2.5992813110351562, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8930187225341797, + "num_tokens": 178020107.0, + "step": 4581 + }, + { + "epoch": 0.5828774965017174, + "grad_norm": 2.897793769836426, + "learning_rate": 1e-06, + "loss": 0.3888, + "mean_token_accuracy": 0.8663261532783508, + "num_tokens": 178055715.0, + "step": 4582 + }, + { + "epoch": 0.5830047067803078, + "grad_norm": 2.8109781742095947, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8682008385658264, + "num_tokens": 178090853.0, + "step": 4583 + }, + { + "epoch": 0.5831319170588983, + "grad_norm": 2.6530516147613525, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8822951316833496, + "num_tokens": 178129644.0, + "step": 4584 + }, + { + "epoch": 0.5832591273374889, + "grad_norm": 2.6397640705108643, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.869041919708252, + "num_tokens": 178170721.0, + "step": 4585 + }, + { + "epoch": 0.5833863376160794, + "grad_norm": 2.700664520263672, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8806758522987366, + "num_tokens": 178207257.0, + "step": 4586 + }, + { + "epoch": 0.5835135478946699, + "grad_norm": 2.8366477489471436, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8800310492515564, + "num_tokens": 178240624.0, + "step": 4587 + }, + { + "epoch": 0.5836407581732604, + "grad_norm": 2.5758719444274902, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8692234754562378, + "num_tokens": 178280574.0, + "step": 4588 + }, + { + "epoch": 0.5837679684518509, + "grad_norm": 2.892947196960449, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8698099255561829, + "num_tokens": 178316285.0, + "step": 4589 + }, + { + "epoch": 0.5838951787304414, + "grad_norm": 2.753962755203247, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8777382373809814, + "num_tokens": 178350357.0, + "step": 4590 + }, + { + "epoch": 0.5840223890090319, + "grad_norm": 2.5169789791107178, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8831855654716492, + "num_tokens": 178390216.0, + "step": 4591 + }, + { + "epoch": 0.5841495992876224, + "grad_norm": 2.6789417266845703, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8828128576278687, + "num_tokens": 178427448.0, + "step": 4592 + }, + { + "epoch": 0.584276809566213, + "grad_norm": 2.7332961559295654, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.8655494451522827, + "num_tokens": 178467871.0, + "step": 4593 + }, + { + "epoch": 0.5844040198448035, + "grad_norm": 2.6120052337646484, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8870584964752197, + "num_tokens": 178502884.0, + "step": 4594 + }, + { + "epoch": 0.5845312301233939, + "grad_norm": 2.545691728591919, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8650445938110352, + "num_tokens": 178550196.0, + "step": 4595 + }, + { + "epoch": 0.5846584404019844, + "grad_norm": 2.491243362426758, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.885410487651825, + "num_tokens": 178588672.0, + "step": 4596 + }, + { + "epoch": 0.584785650680575, + "grad_norm": 2.7131447792053223, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.8651098012924194, + "num_tokens": 178626671.0, + "step": 4597 + }, + { + "epoch": 0.5849128609591655, + "grad_norm": 2.5046546459198, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8854018449783325, + "num_tokens": 178664474.0, + "step": 4598 + }, + { + "epoch": 0.585040071237756, + "grad_norm": 2.552734136581421, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8877263069152832, + "num_tokens": 178701175.0, + "step": 4599 + }, + { + "epoch": 0.5851672815163466, + "grad_norm": 2.636221170425415, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8782746195793152, + "num_tokens": 178738851.0, + "step": 4600 + }, + { + "epoch": 0.585294491794937, + "grad_norm": 2.6104848384857178, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8859293460845947, + "num_tokens": 178778655.0, + "step": 4601 + }, + { + "epoch": 0.5854217020735275, + "grad_norm": 2.8344573974609375, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.8674307465553284, + "num_tokens": 178815099.0, + "step": 4602 + }, + { + "epoch": 0.585548912352118, + "grad_norm": 2.671912670135498, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8834555745124817, + "num_tokens": 178853763.0, + "step": 4603 + }, + { + "epoch": 0.5856761226307086, + "grad_norm": 2.789191246032715, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8775448799133301, + "num_tokens": 178892393.0, + "step": 4604 + }, + { + "epoch": 0.5858033329092991, + "grad_norm": 2.6628589630126953, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8706340789794922, + "num_tokens": 178932305.0, + "step": 4605 + }, + { + "epoch": 0.5859305431878896, + "grad_norm": 2.5537140369415283, + "learning_rate": 1e-06, + "loss": 0.3723, + "mean_token_accuracy": 0.8728621006011963, + "num_tokens": 178972754.0, + "step": 4606 + }, + { + "epoch": 0.5860577534664801, + "grad_norm": 2.5619516372680664, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.8626729249954224, + "num_tokens": 179017734.0, + "step": 4607 + }, + { + "epoch": 0.5861849637450706, + "grad_norm": 2.994960069656372, + "learning_rate": 1e-06, + "loss": 0.4114, + "mean_token_accuracy": 0.8539307117462158, + "num_tokens": 179051138.0, + "step": 4608 + }, + { + "epoch": 0.5863121740236611, + "grad_norm": 2.4297449588775635, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8813857436180115, + "num_tokens": 179098463.0, + "step": 4609 + }, + { + "epoch": 0.5864393843022516, + "grad_norm": 2.7288503646850586, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8755323886871338, + "num_tokens": 179133563.0, + "step": 4610 + }, + { + "epoch": 0.5865665945808421, + "grad_norm": 2.584397792816162, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8844724893569946, + "num_tokens": 179169472.0, + "step": 4611 + }, + { + "epoch": 0.5866938048594327, + "grad_norm": 2.7383155822753906, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8658586740493774, + "num_tokens": 179210594.0, + "step": 4612 + }, + { + "epoch": 0.5868210151380232, + "grad_norm": 2.48524808883667, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8793537616729736, + "num_tokens": 179252011.0, + "step": 4613 + }, + { + "epoch": 0.5869482254166136, + "grad_norm": 2.469482898712158, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8737492561340332, + "num_tokens": 179298219.0, + "step": 4614 + }, + { + "epoch": 0.5870754356952042, + "grad_norm": 2.649695634841919, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8749524354934692, + "num_tokens": 179335762.0, + "step": 4615 + }, + { + "epoch": 0.5872026459737947, + "grad_norm": 2.905545711517334, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8740840554237366, + "num_tokens": 179370519.0, + "step": 4616 + }, + { + "epoch": 0.5873298562523852, + "grad_norm": 2.3943300247192383, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8878984451293945, + "num_tokens": 179415581.0, + "step": 4617 + }, + { + "epoch": 0.5874570665309757, + "grad_norm": 2.451334238052368, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8746594190597534, + "num_tokens": 179461181.0, + "step": 4618 + }, + { + "epoch": 0.5875842768095663, + "grad_norm": 2.692058563232422, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8912346363067627, + "num_tokens": 179493090.0, + "step": 4619 + }, + { + "epoch": 0.5877114870881567, + "grad_norm": 2.851649045944214, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8715922236442566, + "num_tokens": 179527083.0, + "step": 4620 + }, + { + "epoch": 0.5878386973667472, + "grad_norm": 2.7575533390045166, + "learning_rate": 1e-06, + "loss": 0.3975, + "mean_token_accuracy": 0.8684847950935364, + "num_tokens": 179567584.0, + "step": 4621 + }, + { + "epoch": 0.5879659076453377, + "grad_norm": 2.4873366355895996, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8786460161209106, + "num_tokens": 179607077.0, + "step": 4622 + }, + { + "epoch": 0.5880931179239283, + "grad_norm": 2.5464892387390137, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8828599452972412, + "num_tokens": 179647429.0, + "step": 4623 + }, + { + "epoch": 0.5882203282025188, + "grad_norm": 2.541883945465088, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8682490587234497, + "num_tokens": 179690910.0, + "step": 4624 + }, + { + "epoch": 0.5883475384811093, + "grad_norm": 2.755858898162842, + "learning_rate": 1e-06, + "loss": 0.4077, + "mean_token_accuracy": 0.8615313768386841, + "num_tokens": 179732673.0, + "step": 4625 + }, + { + "epoch": 0.5884747487596997, + "grad_norm": 2.5591676235198975, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8678710460662842, + "num_tokens": 179774810.0, + "step": 4626 + }, + { + "epoch": 0.5886019590382903, + "grad_norm": 2.625567674636841, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8990070223808289, + "num_tokens": 179811327.0, + "step": 4627 + }, + { + "epoch": 0.5887291693168808, + "grad_norm": 2.578237533569336, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8808400630950928, + "num_tokens": 179852283.0, + "step": 4628 + }, + { + "epoch": 0.5888563795954713, + "grad_norm": 2.557821273803711, + "learning_rate": 1e-06, + "loss": 0.3906, + "mean_token_accuracy": 0.8636919260025024, + "num_tokens": 179897803.0, + "step": 4629 + }, + { + "epoch": 0.5889835898740619, + "grad_norm": 2.7179243564605713, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8703440427780151, + "num_tokens": 179934221.0, + "step": 4630 + }, + { + "epoch": 0.5891108001526524, + "grad_norm": 2.2930312156677246, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8899467587471008, + "num_tokens": 179979710.0, + "step": 4631 + }, + { + "epoch": 0.5892380104312428, + "grad_norm": 2.5671043395996094, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8812021017074585, + "num_tokens": 180016992.0, + "step": 4632 + }, + { + "epoch": 0.5893652207098333, + "grad_norm": 2.7023584842681885, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8674273490905762, + "num_tokens": 180055774.0, + "step": 4633 + }, + { + "epoch": 0.5894924309884239, + "grad_norm": 2.5399909019470215, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8788083791732788, + "num_tokens": 180095205.0, + "step": 4634 + }, + { + "epoch": 0.5896196412670144, + "grad_norm": 2.4677071571350098, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8807318210601807, + "num_tokens": 180138006.0, + "step": 4635 + }, + { + "epoch": 0.5897468515456049, + "grad_norm": 2.559110641479492, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8655649423599243, + "num_tokens": 180180321.0, + "step": 4636 + }, + { + "epoch": 0.5898740618241954, + "grad_norm": 2.6687891483306885, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.880764365196228, + "num_tokens": 180216654.0, + "step": 4637 + }, + { + "epoch": 0.5900012721027859, + "grad_norm": 2.5190341472625732, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8832122087478638, + "num_tokens": 180257754.0, + "step": 4638 + }, + { + "epoch": 0.5901284823813764, + "grad_norm": 2.5047879219055176, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8819917440414429, + "num_tokens": 180296856.0, + "step": 4639 + }, + { + "epoch": 0.5902556926599669, + "grad_norm": 2.8185620307922363, + "learning_rate": 1e-06, + "loss": 0.4151, + "mean_token_accuracy": 0.8595727682113647, + "num_tokens": 180336711.0, + "step": 4640 + }, + { + "epoch": 0.5903829029385574, + "grad_norm": 2.636359930038452, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8896547555923462, + "num_tokens": 180372890.0, + "step": 4641 + }, + { + "epoch": 0.590510113217148, + "grad_norm": 2.375276565551758, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8815813660621643, + "num_tokens": 180416952.0, + "step": 4642 + }, + { + "epoch": 0.5906373234957385, + "grad_norm": 2.7164299488067627, + "learning_rate": 1e-06, + "loss": 0.418, + "mean_token_accuracy": 0.8577971458435059, + "num_tokens": 180457546.0, + "step": 4643 + }, + { + "epoch": 0.5907645337743289, + "grad_norm": 2.4008419513702393, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8814647793769836, + "num_tokens": 180501450.0, + "step": 4644 + }, + { + "epoch": 0.5908917440529194, + "grad_norm": 2.7742903232574463, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8627538084983826, + "num_tokens": 180536984.0, + "step": 4645 + }, + { + "epoch": 0.59101895433151, + "grad_norm": 2.692115545272827, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8711552619934082, + "num_tokens": 180576966.0, + "step": 4646 + }, + { + "epoch": 0.5911461646101005, + "grad_norm": 2.5268232822418213, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8813412189483643, + "num_tokens": 180615705.0, + "step": 4647 + }, + { + "epoch": 0.591273374888691, + "grad_norm": 2.535853147506714, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8808016777038574, + "num_tokens": 180655924.0, + "step": 4648 + }, + { + "epoch": 0.5914005851672816, + "grad_norm": 2.641568899154663, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8728793859481812, + "num_tokens": 180695769.0, + "step": 4649 + }, + { + "epoch": 0.591527795445872, + "grad_norm": 2.518561840057373, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8858106732368469, + "num_tokens": 180733740.0, + "step": 4650 + }, + { + "epoch": 0.5916550057244625, + "grad_norm": 2.3474104404449463, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8841743469238281, + "num_tokens": 180779841.0, + "step": 4651 + }, + { + "epoch": 0.591782216003053, + "grad_norm": 2.5118513107299805, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8777087926864624, + "num_tokens": 180819712.0, + "step": 4652 + }, + { + "epoch": 0.5919094262816436, + "grad_norm": 3.1044702529907227, + "learning_rate": 1e-06, + "loss": 0.4183, + "mean_token_accuracy": 0.8514260649681091, + "num_tokens": 180856075.0, + "step": 4653 + }, + { + "epoch": 0.5920366365602341, + "grad_norm": 2.9545695781707764, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8646625280380249, + "num_tokens": 180888261.0, + "step": 4654 + }, + { + "epoch": 0.5921638468388246, + "grad_norm": 2.654242515563965, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8880053162574768, + "num_tokens": 180927273.0, + "step": 4655 + }, + { + "epoch": 0.592291057117415, + "grad_norm": 2.4842636585235596, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.88238525390625, + "num_tokens": 180968974.0, + "step": 4656 + }, + { + "epoch": 0.5924182673960056, + "grad_norm": 2.5763072967529297, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8768935799598694, + "num_tokens": 181008194.0, + "step": 4657 + }, + { + "epoch": 0.5925454776745961, + "grad_norm": 2.729321002960205, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8673318028450012, + "num_tokens": 181051921.0, + "step": 4658 + }, + { + "epoch": 0.5926726879531866, + "grad_norm": 2.7650182247161865, + "learning_rate": 1e-06, + "loss": 0.4087, + "mean_token_accuracy": 0.8581175208091736, + "num_tokens": 181090258.0, + "step": 4659 + }, + { + "epoch": 0.5927998982317771, + "grad_norm": 2.7834811210632324, + "learning_rate": 1e-06, + "loss": 0.3998, + "mean_token_accuracy": 0.8633145093917847, + "num_tokens": 181128121.0, + "step": 4660 + }, + { + "epoch": 0.5929271085103677, + "grad_norm": 2.646683931350708, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8736371994018555, + "num_tokens": 181165728.0, + "step": 4661 + }, + { + "epoch": 0.5930543187889582, + "grad_norm": 2.7639176845550537, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8726365566253662, + "num_tokens": 181200749.0, + "step": 4662 + }, + { + "epoch": 0.5931815290675486, + "grad_norm": 2.478912830352783, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8747761845588684, + "num_tokens": 181244583.0, + "step": 4663 + }, + { + "epoch": 0.5933087393461391, + "grad_norm": 2.76192569732666, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8750512599945068, + "num_tokens": 181282689.0, + "step": 4664 + }, + { + "epoch": 0.5934359496247297, + "grad_norm": 2.347921371459961, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8945600390434265, + "num_tokens": 181324459.0, + "step": 4665 + }, + { + "epoch": 0.5935631599033202, + "grad_norm": 2.5098087787628174, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8737497329711914, + "num_tokens": 181370558.0, + "step": 4666 + }, + { + "epoch": 0.5936903701819107, + "grad_norm": 2.443800449371338, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8790156841278076, + "num_tokens": 181413386.0, + "step": 4667 + }, + { + "epoch": 0.5938175804605013, + "grad_norm": 2.662654399871826, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8818745017051697, + "num_tokens": 181453570.0, + "step": 4668 + }, + { + "epoch": 0.5939447907390917, + "grad_norm": 2.5013952255249023, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8788551092147827, + "num_tokens": 181493953.0, + "step": 4669 + }, + { + "epoch": 0.5940720010176822, + "grad_norm": 2.7016046047210693, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8863538503646851, + "num_tokens": 181530338.0, + "step": 4670 + }, + { + "epoch": 0.5941992112962727, + "grad_norm": 2.5889415740966797, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.873542070388794, + "num_tokens": 181571525.0, + "step": 4671 + }, + { + "epoch": 0.5943264215748633, + "grad_norm": 2.7647202014923096, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.86609947681427, + "num_tokens": 181608756.0, + "step": 4672 + }, + { + "epoch": 0.5944536318534538, + "grad_norm": 2.979198455810547, + "learning_rate": 1e-06, + "loss": 0.3968, + "mean_token_accuracy": 0.8636147379875183, + "num_tokens": 181644435.0, + "step": 4673 + }, + { + "epoch": 0.5945808421320443, + "grad_norm": 2.410818576812744, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8814550042152405, + "num_tokens": 181687036.0, + "step": 4674 + }, + { + "epoch": 0.5947080524106347, + "grad_norm": 2.8247599601745605, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.877063512802124, + "num_tokens": 181724476.0, + "step": 4675 + }, + { + "epoch": 0.5948352626892253, + "grad_norm": 2.5768065452575684, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8806257247924805, + "num_tokens": 181762313.0, + "step": 4676 + }, + { + "epoch": 0.5949624729678158, + "grad_norm": 2.962832450866699, + "learning_rate": 1e-06, + "loss": 0.3868, + "mean_token_accuracy": 0.8636797666549683, + "num_tokens": 181794463.0, + "step": 4677 + }, + { + "epoch": 0.5950896832464063, + "grad_norm": 2.6923303604125977, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.864483654499054, + "num_tokens": 181835440.0, + "step": 4678 + }, + { + "epoch": 0.5952168935249968, + "grad_norm": 2.555152416229248, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8689112663269043, + "num_tokens": 181879810.0, + "step": 4679 + }, + { + "epoch": 0.5953441038035874, + "grad_norm": 2.645338773727417, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8803671598434448, + "num_tokens": 181915911.0, + "step": 4680 + }, + { + "epoch": 0.5954713140821778, + "grad_norm": 2.387507200241089, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8918389081954956, + "num_tokens": 181956947.0, + "step": 4681 + }, + { + "epoch": 0.5955985243607683, + "grad_norm": 2.6106114387512207, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8704758882522583, + "num_tokens": 181995411.0, + "step": 4682 + }, + { + "epoch": 0.5957257346393589, + "grad_norm": 2.765691041946411, + "learning_rate": 1e-06, + "loss": 0.3904, + "mean_token_accuracy": 0.8653792142868042, + "num_tokens": 182031640.0, + "step": 4683 + }, + { + "epoch": 0.5958529449179494, + "grad_norm": 3.060049295425415, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8722227811813354, + "num_tokens": 182062083.0, + "step": 4684 + }, + { + "epoch": 0.5959801551965399, + "grad_norm": 2.541294574737549, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8921300172805786, + "num_tokens": 182097808.0, + "step": 4685 + }, + { + "epoch": 0.5961073654751304, + "grad_norm": 2.45845365524292, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8795469403266907, + "num_tokens": 182142016.0, + "step": 4686 + }, + { + "epoch": 0.5962345757537209, + "grad_norm": 3.2101054191589355, + "learning_rate": 1e-06, + "loss": 0.4772, + "mean_token_accuracy": 0.8349912166595459, + "num_tokens": 182173984.0, + "step": 4687 + }, + { + "epoch": 0.5963617860323114, + "grad_norm": 2.6788599491119385, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8723785877227783, + "num_tokens": 182215111.0, + "step": 4688 + }, + { + "epoch": 0.5964889963109019, + "grad_norm": 2.5919435024261475, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8685261011123657, + "num_tokens": 182256734.0, + "step": 4689 + }, + { + "epoch": 0.5966162065894924, + "grad_norm": 2.4357407093048096, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8865131735801697, + "num_tokens": 182298336.0, + "step": 4690 + }, + { + "epoch": 0.596743416868083, + "grad_norm": 2.9335739612579346, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.876571536064148, + "num_tokens": 182331281.0, + "step": 4691 + }, + { + "epoch": 0.5968706271466735, + "grad_norm": 2.6176536083221436, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8688161373138428, + "num_tokens": 182371957.0, + "step": 4692 + }, + { + "epoch": 0.5969978374252639, + "grad_norm": 2.4308009147644043, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8862248659133911, + "num_tokens": 182414112.0, + "step": 4693 + }, + { + "epoch": 0.5971250477038544, + "grad_norm": 2.7770955562591553, + "learning_rate": 1e-06, + "loss": 0.3935, + "mean_token_accuracy": 0.8624222874641418, + "num_tokens": 182454632.0, + "step": 4694 + }, + { + "epoch": 0.597252257982445, + "grad_norm": 2.8457298278808594, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8725310564041138, + "num_tokens": 182488252.0, + "step": 4695 + }, + { + "epoch": 0.5973794682610355, + "grad_norm": 2.6418588161468506, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8793606758117676, + "num_tokens": 182523219.0, + "step": 4696 + }, + { + "epoch": 0.597506678539626, + "grad_norm": 2.812479257583618, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8754457235336304, + "num_tokens": 182557813.0, + "step": 4697 + }, + { + "epoch": 0.5976338888182166, + "grad_norm": 2.507450819015503, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8693366050720215, + "num_tokens": 182602433.0, + "step": 4698 + }, + { + "epoch": 0.597761099096807, + "grad_norm": 2.6541550159454346, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8818238973617554, + "num_tokens": 182639658.0, + "step": 4699 + }, + { + "epoch": 0.5978883093753975, + "grad_norm": 2.486302614212036, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8831490278244019, + "num_tokens": 182678548.0, + "step": 4700 + }, + { + "epoch": 0.598015519653988, + "grad_norm": 2.6911795139312744, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8776458501815796, + "num_tokens": 182716638.0, + "step": 4701 + }, + { + "epoch": 0.5981427299325786, + "grad_norm": 2.6819887161254883, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8728130459785461, + "num_tokens": 182754776.0, + "step": 4702 + }, + { + "epoch": 0.5982699402111691, + "grad_norm": 2.5455164909362793, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.882946252822876, + "num_tokens": 182793664.0, + "step": 4703 + }, + { + "epoch": 0.5983971504897596, + "grad_norm": 2.578732967376709, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8725902438163757, + "num_tokens": 182833619.0, + "step": 4704 + }, + { + "epoch": 0.59852436076835, + "grad_norm": 2.7949938774108887, + "learning_rate": 1e-06, + "loss": 0.394, + "mean_token_accuracy": 0.8653171062469482, + "num_tokens": 182870002.0, + "step": 4705 + }, + { + "epoch": 0.5986515710469406, + "grad_norm": 2.509852647781372, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8739180564880371, + "num_tokens": 182911622.0, + "step": 4706 + }, + { + "epoch": 0.5987787813255311, + "grad_norm": 2.5762035846710205, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8826029300689697, + "num_tokens": 182951855.0, + "step": 4707 + }, + { + "epoch": 0.5989059916041216, + "grad_norm": 2.6142122745513916, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8712880611419678, + "num_tokens": 182989955.0, + "step": 4708 + }, + { + "epoch": 0.5990332018827121, + "grad_norm": 2.6682450771331787, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8752186298370361, + "num_tokens": 183029773.0, + "step": 4709 + }, + { + "epoch": 0.5991604121613027, + "grad_norm": 2.463148593902588, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8775049448013306, + "num_tokens": 183071915.0, + "step": 4710 + }, + { + "epoch": 0.5992876224398932, + "grad_norm": 2.5260140895843506, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8867557644844055, + "num_tokens": 183110577.0, + "step": 4711 + }, + { + "epoch": 0.5994148327184836, + "grad_norm": 2.8947319984436035, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8725301027297974, + "num_tokens": 183145692.0, + "step": 4712 + }, + { + "epoch": 0.5995420429970741, + "grad_norm": 2.600189208984375, + "learning_rate": 1e-06, + "loss": 0.3948, + "mean_token_accuracy": 0.8637036681175232, + "num_tokens": 183188775.0, + "step": 4713 + }, + { + "epoch": 0.5996692532756647, + "grad_norm": 2.536503314971924, + "learning_rate": 1e-06, + "loss": 0.3956, + "mean_token_accuracy": 0.86372309923172, + "num_tokens": 183235204.0, + "step": 4714 + }, + { + "epoch": 0.5997964635542552, + "grad_norm": 2.5527915954589844, + "learning_rate": 1e-06, + "loss": 0.3885, + "mean_token_accuracy": 0.866464376449585, + "num_tokens": 183278167.0, + "step": 4715 + }, + { + "epoch": 0.5999236738328457, + "grad_norm": 2.7043516635894775, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8826684951782227, + "num_tokens": 183312242.0, + "step": 4716 + }, + { + "epoch": 0.6000508841114363, + "grad_norm": 2.505284309387207, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8787308931350708, + "num_tokens": 183354869.0, + "step": 4717 + }, + { + "epoch": 0.6001780943900267, + "grad_norm": 2.6886160373687744, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8792451620101929, + "num_tokens": 183389264.0, + "step": 4718 + }, + { + "epoch": 0.6003053046686172, + "grad_norm": 3.074697732925415, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8679388165473938, + "num_tokens": 183421173.0, + "step": 4719 + }, + { + "epoch": 0.6004325149472077, + "grad_norm": 2.658130645751953, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8759036660194397, + "num_tokens": 183461473.0, + "step": 4720 + }, + { + "epoch": 0.6005597252257983, + "grad_norm": 2.4040427207946777, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8908388614654541, + "num_tokens": 183503195.0, + "step": 4721 + }, + { + "epoch": 0.6006869355043888, + "grad_norm": 2.7246627807617188, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8754587173461914, + "num_tokens": 183546318.0, + "step": 4722 + }, + { + "epoch": 0.6008141457829793, + "grad_norm": 2.551637649536133, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8758957386016846, + "num_tokens": 183587634.0, + "step": 4723 + }, + { + "epoch": 0.6009413560615697, + "grad_norm": 2.8286232948303223, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8717784285545349, + "num_tokens": 183620728.0, + "step": 4724 + }, + { + "epoch": 0.6010685663401603, + "grad_norm": 2.5461058616638184, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8743505477905273, + "num_tokens": 183661567.0, + "step": 4725 + }, + { + "epoch": 0.6011957766187508, + "grad_norm": 2.5364456176757812, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8884017467498779, + "num_tokens": 183698966.0, + "step": 4726 + }, + { + "epoch": 0.6013229868973413, + "grad_norm": 2.7206685543060303, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8718040585517883, + "num_tokens": 183736966.0, + "step": 4727 + }, + { + "epoch": 0.6014501971759318, + "grad_norm": 2.73264217376709, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8735178709030151, + "num_tokens": 183774329.0, + "step": 4728 + }, + { + "epoch": 0.6015774074545224, + "grad_norm": 2.6724565029144287, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8747348189353943, + "num_tokens": 183815938.0, + "step": 4729 + }, + { + "epoch": 0.6017046177331128, + "grad_norm": 2.6793508529663086, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8767988681793213, + "num_tokens": 183857312.0, + "step": 4730 + }, + { + "epoch": 0.6018318280117033, + "grad_norm": 2.731922149658203, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8698711395263672, + "num_tokens": 183897397.0, + "step": 4731 + }, + { + "epoch": 0.6019590382902938, + "grad_norm": 2.689699411392212, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.87505042552948, + "num_tokens": 183934320.0, + "step": 4732 + }, + { + "epoch": 0.6020862485688844, + "grad_norm": 2.6501340866088867, + "learning_rate": 1e-06, + "loss": 0.3879, + "mean_token_accuracy": 0.864470362663269, + "num_tokens": 183975488.0, + "step": 4733 + }, + { + "epoch": 0.6022134588474749, + "grad_norm": 2.6008269786834717, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.880445122718811, + "num_tokens": 184015617.0, + "step": 4734 + }, + { + "epoch": 0.6023406691260654, + "grad_norm": 2.7153995037078857, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8729130625724792, + "num_tokens": 184052480.0, + "step": 4735 + }, + { + "epoch": 0.6024678794046558, + "grad_norm": 2.859938144683838, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8744058609008789, + "num_tokens": 184088407.0, + "step": 4736 + }, + { + "epoch": 0.6025950896832464, + "grad_norm": 2.878706932067871, + "learning_rate": 1e-06, + "loss": 0.3676, + "mean_token_accuracy": 0.8715033531188965, + "num_tokens": 184123280.0, + "step": 4737 + }, + { + "epoch": 0.6027222999618369, + "grad_norm": 2.2893028259277344, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8977935314178467, + "num_tokens": 184164300.0, + "step": 4738 + }, + { + "epoch": 0.6028495102404274, + "grad_norm": 2.6481308937072754, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.88034987449646, + "num_tokens": 184202155.0, + "step": 4739 + }, + { + "epoch": 0.602976720519018, + "grad_norm": 2.820831537246704, + "learning_rate": 1e-06, + "loss": 0.4207, + "mean_token_accuracy": 0.8551281094551086, + "num_tokens": 184240487.0, + "step": 4740 + }, + { + "epoch": 0.6031039307976085, + "grad_norm": 2.6619279384613037, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8795130252838135, + "num_tokens": 184277828.0, + "step": 4741 + }, + { + "epoch": 0.6032311410761989, + "grad_norm": 2.3896989822387695, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8865015506744385, + "num_tokens": 184319734.0, + "step": 4742 + }, + { + "epoch": 0.6033583513547894, + "grad_norm": 2.402118682861328, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8648180961608887, + "num_tokens": 184366992.0, + "step": 4743 + }, + { + "epoch": 0.60348556163338, + "grad_norm": 2.7415642738342285, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8771138191223145, + "num_tokens": 184403385.0, + "step": 4744 + }, + { + "epoch": 0.6036127719119705, + "grad_norm": 2.679936647415161, + "learning_rate": 1e-06, + "loss": 0.4291, + "mean_token_accuracy": 0.8550885915756226, + "num_tokens": 184442724.0, + "step": 4745 + }, + { + "epoch": 0.603739982190561, + "grad_norm": 2.781201124191284, + "learning_rate": 1e-06, + "loss": 0.4183, + "mean_token_accuracy": 0.858519434928894, + "num_tokens": 184484107.0, + "step": 4746 + }, + { + "epoch": 0.6038671924691515, + "grad_norm": 2.627241373062134, + "learning_rate": 1e-06, + "loss": 0.3823, + "mean_token_accuracy": 0.8722543716430664, + "num_tokens": 184526024.0, + "step": 4747 + }, + { + "epoch": 0.603994402747742, + "grad_norm": 2.7181508541107178, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8704725503921509, + "num_tokens": 184570331.0, + "step": 4748 + }, + { + "epoch": 0.6041216130263325, + "grad_norm": 3.088484525680542, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8674099445343018, + "num_tokens": 184603992.0, + "step": 4749 + }, + { + "epoch": 0.604248823304923, + "grad_norm": 2.734586715698242, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8820222616195679, + "num_tokens": 184640751.0, + "step": 4750 + }, + { + "epoch": 0.6043760335835135, + "grad_norm": 2.806600332260132, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8773775100708008, + "num_tokens": 184675534.0, + "step": 4751 + }, + { + "epoch": 0.6045032438621041, + "grad_norm": 2.802234411239624, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8831517100334167, + "num_tokens": 184707285.0, + "step": 4752 + }, + { + "epoch": 0.6046304541406946, + "grad_norm": 2.801687479019165, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8746911287307739, + "num_tokens": 184745905.0, + "step": 4753 + }, + { + "epoch": 0.604757664419285, + "grad_norm": 2.8021748065948486, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8776422739028931, + "num_tokens": 184783197.0, + "step": 4754 + }, + { + "epoch": 0.6048848746978756, + "grad_norm": 2.4630191326141357, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8858460187911987, + "num_tokens": 184824004.0, + "step": 4755 + }, + { + "epoch": 0.6050120849764661, + "grad_norm": 2.8489747047424316, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8740415573120117, + "num_tokens": 184862966.0, + "step": 4756 + }, + { + "epoch": 0.6051392952550566, + "grad_norm": 2.5640158653259277, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8654246926307678, + "num_tokens": 184907705.0, + "step": 4757 + }, + { + "epoch": 0.6052665055336471, + "grad_norm": 2.7181613445281982, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8719371557235718, + "num_tokens": 184947849.0, + "step": 4758 + }, + { + "epoch": 0.6053937158122377, + "grad_norm": 2.7003684043884277, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8723034858703613, + "num_tokens": 184990597.0, + "step": 4759 + }, + { + "epoch": 0.6055209260908282, + "grad_norm": 2.6193549633026123, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8793693780899048, + "num_tokens": 185028710.0, + "step": 4760 + }, + { + "epoch": 0.6056481363694186, + "grad_norm": 2.7963130474090576, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8799216747283936, + "num_tokens": 185064306.0, + "step": 4761 + }, + { + "epoch": 0.6057753466480091, + "grad_norm": 2.625972032546997, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8811858296394348, + "num_tokens": 185102095.0, + "step": 4762 + }, + { + "epoch": 0.6059025569265997, + "grad_norm": 2.8018250465393066, + "learning_rate": 1e-06, + "loss": 0.3936, + "mean_token_accuracy": 0.8635622262954712, + "num_tokens": 185139120.0, + "step": 4763 + }, + { + "epoch": 0.6060297672051902, + "grad_norm": 2.3624539375305176, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8744382262229919, + "num_tokens": 185183321.0, + "step": 4764 + }, + { + "epoch": 0.6061569774837807, + "grad_norm": 2.646210193634033, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8864946365356445, + "num_tokens": 185218194.0, + "step": 4765 + }, + { + "epoch": 0.6062841877623713, + "grad_norm": 2.5006232261657715, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8653824329376221, + "num_tokens": 185262413.0, + "step": 4766 + }, + { + "epoch": 0.6064113980409617, + "grad_norm": 2.559418201446533, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8752312660217285, + "num_tokens": 185304396.0, + "step": 4767 + }, + { + "epoch": 0.6065386083195522, + "grad_norm": 2.599170207977295, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8816519975662231, + "num_tokens": 185342992.0, + "step": 4768 + }, + { + "epoch": 0.6066658185981427, + "grad_norm": 2.68168568611145, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8756253123283386, + "num_tokens": 185379950.0, + "step": 4769 + }, + { + "epoch": 0.6067930288767333, + "grad_norm": 2.7240829467773438, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8824936151504517, + "num_tokens": 185413860.0, + "step": 4770 + }, + { + "epoch": 0.6069202391553238, + "grad_norm": 2.356736183166504, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8817459344863892, + "num_tokens": 185459056.0, + "step": 4771 + }, + { + "epoch": 0.6070474494339143, + "grad_norm": 2.497143268585205, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8930836319923401, + "num_tokens": 185497773.0, + "step": 4772 + }, + { + "epoch": 0.6071746597125047, + "grad_norm": 2.821039915084839, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8720743060112, + "num_tokens": 185536662.0, + "step": 4773 + }, + { + "epoch": 0.6073018699910953, + "grad_norm": 2.8141674995422363, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.8709622025489807, + "num_tokens": 185575426.0, + "step": 4774 + }, + { + "epoch": 0.6074290802696858, + "grad_norm": 2.4794468879699707, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8734396696090698, + "num_tokens": 185619347.0, + "step": 4775 + }, + { + "epoch": 0.6075562905482763, + "grad_norm": 2.6148970127105713, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8798443078994751, + "num_tokens": 185658239.0, + "step": 4776 + }, + { + "epoch": 0.6076835008268668, + "grad_norm": 2.832362413406372, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8868634700775146, + "num_tokens": 185697425.0, + "step": 4777 + }, + { + "epoch": 0.6078107111054574, + "grad_norm": 2.9621520042419434, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8658614158630371, + "num_tokens": 185731571.0, + "step": 4778 + }, + { + "epoch": 0.6079379213840478, + "grad_norm": 2.833024501800537, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8789898753166199, + "num_tokens": 185768068.0, + "step": 4779 + }, + { + "epoch": 0.6080651316626383, + "grad_norm": 2.70162296295166, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8676337003707886, + "num_tokens": 185811808.0, + "step": 4780 + }, + { + "epoch": 0.6081923419412288, + "grad_norm": 2.330549478530884, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8849557638168335, + "num_tokens": 185858723.0, + "step": 4781 + }, + { + "epoch": 0.6083195522198194, + "grad_norm": 2.485809326171875, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8796994090080261, + "num_tokens": 185899694.0, + "step": 4782 + }, + { + "epoch": 0.6084467624984099, + "grad_norm": 2.7341396808624268, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8754940032958984, + "num_tokens": 185937213.0, + "step": 4783 + }, + { + "epoch": 0.6085739727770004, + "grad_norm": 2.5674076080322266, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.869686484336853, + "num_tokens": 185977490.0, + "step": 4784 + }, + { + "epoch": 0.6087011830555908, + "grad_norm": 2.6541574001312256, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8775380253791809, + "num_tokens": 186014231.0, + "step": 4785 + }, + { + "epoch": 0.6088283933341814, + "grad_norm": 2.765425443649292, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8749834895133972, + "num_tokens": 186048262.0, + "step": 4786 + }, + { + "epoch": 0.6089556036127719, + "grad_norm": 2.6723546981811523, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8787281513214111, + "num_tokens": 186087082.0, + "step": 4787 + }, + { + "epoch": 0.6090828138913624, + "grad_norm": 2.6904287338256836, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8809349536895752, + "num_tokens": 186123157.0, + "step": 4788 + }, + { + "epoch": 0.609210024169953, + "grad_norm": 2.758498430252075, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8749221563339233, + "num_tokens": 186156614.0, + "step": 4789 + }, + { + "epoch": 0.6093372344485435, + "grad_norm": 2.567908525466919, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8808263540267944, + "num_tokens": 186195098.0, + "step": 4790 + }, + { + "epoch": 0.6094644447271339, + "grad_norm": 2.7178797721862793, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8703896999359131, + "num_tokens": 186233971.0, + "step": 4791 + }, + { + "epoch": 0.6095916550057244, + "grad_norm": 2.90600323677063, + "learning_rate": 1e-06, + "loss": 0.4043, + "mean_token_accuracy": 0.8629124760627747, + "num_tokens": 186270929.0, + "step": 4792 + }, + { + "epoch": 0.609718865284315, + "grad_norm": 2.4283854961395264, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8877567052841187, + "num_tokens": 186312227.0, + "step": 4793 + }, + { + "epoch": 0.6098460755629055, + "grad_norm": 2.5786452293395996, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8840885758399963, + "num_tokens": 186347568.0, + "step": 4794 + }, + { + "epoch": 0.609973285841496, + "grad_norm": 2.639822483062744, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8966768383979797, + "num_tokens": 186382190.0, + "step": 4795 + }, + { + "epoch": 0.6101004961200865, + "grad_norm": 2.712085247039795, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8750594854354858, + "num_tokens": 186418584.0, + "step": 4796 + }, + { + "epoch": 0.610227706398677, + "grad_norm": 2.6899802684783936, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8868478536605835, + "num_tokens": 186453550.0, + "step": 4797 + }, + { + "epoch": 0.6103549166772675, + "grad_norm": 2.7133917808532715, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8804634809494019, + "num_tokens": 186488858.0, + "step": 4798 + }, + { + "epoch": 0.610482126955858, + "grad_norm": 2.4600653648376465, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.885325014591217, + "num_tokens": 186529096.0, + "step": 4799 + }, + { + "epoch": 0.6106093372344485, + "grad_norm": 2.385491132736206, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8857194185256958, + "num_tokens": 186571240.0, + "step": 4800 + }, + { + "epoch": 0.6107365475130391, + "grad_norm": 2.7285611629486084, + "learning_rate": 1e-06, + "loss": 0.4017, + "mean_token_accuracy": 0.8644937872886658, + "num_tokens": 186610276.0, + "step": 4801 + }, + { + "epoch": 0.6108637577916296, + "grad_norm": 2.6667542457580566, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.879228949546814, + "num_tokens": 186649457.0, + "step": 4802 + }, + { + "epoch": 0.61099096807022, + "grad_norm": 2.743492841720581, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.8623533248901367, + "num_tokens": 186689436.0, + "step": 4803 + }, + { + "epoch": 0.6111181783488105, + "grad_norm": 2.802747964859009, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8690127730369568, + "num_tokens": 186725358.0, + "step": 4804 + }, + { + "epoch": 0.6112453886274011, + "grad_norm": 2.494474411010742, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.8666678667068481, + "num_tokens": 186770553.0, + "step": 4805 + }, + { + "epoch": 0.6113725989059916, + "grad_norm": 2.375284433364868, + "learning_rate": 1e-06, + "loss": 0.361, + "mean_token_accuracy": 0.8738324642181396, + "num_tokens": 186815482.0, + "step": 4806 + }, + { + "epoch": 0.6114998091845821, + "grad_norm": 2.498511791229248, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8837083578109741, + "num_tokens": 186853433.0, + "step": 4807 + }, + { + "epoch": 0.6116270194631727, + "grad_norm": 2.4861865043640137, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8682630062103271, + "num_tokens": 186897317.0, + "step": 4808 + }, + { + "epoch": 0.6117542297417632, + "grad_norm": 2.730656147003174, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8745018243789673, + "num_tokens": 186934395.0, + "step": 4809 + }, + { + "epoch": 0.6118814400203536, + "grad_norm": 2.7143714427948, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8760198354721069, + "num_tokens": 186971092.0, + "step": 4810 + }, + { + "epoch": 0.6120086502989441, + "grad_norm": 2.7159221172332764, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8839273452758789, + "num_tokens": 187007135.0, + "step": 4811 + }, + { + "epoch": 0.6121358605775347, + "grad_norm": 2.4038820266723633, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.87901771068573, + "num_tokens": 187051177.0, + "step": 4812 + }, + { + "epoch": 0.6122630708561252, + "grad_norm": 2.741490125656128, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8806284666061401, + "num_tokens": 187084980.0, + "step": 4813 + }, + { + "epoch": 0.6123902811347157, + "grad_norm": 2.805462598800659, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.865006148815155, + "num_tokens": 187122997.0, + "step": 4814 + }, + { + "epoch": 0.6125174914133062, + "grad_norm": 2.346074104309082, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8764903545379639, + "num_tokens": 187167828.0, + "step": 4815 + }, + { + "epoch": 0.6126447016918967, + "grad_norm": 2.5187606811523438, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8838410973548889, + "num_tokens": 187210209.0, + "step": 4816 + }, + { + "epoch": 0.6127719119704872, + "grad_norm": 2.574099540710449, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8890049457550049, + "num_tokens": 187243665.0, + "step": 4817 + }, + { + "epoch": 0.6128991222490777, + "grad_norm": 2.5379834175109863, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8793063759803772, + "num_tokens": 187282566.0, + "step": 4818 + }, + { + "epoch": 0.6130263325276682, + "grad_norm": 2.605745792388916, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8777603507041931, + "num_tokens": 187323130.0, + "step": 4819 + }, + { + "epoch": 0.6131535428062588, + "grad_norm": 2.716364860534668, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8757749795913696, + "num_tokens": 187359128.0, + "step": 4820 + }, + { + "epoch": 0.6132807530848493, + "grad_norm": 2.516711711883545, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8727507591247559, + "num_tokens": 187400691.0, + "step": 4821 + }, + { + "epoch": 0.6134079633634397, + "grad_norm": 2.5207152366638184, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8697937726974487, + "num_tokens": 187442416.0, + "step": 4822 + }, + { + "epoch": 0.6135351736420303, + "grad_norm": 2.533252716064453, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8910655975341797, + "num_tokens": 187481224.0, + "step": 4823 + }, + { + "epoch": 0.6136623839206208, + "grad_norm": 2.8483633995056152, + "learning_rate": 1e-06, + "loss": 0.3973, + "mean_token_accuracy": 0.8590234518051147, + "num_tokens": 187518612.0, + "step": 4824 + }, + { + "epoch": 0.6137895941992113, + "grad_norm": 2.6278555393218994, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8926103115081787, + "num_tokens": 187555496.0, + "step": 4825 + }, + { + "epoch": 0.6139168044778018, + "grad_norm": 2.378753900527954, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8781519532203674, + "num_tokens": 187600938.0, + "step": 4826 + }, + { + "epoch": 0.6140440147563924, + "grad_norm": 2.555997610092163, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8699544072151184, + "num_tokens": 187640148.0, + "step": 4827 + }, + { + "epoch": 0.6141712250349828, + "grad_norm": 2.748655080795288, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8751722574234009, + "num_tokens": 187674740.0, + "step": 4828 + }, + { + "epoch": 0.6142984353135733, + "grad_norm": 2.6650443077087402, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8880232572555542, + "num_tokens": 187710342.0, + "step": 4829 + }, + { + "epoch": 0.6144256455921638, + "grad_norm": 2.410745859146118, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8799335956573486, + "num_tokens": 187756434.0, + "step": 4830 + }, + { + "epoch": 0.6145528558707544, + "grad_norm": 2.696300983428955, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8780237436294556, + "num_tokens": 187794321.0, + "step": 4831 + }, + { + "epoch": 0.6146800661493449, + "grad_norm": 2.3132741451263428, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8815089464187622, + "num_tokens": 187838943.0, + "step": 4832 + }, + { + "epoch": 0.6148072764279354, + "grad_norm": 2.436504364013672, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8884720802307129, + "num_tokens": 187874892.0, + "step": 4833 + }, + { + "epoch": 0.6149344867065258, + "grad_norm": 2.520507335662842, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8759622573852539, + "num_tokens": 187915313.0, + "step": 4834 + }, + { + "epoch": 0.6150616969851164, + "grad_norm": 2.4314095973968506, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.8676431179046631, + "num_tokens": 187961130.0, + "step": 4835 + }, + { + "epoch": 0.6151889072637069, + "grad_norm": 2.5547194480895996, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8921723365783691, + "num_tokens": 187998846.0, + "step": 4836 + }, + { + "epoch": 0.6153161175422974, + "grad_norm": 2.443817138671875, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8832178115844727, + "num_tokens": 188038966.0, + "step": 4837 + }, + { + "epoch": 0.615443327820888, + "grad_norm": 2.5846855640411377, + "learning_rate": 1e-06, + "loss": 0.422, + "mean_token_accuracy": 0.8546749949455261, + "num_tokens": 188081862.0, + "step": 4838 + }, + { + "epoch": 0.6155705380994785, + "grad_norm": 2.5744943618774414, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8829062581062317, + "num_tokens": 188120449.0, + "step": 4839 + }, + { + "epoch": 0.6156977483780689, + "grad_norm": 2.4756250381469727, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8794890642166138, + "num_tokens": 188161711.0, + "step": 4840 + }, + { + "epoch": 0.6158249586566594, + "grad_norm": 2.6215357780456543, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8820452690124512, + "num_tokens": 188198918.0, + "step": 4841 + }, + { + "epoch": 0.61595216893525, + "grad_norm": 2.5795645713806152, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8755819797515869, + "num_tokens": 188240691.0, + "step": 4842 + }, + { + "epoch": 0.6160793792138405, + "grad_norm": 2.5127246379852295, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8707184791564941, + "num_tokens": 188282591.0, + "step": 4843 + }, + { + "epoch": 0.616206589492431, + "grad_norm": 2.725712537765503, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8644402623176575, + "num_tokens": 188322266.0, + "step": 4844 + }, + { + "epoch": 0.6163337997710215, + "grad_norm": 2.482973337173462, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8792543411254883, + "num_tokens": 188362297.0, + "step": 4845 + }, + { + "epoch": 0.616461010049612, + "grad_norm": 2.531921863555908, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8819935917854309, + "num_tokens": 188398592.0, + "step": 4846 + }, + { + "epoch": 0.6165882203282025, + "grad_norm": 2.6426985263824463, + "learning_rate": 1e-06, + "loss": 0.3795, + "mean_token_accuracy": 0.8695930242538452, + "num_tokens": 188437512.0, + "step": 4847 + }, + { + "epoch": 0.616715430606793, + "grad_norm": 2.958712339401245, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.880836009979248, + "num_tokens": 188470709.0, + "step": 4848 + }, + { + "epoch": 0.6168426408853835, + "grad_norm": 2.8433213233947754, + "learning_rate": 1e-06, + "loss": 0.4348, + "mean_token_accuracy": 0.8507777452468872, + "num_tokens": 188512069.0, + "step": 4849 + }, + { + "epoch": 0.6169698511639741, + "grad_norm": 2.6917295455932617, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.866765022277832, + "num_tokens": 188549562.0, + "step": 4850 + }, + { + "epoch": 0.6170970614425646, + "grad_norm": 2.5554139614105225, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8740509748458862, + "num_tokens": 188589195.0, + "step": 4851 + }, + { + "epoch": 0.617224271721155, + "grad_norm": 2.597822427749634, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8908833265304565, + "num_tokens": 188624614.0, + "step": 4852 + }, + { + "epoch": 0.6173514819997455, + "grad_norm": 2.557523488998413, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9004451632499695, + "num_tokens": 188659337.0, + "step": 4853 + }, + { + "epoch": 0.6174786922783361, + "grad_norm": 2.5444233417510986, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.879220187664032, + "num_tokens": 188701772.0, + "step": 4854 + }, + { + "epoch": 0.6176059025569266, + "grad_norm": 2.622121572494507, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8629844784736633, + "num_tokens": 188740033.0, + "step": 4855 + }, + { + "epoch": 0.6177331128355171, + "grad_norm": 2.6030094623565674, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8700832724571228, + "num_tokens": 188779539.0, + "step": 4856 + }, + { + "epoch": 0.6178603231141077, + "grad_norm": 2.6082770824432373, + "learning_rate": 1e-06, + "loss": 0.3912, + "mean_token_accuracy": 0.8651117086410522, + "num_tokens": 188822164.0, + "step": 4857 + }, + { + "epoch": 0.6179875333926982, + "grad_norm": 2.9406144618988037, + "learning_rate": 1e-06, + "loss": 0.4021, + "mean_token_accuracy": 0.8615436553955078, + "num_tokens": 188858497.0, + "step": 4858 + }, + { + "epoch": 0.6181147436712886, + "grad_norm": 2.420607805252075, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8724738955497742, + "num_tokens": 188905419.0, + "step": 4859 + }, + { + "epoch": 0.6182419539498791, + "grad_norm": 2.5632221698760986, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8763466477394104, + "num_tokens": 188945964.0, + "step": 4860 + }, + { + "epoch": 0.6183691642284697, + "grad_norm": 2.65222430229187, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8699081540107727, + "num_tokens": 188986363.0, + "step": 4861 + }, + { + "epoch": 0.6184963745070602, + "grad_norm": 2.6970419883728027, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8897175788879395, + "num_tokens": 189017720.0, + "step": 4862 + }, + { + "epoch": 0.6186235847856507, + "grad_norm": 2.9262747764587402, + "learning_rate": 1e-06, + "loss": 0.4001, + "mean_token_accuracy": 0.8603882193565369, + "num_tokens": 189050384.0, + "step": 4863 + }, + { + "epoch": 0.6187507950642412, + "grad_norm": 2.4751930236816406, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8800168037414551, + "num_tokens": 189093474.0, + "step": 4864 + }, + { + "epoch": 0.6188780053428317, + "grad_norm": 2.46295166015625, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.871546745300293, + "num_tokens": 189135725.0, + "step": 4865 + }, + { + "epoch": 0.6190052156214222, + "grad_norm": 2.3975882530212402, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8791714906692505, + "num_tokens": 189179948.0, + "step": 4866 + }, + { + "epoch": 0.6191324259000127, + "grad_norm": 2.6194679737091064, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8800554275512695, + "num_tokens": 189216698.0, + "step": 4867 + }, + { + "epoch": 0.6192596361786032, + "grad_norm": 2.6291298866271973, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8885272145271301, + "num_tokens": 189254884.0, + "step": 4868 + }, + { + "epoch": 0.6193868464571938, + "grad_norm": 2.6774845123291016, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8738710880279541, + "num_tokens": 189292652.0, + "step": 4869 + }, + { + "epoch": 0.6195140567357843, + "grad_norm": 2.567460536956787, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8802741765975952, + "num_tokens": 189328120.0, + "step": 4870 + }, + { + "epoch": 0.6196412670143747, + "grad_norm": 2.376767873764038, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9017086625099182, + "num_tokens": 189365167.0, + "step": 4871 + }, + { + "epoch": 0.6197684772929652, + "grad_norm": 2.5991086959838867, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8732719421386719, + "num_tokens": 189403556.0, + "step": 4872 + }, + { + "epoch": 0.6198956875715558, + "grad_norm": 2.6233437061309814, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.872585654258728, + "num_tokens": 189444200.0, + "step": 4873 + }, + { + "epoch": 0.6200228978501463, + "grad_norm": 2.6810553073883057, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8687834739685059, + "num_tokens": 189482714.0, + "step": 4874 + }, + { + "epoch": 0.6201501081287368, + "grad_norm": 2.460559606552124, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8832204341888428, + "num_tokens": 189522633.0, + "step": 4875 + }, + { + "epoch": 0.6202773184073274, + "grad_norm": 2.7940402030944824, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.871416449546814, + "num_tokens": 189557918.0, + "step": 4876 + }, + { + "epoch": 0.6204045286859178, + "grad_norm": 2.768862247467041, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8707196116447449, + "num_tokens": 189596413.0, + "step": 4877 + }, + { + "epoch": 0.6205317389645083, + "grad_norm": 3.127981424331665, + "learning_rate": 1e-06, + "loss": 0.4448, + "mean_token_accuracy": 0.851919412612915, + "num_tokens": 189631861.0, + "step": 4878 + }, + { + "epoch": 0.6206589492430988, + "grad_norm": 2.757934093475342, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8884379863739014, + "num_tokens": 189668248.0, + "step": 4879 + }, + { + "epoch": 0.6207861595216894, + "grad_norm": 2.4132680892944336, + "learning_rate": 1e-06, + "loss": 0.4027, + "mean_token_accuracy": 0.8636484146118164, + "num_tokens": 189715825.0, + "step": 4880 + }, + { + "epoch": 0.6209133698002799, + "grad_norm": 2.739300489425659, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8871833682060242, + "num_tokens": 189751182.0, + "step": 4881 + }, + { + "epoch": 0.6210405800788704, + "grad_norm": 2.7790679931640625, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8805395364761353, + "num_tokens": 189787807.0, + "step": 4882 + }, + { + "epoch": 0.6211677903574608, + "grad_norm": 2.9001266956329346, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.876983106136322, + "num_tokens": 189823487.0, + "step": 4883 + }, + { + "epoch": 0.6212950006360514, + "grad_norm": 2.69081974029541, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.883380651473999, + "num_tokens": 189857645.0, + "step": 4884 + }, + { + "epoch": 0.6214222109146419, + "grad_norm": 2.4544076919555664, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8775877952575684, + "num_tokens": 189897817.0, + "step": 4885 + }, + { + "epoch": 0.6215494211932324, + "grad_norm": 2.4905593395233154, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8766267895698547, + "num_tokens": 189939103.0, + "step": 4886 + }, + { + "epoch": 0.621676631471823, + "grad_norm": 2.6465253829956055, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.868470311164856, + "num_tokens": 189977861.0, + "step": 4887 + }, + { + "epoch": 0.6218038417504135, + "grad_norm": 2.6215851306915283, + "learning_rate": 1e-06, + "loss": 0.3901, + "mean_token_accuracy": 0.8658034801483154, + "num_tokens": 190016285.0, + "step": 4888 + }, + { + "epoch": 0.6219310520290039, + "grad_norm": 2.5354998111724854, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8763691186904907, + "num_tokens": 190057870.0, + "step": 4889 + }, + { + "epoch": 0.6220582623075944, + "grad_norm": 2.556572198867798, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8752233982086182, + "num_tokens": 190099723.0, + "step": 4890 + }, + { + "epoch": 0.622185472586185, + "grad_norm": 2.4004204273223877, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8730083107948303, + "num_tokens": 190143933.0, + "step": 4891 + }, + { + "epoch": 0.6223126828647755, + "grad_norm": 2.8254175186157227, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8811691403388977, + "num_tokens": 190179380.0, + "step": 4892 + }, + { + "epoch": 0.622439893143366, + "grad_norm": 2.7403202056884766, + "learning_rate": 1e-06, + "loss": 0.3833, + "mean_token_accuracy": 0.8663251399993896, + "num_tokens": 190216086.0, + "step": 4893 + }, + { + "epoch": 0.6225671034219565, + "grad_norm": 2.6453192234039307, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8887283802032471, + "num_tokens": 190253073.0, + "step": 4894 + }, + { + "epoch": 0.622694313700547, + "grad_norm": 2.592165946960449, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8819355964660645, + "num_tokens": 190291774.0, + "step": 4895 + }, + { + "epoch": 0.6228215239791375, + "grad_norm": 2.86022686958313, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8728505373001099, + "num_tokens": 190326869.0, + "step": 4896 + }, + { + "epoch": 0.622948734257728, + "grad_norm": 2.3512229919433594, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8797868490219116, + "num_tokens": 190372895.0, + "step": 4897 + }, + { + "epoch": 0.6230759445363185, + "grad_norm": 2.603607177734375, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.886768639087677, + "num_tokens": 190408233.0, + "step": 4898 + }, + { + "epoch": 0.6232031548149091, + "grad_norm": 2.6760458946228027, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8803256750106812, + "num_tokens": 190443222.0, + "step": 4899 + }, + { + "epoch": 0.6233303650934996, + "grad_norm": 2.7480554580688477, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.879928469657898, + "num_tokens": 190479487.0, + "step": 4900 + }, + { + "epoch": 0.62345757537209, + "grad_norm": 2.704493999481201, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8672133684158325, + "num_tokens": 190520311.0, + "step": 4901 + }, + { + "epoch": 0.6235847856506805, + "grad_norm": 2.592275619506836, + "learning_rate": 1e-06, + "loss": 0.3727, + "mean_token_accuracy": 0.8725274205207825, + "num_tokens": 190564863.0, + "step": 4902 + }, + { + "epoch": 0.6237119959292711, + "grad_norm": 2.780663013458252, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8731346130371094, + "num_tokens": 190603414.0, + "step": 4903 + }, + { + "epoch": 0.6238392062078616, + "grad_norm": 2.835153341293335, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.871732234954834, + "num_tokens": 190638660.0, + "step": 4904 + }, + { + "epoch": 0.6239664164864521, + "grad_norm": 2.5905423164367676, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.882672905921936, + "num_tokens": 190675814.0, + "step": 4905 + }, + { + "epoch": 0.6240936267650427, + "grad_norm": 2.535332202911377, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8895891308784485, + "num_tokens": 190716792.0, + "step": 4906 + }, + { + "epoch": 0.6242208370436332, + "grad_norm": 2.3921256065368652, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8895431756973267, + "num_tokens": 190757493.0, + "step": 4907 + }, + { + "epoch": 0.6243480473222236, + "grad_norm": 2.642498254776001, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8904072046279907, + "num_tokens": 190791481.0, + "step": 4908 + }, + { + "epoch": 0.6244752576008141, + "grad_norm": 2.934044361114502, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8914669752120972, + "num_tokens": 190821801.0, + "step": 4909 + }, + { + "epoch": 0.6246024678794047, + "grad_norm": 2.5393073558807373, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8739563226699829, + "num_tokens": 190863081.0, + "step": 4910 + }, + { + "epoch": 0.6247296781579952, + "grad_norm": 2.5859310626983643, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.8663187026977539, + "num_tokens": 190904146.0, + "step": 4911 + }, + { + "epoch": 0.6248568884365857, + "grad_norm": 2.6344940662384033, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8824694752693176, + "num_tokens": 190941637.0, + "step": 4912 + }, + { + "epoch": 0.6249840987151762, + "grad_norm": 2.576603412628174, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8793076276779175, + "num_tokens": 190978407.0, + "step": 4913 + }, + { + "epoch": 0.6251113089937667, + "grad_norm": 2.344993829727173, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8878948092460632, + "num_tokens": 191019985.0, + "step": 4914 + }, + { + "epoch": 0.6252385192723572, + "grad_norm": 2.483635902404785, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8779844045639038, + "num_tokens": 191060648.0, + "step": 4915 + }, + { + "epoch": 0.6253657295509477, + "grad_norm": 2.381756067276001, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8883209228515625, + "num_tokens": 191099809.0, + "step": 4916 + }, + { + "epoch": 0.6254929398295382, + "grad_norm": 2.7934343814849854, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8673114776611328, + "num_tokens": 191133605.0, + "step": 4917 + }, + { + "epoch": 0.6256201501081288, + "grad_norm": 2.4428460597991943, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8799092769622803, + "num_tokens": 191178556.0, + "step": 4918 + }, + { + "epoch": 0.6257473603867193, + "grad_norm": 2.657674789428711, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8663958311080933, + "num_tokens": 191217293.0, + "step": 4919 + }, + { + "epoch": 0.6258745706653097, + "grad_norm": 3.021049976348877, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8641685247421265, + "num_tokens": 191253706.0, + "step": 4920 + }, + { + "epoch": 0.6260017809439002, + "grad_norm": 2.3736612796783447, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8890079259872437, + "num_tokens": 191297053.0, + "step": 4921 + }, + { + "epoch": 0.6261289912224908, + "grad_norm": 2.472663402557373, + "learning_rate": 1e-06, + "loss": 0.387, + "mean_token_accuracy": 0.8678362369537354, + "num_tokens": 191340179.0, + "step": 4922 + }, + { + "epoch": 0.6262562015010813, + "grad_norm": 2.8894455432891846, + "learning_rate": 1e-06, + "loss": 0.3916, + "mean_token_accuracy": 0.8674220442771912, + "num_tokens": 191378012.0, + "step": 4923 + }, + { + "epoch": 0.6263834117796718, + "grad_norm": 2.848552703857422, + "learning_rate": 1e-06, + "loss": 0.3876, + "mean_token_accuracy": 0.8670732378959656, + "num_tokens": 191414999.0, + "step": 4924 + }, + { + "epoch": 0.6265106220582624, + "grad_norm": 2.6024510860443115, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8731569051742554, + "num_tokens": 191456337.0, + "step": 4925 + }, + { + "epoch": 0.6266378323368528, + "grad_norm": 2.556551694869995, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8810652494430542, + "num_tokens": 191494505.0, + "step": 4926 + }, + { + "epoch": 0.6267650426154433, + "grad_norm": 2.589289665222168, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8732509016990662, + "num_tokens": 191535316.0, + "step": 4927 + }, + { + "epoch": 0.6268922528940338, + "grad_norm": 2.5318546295166016, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8782845139503479, + "num_tokens": 191576200.0, + "step": 4928 + }, + { + "epoch": 0.6270194631726244, + "grad_norm": 2.595174789428711, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8871057033538818, + "num_tokens": 191612200.0, + "step": 4929 + }, + { + "epoch": 0.6271466734512149, + "grad_norm": 2.809365749359131, + "learning_rate": 1e-06, + "loss": 0.444, + "mean_token_accuracy": 0.8496330976486206, + "num_tokens": 191653639.0, + "step": 4930 + }, + { + "epoch": 0.6272738837298054, + "grad_norm": 2.477146863937378, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8710063695907593, + "num_tokens": 191700038.0, + "step": 4931 + }, + { + "epoch": 0.6274010940083958, + "grad_norm": 2.837538957595825, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8866345882415771, + "num_tokens": 191728023.0, + "step": 4932 + }, + { + "epoch": 0.6275283042869864, + "grad_norm": 2.7544925212860107, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8752210140228271, + "num_tokens": 191765961.0, + "step": 4933 + }, + { + "epoch": 0.6276555145655769, + "grad_norm": 2.6464571952819824, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8854048252105713, + "num_tokens": 191801402.0, + "step": 4934 + }, + { + "epoch": 0.6277827248441674, + "grad_norm": 2.643671751022339, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.868632972240448, + "num_tokens": 191842165.0, + "step": 4935 + }, + { + "epoch": 0.627909935122758, + "grad_norm": 2.6850507259368896, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8813728094100952, + "num_tokens": 191877903.0, + "step": 4936 + }, + { + "epoch": 0.6280371454013485, + "grad_norm": 2.686661958694458, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8708042502403259, + "num_tokens": 191913876.0, + "step": 4937 + }, + { + "epoch": 0.6281643556799389, + "grad_norm": 3.3213682174682617, + "learning_rate": 1e-06, + "loss": 0.432, + "mean_token_accuracy": 0.8585029244422913, + "num_tokens": 191942733.0, + "step": 4938 + }, + { + "epoch": 0.6282915659585294, + "grad_norm": 2.672215700149536, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8781835436820984, + "num_tokens": 191978034.0, + "step": 4939 + }, + { + "epoch": 0.62841877623712, + "grad_norm": 2.637500047683716, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.882124125957489, + "num_tokens": 192016675.0, + "step": 4940 + }, + { + "epoch": 0.6285459865157105, + "grad_norm": 2.6268222332000732, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8718376159667969, + "num_tokens": 192056137.0, + "step": 4941 + }, + { + "epoch": 0.628673196794301, + "grad_norm": 2.5429084300994873, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8776557445526123, + "num_tokens": 192095427.0, + "step": 4942 + }, + { + "epoch": 0.6288004070728915, + "grad_norm": 2.6980342864990234, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8833187818527222, + "num_tokens": 192129650.0, + "step": 4943 + }, + { + "epoch": 0.628927617351482, + "grad_norm": 2.9148237705230713, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8720409274101257, + "num_tokens": 192160988.0, + "step": 4944 + }, + { + "epoch": 0.6290548276300725, + "grad_norm": 2.6910691261291504, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8818603157997131, + "num_tokens": 192198761.0, + "step": 4945 + }, + { + "epoch": 0.629182037908663, + "grad_norm": 2.4723408222198486, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.888679027557373, + "num_tokens": 192237563.0, + "step": 4946 + }, + { + "epoch": 0.6293092481872535, + "grad_norm": 2.554551124572754, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.878949761390686, + "num_tokens": 192279095.0, + "step": 4947 + }, + { + "epoch": 0.6294364584658441, + "grad_norm": 2.801521062850952, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8732832074165344, + "num_tokens": 192314494.0, + "step": 4948 + }, + { + "epoch": 0.6295636687444346, + "grad_norm": 2.931525468826294, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8785021901130676, + "num_tokens": 192350142.0, + "step": 4949 + }, + { + "epoch": 0.629690879023025, + "grad_norm": 2.6947288513183594, + "learning_rate": 1e-06, + "loss": 0.3965, + "mean_token_accuracy": 0.8623279929161072, + "num_tokens": 192392436.0, + "step": 4950 + }, + { + "epoch": 0.6298180893016155, + "grad_norm": 2.5813050270080566, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8808119893074036, + "num_tokens": 192431294.0, + "step": 4951 + }, + { + "epoch": 0.6299452995802061, + "grad_norm": 2.656538486480713, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8830173015594482, + "num_tokens": 192467705.0, + "step": 4952 + }, + { + "epoch": 0.6300725098587966, + "grad_norm": 2.725952386856079, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8777343034744263, + "num_tokens": 192504716.0, + "step": 4953 + }, + { + "epoch": 0.6301997201373871, + "grad_norm": 2.669543504714966, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.883604884147644, + "num_tokens": 192544069.0, + "step": 4954 + }, + { + "epoch": 0.6303269304159776, + "grad_norm": 2.357783317565918, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8816432356834412, + "num_tokens": 192586316.0, + "step": 4955 + }, + { + "epoch": 0.6304541406945681, + "grad_norm": 2.620819568634033, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8782490491867065, + "num_tokens": 192627208.0, + "step": 4956 + }, + { + "epoch": 0.6305813509731586, + "grad_norm": 2.5521485805511475, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8797860741615295, + "num_tokens": 192667203.0, + "step": 4957 + }, + { + "epoch": 0.6307085612517491, + "grad_norm": 2.5042965412139893, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8890286087989807, + "num_tokens": 192703977.0, + "step": 4958 + }, + { + "epoch": 0.6308357715303397, + "grad_norm": 2.539964437484741, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8870136737823486, + "num_tokens": 192742011.0, + "step": 4959 + }, + { + "epoch": 0.6309629818089302, + "grad_norm": 2.829249858856201, + "learning_rate": 1e-06, + "loss": 0.3884, + "mean_token_accuracy": 0.8660465478897095, + "num_tokens": 192779047.0, + "step": 4960 + }, + { + "epoch": 0.6310901920875207, + "grad_norm": 2.2459418773651123, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.87336266040802, + "num_tokens": 192833184.0, + "step": 4961 + }, + { + "epoch": 0.6312174023661112, + "grad_norm": 2.626016855239868, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8728766441345215, + "num_tokens": 192873292.0, + "step": 4962 + }, + { + "epoch": 0.6313446126447017, + "grad_norm": 2.35941219329834, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8809781074523926, + "num_tokens": 192919518.0, + "step": 4963 + }, + { + "epoch": 0.6314718229232922, + "grad_norm": 2.6627559661865234, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8720026016235352, + "num_tokens": 192956520.0, + "step": 4964 + }, + { + "epoch": 0.6315990332018827, + "grad_norm": 2.4831459522247314, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8734398484230042, + "num_tokens": 193000890.0, + "step": 4965 + }, + { + "epoch": 0.6317262434804732, + "grad_norm": 2.5613784790039062, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8758755922317505, + "num_tokens": 193038487.0, + "step": 4966 + }, + { + "epoch": 0.6318534537590638, + "grad_norm": 2.692692756652832, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8696933388710022, + "num_tokens": 193075105.0, + "step": 4967 + }, + { + "epoch": 0.6319806640376543, + "grad_norm": 2.525808811187744, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8715718984603882, + "num_tokens": 193115029.0, + "step": 4968 + }, + { + "epoch": 0.6321078743162447, + "grad_norm": 2.555912971496582, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8720247149467468, + "num_tokens": 193156602.0, + "step": 4969 + }, + { + "epoch": 0.6322350845948352, + "grad_norm": 2.4986841678619385, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8752685785293579, + "num_tokens": 193196103.0, + "step": 4970 + }, + { + "epoch": 0.6323622948734258, + "grad_norm": 2.4537949562072754, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8678779006004333, + "num_tokens": 193240231.0, + "step": 4971 + }, + { + "epoch": 0.6324895051520163, + "grad_norm": 2.557913064956665, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8770803213119507, + "num_tokens": 193280888.0, + "step": 4972 + }, + { + "epoch": 0.6326167154306068, + "grad_norm": 2.5236661434173584, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8906450271606445, + "num_tokens": 193318695.0, + "step": 4973 + }, + { + "epoch": 0.6327439257091974, + "grad_norm": 2.618445634841919, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.864326000213623, + "num_tokens": 193358500.0, + "step": 4974 + }, + { + "epoch": 0.6328711359877878, + "grad_norm": 2.8342206478118896, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8636712431907654, + "num_tokens": 193393600.0, + "step": 4975 + }, + { + "epoch": 0.6329983462663783, + "grad_norm": 2.4805431365966797, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8761630654335022, + "num_tokens": 193435961.0, + "step": 4976 + }, + { + "epoch": 0.6331255565449688, + "grad_norm": 2.7181334495544434, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8806809782981873, + "num_tokens": 193470023.0, + "step": 4977 + }, + { + "epoch": 0.6332527668235594, + "grad_norm": 2.492828845977783, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8759728670120239, + "num_tokens": 193510977.0, + "step": 4978 + }, + { + "epoch": 0.6333799771021499, + "grad_norm": 2.394120693206787, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8783546090126038, + "num_tokens": 193556744.0, + "step": 4979 + }, + { + "epoch": 0.6335071873807404, + "grad_norm": 2.852307081222534, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8710469007492065, + "num_tokens": 193592222.0, + "step": 4980 + }, + { + "epoch": 0.6336343976593308, + "grad_norm": 2.608175754547119, + "learning_rate": 1e-06, + "loss": 0.4068, + "mean_token_accuracy": 0.859757661819458, + "num_tokens": 193633570.0, + "step": 4981 + }, + { + "epoch": 0.6337616079379214, + "grad_norm": 2.582301616668701, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8798567652702332, + "num_tokens": 193669644.0, + "step": 4982 + }, + { + "epoch": 0.6338888182165119, + "grad_norm": 2.5559587478637695, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8758077025413513, + "num_tokens": 193711367.0, + "step": 4983 + }, + { + "epoch": 0.6340160284951024, + "grad_norm": 2.7041714191436768, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8809252977371216, + "num_tokens": 193746244.0, + "step": 4984 + }, + { + "epoch": 0.6341432387736929, + "grad_norm": 2.4170711040496826, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8734139204025269, + "num_tokens": 193788688.0, + "step": 4985 + }, + { + "epoch": 0.6342704490522835, + "grad_norm": 2.4876954555511475, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8854419589042664, + "num_tokens": 193828035.0, + "step": 4986 + }, + { + "epoch": 0.6343976593308739, + "grad_norm": 2.5417189598083496, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8748689889907837, + "num_tokens": 193870005.0, + "step": 4987 + }, + { + "epoch": 0.6345248696094644, + "grad_norm": 2.903329610824585, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8729743957519531, + "num_tokens": 193909318.0, + "step": 4988 + }, + { + "epoch": 0.6346520798880549, + "grad_norm": 2.571934700012207, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8771134614944458, + "num_tokens": 193945047.0, + "step": 4989 + }, + { + "epoch": 0.6347792901666455, + "grad_norm": 2.5299735069274902, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8875405788421631, + "num_tokens": 193981325.0, + "step": 4990 + }, + { + "epoch": 0.634906500445236, + "grad_norm": 2.8007142543792725, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8733680844306946, + "num_tokens": 194018914.0, + "step": 4991 + }, + { + "epoch": 0.6350337107238265, + "grad_norm": 2.550844192504883, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8768037557601929, + "num_tokens": 194059552.0, + "step": 4992 + }, + { + "epoch": 0.635160921002417, + "grad_norm": 2.6315340995788574, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8722782731056213, + "num_tokens": 194099466.0, + "step": 4993 + }, + { + "epoch": 0.6352881312810075, + "grad_norm": 2.6814191341400146, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8708441257476807, + "num_tokens": 194136714.0, + "step": 4994 + }, + { + "epoch": 0.635415341559598, + "grad_norm": 2.554112195968628, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8742340803146362, + "num_tokens": 194178194.0, + "step": 4995 + }, + { + "epoch": 0.6355425518381885, + "grad_norm": 2.8037569522857666, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.881669282913208, + "num_tokens": 194211374.0, + "step": 4996 + }, + { + "epoch": 0.6356697621167791, + "grad_norm": 2.908374547958374, + "learning_rate": 1e-06, + "loss": 0.3978, + "mean_token_accuracy": 0.863704264163971, + "num_tokens": 194248960.0, + "step": 4997 + }, + { + "epoch": 0.6357969723953696, + "grad_norm": 2.75290846824646, + "learning_rate": 1e-06, + "loss": 0.3864, + "mean_token_accuracy": 0.8683906197547913, + "num_tokens": 194286566.0, + "step": 4998 + }, + { + "epoch": 0.63592418267396, + "grad_norm": 2.683445453643799, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.8799086213111877, + "num_tokens": 194323041.0, + "step": 4999 + }, + { + "epoch": 0.6360513929525505, + "grad_norm": 2.6426587104797363, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.877778172492981, + "num_tokens": 194359635.0, + "step": 5000 + }, + { + "epoch": 0.6361786032311411, + "grad_norm": 2.814911127090454, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8677070140838623, + "num_tokens": 194394791.0, + "step": 5001 + }, + { + "epoch": 0.6363058135097316, + "grad_norm": 2.6872918605804443, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8743726015090942, + "num_tokens": 194433943.0, + "step": 5002 + }, + { + "epoch": 0.6364330237883221, + "grad_norm": 2.7139358520507812, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8944089412689209, + "num_tokens": 194465945.0, + "step": 5003 + }, + { + "epoch": 0.6365602340669126, + "grad_norm": 2.9365618228912354, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8777503967285156, + "num_tokens": 194496174.0, + "step": 5004 + }, + { + "epoch": 0.6366874443455031, + "grad_norm": 2.7547199726104736, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8754605054855347, + "num_tokens": 194537110.0, + "step": 5005 + }, + { + "epoch": 0.6368146546240936, + "grad_norm": 2.610687017440796, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8874802589416504, + "num_tokens": 194576783.0, + "step": 5006 + }, + { + "epoch": 0.6369418649026841, + "grad_norm": 2.7254011631011963, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8827410936355591, + "num_tokens": 194615188.0, + "step": 5007 + }, + { + "epoch": 0.6370690751812746, + "grad_norm": 2.6869723796844482, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8754944801330566, + "num_tokens": 194651640.0, + "step": 5008 + }, + { + "epoch": 0.6371962854598652, + "grad_norm": 2.6282119750976562, + "learning_rate": 1e-06, + "loss": 0.4172, + "mean_token_accuracy": 0.8572189807891846, + "num_tokens": 194697257.0, + "step": 5009 + }, + { + "epoch": 0.6373234957384557, + "grad_norm": 2.6106042861938477, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8879194259643555, + "num_tokens": 194736560.0, + "step": 5010 + }, + { + "epoch": 0.6374507060170462, + "grad_norm": 2.749215841293335, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8757101893424988, + "num_tokens": 194773831.0, + "step": 5011 + }, + { + "epoch": 0.6375779162956366, + "grad_norm": 2.8080408573150635, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8738635182380676, + "num_tokens": 194815452.0, + "step": 5012 + }, + { + "epoch": 0.6377051265742272, + "grad_norm": 2.5428895950317383, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8742489814758301, + "num_tokens": 194856626.0, + "step": 5013 + }, + { + "epoch": 0.6378323368528177, + "grad_norm": 2.7465429306030273, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.8642292022705078, + "num_tokens": 194892456.0, + "step": 5014 + }, + { + "epoch": 0.6379595471314082, + "grad_norm": 2.5043299198150635, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8882201313972473, + "num_tokens": 194931462.0, + "step": 5015 + }, + { + "epoch": 0.6380867574099988, + "grad_norm": 2.999177932739258, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8817278146743774, + "num_tokens": 194960427.0, + "step": 5016 + }, + { + "epoch": 0.6382139676885893, + "grad_norm": 2.53727126121521, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8835405111312866, + "num_tokens": 194999090.0, + "step": 5017 + }, + { + "epoch": 0.6383411779671797, + "grad_norm": 2.6753580570220947, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8723539113998413, + "num_tokens": 195038442.0, + "step": 5018 + }, + { + "epoch": 0.6384683882457702, + "grad_norm": 2.7340188026428223, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.870481550693512, + "num_tokens": 195070997.0, + "step": 5019 + }, + { + "epoch": 0.6385955985243608, + "grad_norm": 2.5372936725616455, + "learning_rate": 1e-06, + "loss": 0.3715, + "mean_token_accuracy": 0.8681297898292542, + "num_tokens": 195112226.0, + "step": 5020 + }, + { + "epoch": 0.6387228088029513, + "grad_norm": 2.6122543811798096, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8784189224243164, + "num_tokens": 195151609.0, + "step": 5021 + }, + { + "epoch": 0.6388500190815418, + "grad_norm": 2.4903643131256104, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8939653038978577, + "num_tokens": 195189834.0, + "step": 5022 + }, + { + "epoch": 0.6389772293601323, + "grad_norm": 2.733872413635254, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8810901641845703, + "num_tokens": 195224095.0, + "step": 5023 + }, + { + "epoch": 0.6391044396387228, + "grad_norm": 2.60368013381958, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.8607218265533447, + "num_tokens": 195266980.0, + "step": 5024 + }, + { + "epoch": 0.6392316499173133, + "grad_norm": 2.3983631134033203, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8796689510345459, + "num_tokens": 195313239.0, + "step": 5025 + }, + { + "epoch": 0.6393588601959038, + "grad_norm": 2.708514928817749, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.863661527633667, + "num_tokens": 195349682.0, + "step": 5026 + }, + { + "epoch": 0.6394860704744944, + "grad_norm": 2.769622802734375, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8653969168663025, + "num_tokens": 195385670.0, + "step": 5027 + }, + { + "epoch": 0.6396132807530849, + "grad_norm": 2.6061670780181885, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8824857473373413, + "num_tokens": 195423838.0, + "step": 5028 + }, + { + "epoch": 0.6397404910316754, + "grad_norm": 2.6379311084747314, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8848071098327637, + "num_tokens": 195460328.0, + "step": 5029 + }, + { + "epoch": 0.6398677013102658, + "grad_norm": 2.8238461017608643, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8840720653533936, + "num_tokens": 195491730.0, + "step": 5030 + }, + { + "epoch": 0.6399949115888564, + "grad_norm": 2.41465425491333, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.889991283416748, + "num_tokens": 195533433.0, + "step": 5031 + }, + { + "epoch": 0.6401221218674469, + "grad_norm": 2.4637417793273926, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8716310262680054, + "num_tokens": 195582277.0, + "step": 5032 + }, + { + "epoch": 0.6402493321460374, + "grad_norm": 2.5514557361602783, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8820053339004517, + "num_tokens": 195621362.0, + "step": 5033 + }, + { + "epoch": 0.6403765424246279, + "grad_norm": 2.6150259971618652, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8816965818405151, + "num_tokens": 195656814.0, + "step": 5034 + }, + { + "epoch": 0.6405037527032185, + "grad_norm": 2.4076764583587646, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8746050596237183, + "num_tokens": 195701224.0, + "step": 5035 + }, + { + "epoch": 0.6406309629818089, + "grad_norm": 2.404611825942993, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8879563808441162, + "num_tokens": 195743561.0, + "step": 5036 + }, + { + "epoch": 0.6407581732603994, + "grad_norm": 2.8173129558563232, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8871015310287476, + "num_tokens": 195776363.0, + "step": 5037 + }, + { + "epoch": 0.6408853835389899, + "grad_norm": 2.4020066261291504, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8950311541557312, + "num_tokens": 195818568.0, + "step": 5038 + }, + { + "epoch": 0.6410125938175805, + "grad_norm": 2.395850658416748, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8797358274459839, + "num_tokens": 195862560.0, + "step": 5039 + }, + { + "epoch": 0.641139804096171, + "grad_norm": 3.029601812362671, + "learning_rate": 1e-06, + "loss": 0.4124, + "mean_token_accuracy": 0.8580161929130554, + "num_tokens": 195895758.0, + "step": 5040 + }, + { + "epoch": 0.6412670143747615, + "grad_norm": 2.9445064067840576, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8829042315483093, + "num_tokens": 195929227.0, + "step": 5041 + }, + { + "epoch": 0.6413942246533519, + "grad_norm": 2.785912275314331, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8702263832092285, + "num_tokens": 195965126.0, + "step": 5042 + }, + { + "epoch": 0.6415214349319425, + "grad_norm": 2.565545082092285, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8779284358024597, + "num_tokens": 196005189.0, + "step": 5043 + }, + { + "epoch": 0.641648645210533, + "grad_norm": 2.7274978160858154, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.876852810382843, + "num_tokens": 196042803.0, + "step": 5044 + }, + { + "epoch": 0.6417758554891235, + "grad_norm": 2.7282567024230957, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8775687217712402, + "num_tokens": 196078564.0, + "step": 5045 + }, + { + "epoch": 0.641903065767714, + "grad_norm": 2.4235475063323975, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8811898827552795, + "num_tokens": 196126760.0, + "step": 5046 + }, + { + "epoch": 0.6420302760463046, + "grad_norm": 2.584911823272705, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8873099684715271, + "num_tokens": 196163162.0, + "step": 5047 + }, + { + "epoch": 0.642157486324895, + "grad_norm": 2.593606472015381, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8849719762802124, + "num_tokens": 196201838.0, + "step": 5048 + }, + { + "epoch": 0.6422846966034855, + "grad_norm": 2.580228567123413, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8752045035362244, + "num_tokens": 196241505.0, + "step": 5049 + }, + { + "epoch": 0.6424119068820761, + "grad_norm": 2.7048864364624023, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8802376985549927, + "num_tokens": 196275348.0, + "step": 5050 + }, + { + "epoch": 0.6425391171606666, + "grad_norm": 2.5892691612243652, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.8760610818862915, + "num_tokens": 196314804.0, + "step": 5051 + }, + { + "epoch": 0.6426663274392571, + "grad_norm": 2.5663299560546875, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.861054539680481, + "num_tokens": 196357727.0, + "step": 5052 + }, + { + "epoch": 0.6427935377178476, + "grad_norm": 2.547351360321045, + "learning_rate": 1e-06, + "loss": 0.3803, + "mean_token_accuracy": 0.8713939189910889, + "num_tokens": 196402538.0, + "step": 5053 + }, + { + "epoch": 0.6429207479964381, + "grad_norm": 2.5972344875335693, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8692380785942078, + "num_tokens": 196443095.0, + "step": 5054 + }, + { + "epoch": 0.6430479582750286, + "grad_norm": 2.6435272693634033, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.876676082611084, + "num_tokens": 196481468.0, + "step": 5055 + }, + { + "epoch": 0.6431751685536191, + "grad_norm": 2.5532143115997314, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8779594898223877, + "num_tokens": 196521947.0, + "step": 5056 + }, + { + "epoch": 0.6433023788322096, + "grad_norm": 2.400799512863159, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8827239274978638, + "num_tokens": 196564590.0, + "step": 5057 + }, + { + "epoch": 0.6434295891108002, + "grad_norm": 2.5939619541168213, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8814345598220825, + "num_tokens": 196601197.0, + "step": 5058 + }, + { + "epoch": 0.6435567993893907, + "grad_norm": 2.4941649436950684, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8792080879211426, + "num_tokens": 196643450.0, + "step": 5059 + }, + { + "epoch": 0.6436840096679812, + "grad_norm": 2.473552703857422, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8929393291473389, + "num_tokens": 196683231.0, + "step": 5060 + }, + { + "epoch": 0.6438112199465716, + "grad_norm": 2.5587639808654785, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8813478946685791, + "num_tokens": 196722665.0, + "step": 5061 + }, + { + "epoch": 0.6439384302251622, + "grad_norm": 2.70072603225708, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8797561526298523, + "num_tokens": 196761918.0, + "step": 5062 + }, + { + "epoch": 0.6440656405037527, + "grad_norm": 2.7772936820983887, + "learning_rate": 1e-06, + "loss": 0.4158, + "mean_token_accuracy": 0.855971097946167, + "num_tokens": 196802883.0, + "step": 5063 + }, + { + "epoch": 0.6441928507823432, + "grad_norm": 2.923206090927124, + "learning_rate": 1e-06, + "loss": 0.4252, + "mean_token_accuracy": 0.852358341217041, + "num_tokens": 196838169.0, + "step": 5064 + }, + { + "epoch": 0.6443200610609338, + "grad_norm": 2.3493916988372803, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8819292783737183, + "num_tokens": 196883389.0, + "step": 5065 + }, + { + "epoch": 0.6444472713395243, + "grad_norm": 2.5479395389556885, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8876892924308777, + "num_tokens": 196922710.0, + "step": 5066 + }, + { + "epoch": 0.6445744816181147, + "grad_norm": 2.6014349460601807, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8853487968444824, + "num_tokens": 196960073.0, + "step": 5067 + }, + { + "epoch": 0.6447016918967052, + "grad_norm": 2.707543134689331, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8665698766708374, + "num_tokens": 197001367.0, + "step": 5068 + }, + { + "epoch": 0.6448289021752958, + "grad_norm": 2.32511568069458, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8791465759277344, + "num_tokens": 197046079.0, + "step": 5069 + }, + { + "epoch": 0.6449561124538863, + "grad_norm": 3.047380208969116, + "learning_rate": 1e-06, + "loss": 0.4003, + "mean_token_accuracy": 0.8611413836479187, + "num_tokens": 197078513.0, + "step": 5070 + }, + { + "epoch": 0.6450833227324768, + "grad_norm": 2.570737838745117, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.884087324142456, + "num_tokens": 197115836.0, + "step": 5071 + }, + { + "epoch": 0.6452105330110673, + "grad_norm": 2.933685064315796, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.869828999042511, + "num_tokens": 197149374.0, + "step": 5072 + }, + { + "epoch": 0.6453377432896578, + "grad_norm": 2.7233729362487793, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8757888674736023, + "num_tokens": 197186878.0, + "step": 5073 + }, + { + "epoch": 0.6454649535682483, + "grad_norm": 2.4976165294647217, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8852818012237549, + "num_tokens": 197225066.0, + "step": 5074 + }, + { + "epoch": 0.6455921638468388, + "grad_norm": 2.610356092453003, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8900339603424072, + "num_tokens": 197259052.0, + "step": 5075 + }, + { + "epoch": 0.6457193741254293, + "grad_norm": 2.5582971572875977, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8775283098220825, + "num_tokens": 197298640.0, + "step": 5076 + }, + { + "epoch": 0.6458465844040199, + "grad_norm": 2.5430431365966797, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8882818818092346, + "num_tokens": 197336084.0, + "step": 5077 + }, + { + "epoch": 0.6459737946826104, + "grad_norm": 2.413742780685425, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8874573707580566, + "num_tokens": 197376535.0, + "step": 5078 + }, + { + "epoch": 0.6461010049612008, + "grad_norm": 2.48669695854187, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8885002136230469, + "num_tokens": 197412997.0, + "step": 5079 + }, + { + "epoch": 0.6462282152397913, + "grad_norm": 2.710146903991699, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.871407151222229, + "num_tokens": 197451091.0, + "step": 5080 + }, + { + "epoch": 0.6463554255183819, + "grad_norm": 2.71071720123291, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.8682787418365479, + "num_tokens": 197487096.0, + "step": 5081 + }, + { + "epoch": 0.6464826357969724, + "grad_norm": 2.5556235313415527, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.876651406288147, + "num_tokens": 197529711.0, + "step": 5082 + }, + { + "epoch": 0.6466098460755629, + "grad_norm": 2.581113338470459, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.877521276473999, + "num_tokens": 197568887.0, + "step": 5083 + }, + { + "epoch": 0.6467370563541535, + "grad_norm": 2.555682897567749, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8754564523696899, + "num_tokens": 197611124.0, + "step": 5084 + }, + { + "epoch": 0.6468642666327439, + "grad_norm": 2.490203857421875, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8811963796615601, + "num_tokens": 197651161.0, + "step": 5085 + }, + { + "epoch": 0.6469914769113344, + "grad_norm": 2.622856855392456, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8847188949584961, + "num_tokens": 197687731.0, + "step": 5086 + }, + { + "epoch": 0.6471186871899249, + "grad_norm": 2.54144287109375, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8800904750823975, + "num_tokens": 197727086.0, + "step": 5087 + }, + { + "epoch": 0.6472458974685155, + "grad_norm": 3.1429789066314697, + "learning_rate": 1e-06, + "loss": 0.4251, + "mean_token_accuracy": 0.8543218374252319, + "num_tokens": 197763556.0, + "step": 5088 + }, + { + "epoch": 0.647373107747106, + "grad_norm": 2.5984842777252197, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.8718737363815308, + "num_tokens": 197804347.0, + "step": 5089 + }, + { + "epoch": 0.6475003180256965, + "grad_norm": 2.6551122665405273, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8739156126976013, + "num_tokens": 197841598.0, + "step": 5090 + }, + { + "epoch": 0.6476275283042869, + "grad_norm": 2.667663812637329, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8809495568275452, + "num_tokens": 197882025.0, + "step": 5091 + }, + { + "epoch": 0.6477547385828775, + "grad_norm": 2.7916088104248047, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8792169690132141, + "num_tokens": 197914375.0, + "step": 5092 + }, + { + "epoch": 0.647881948861468, + "grad_norm": 2.6128737926483154, + "learning_rate": 1e-06, + "loss": 0.4109, + "mean_token_accuracy": 0.8544972538948059, + "num_tokens": 197960333.0, + "step": 5093 + }, + { + "epoch": 0.6480091591400585, + "grad_norm": 2.6682510375976562, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8731594085693359, + "num_tokens": 198000079.0, + "step": 5094 + }, + { + "epoch": 0.648136369418649, + "grad_norm": 2.527125120162964, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8913254737854004, + "num_tokens": 198037246.0, + "step": 5095 + }, + { + "epoch": 0.6482635796972396, + "grad_norm": 2.516108274459839, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8828624486923218, + "num_tokens": 198079892.0, + "step": 5096 + }, + { + "epoch": 0.64839078997583, + "grad_norm": 2.3530330657958984, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8740958571434021, + "num_tokens": 198127490.0, + "step": 5097 + }, + { + "epoch": 0.6485180002544205, + "grad_norm": 2.839085102081299, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.8665872812271118, + "num_tokens": 198163266.0, + "step": 5098 + }, + { + "epoch": 0.648645210533011, + "grad_norm": 2.5032849311828613, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8864728212356567, + "num_tokens": 198203397.0, + "step": 5099 + }, + { + "epoch": 0.6487724208116016, + "grad_norm": 2.7348949909210205, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.871319055557251, + "num_tokens": 198239335.0, + "step": 5100 + }, + { + "epoch": 0.6488996310901921, + "grad_norm": 2.82442307472229, + "learning_rate": 1e-06, + "loss": 0.3957, + "mean_token_accuracy": 0.8620542287826538, + "num_tokens": 198277385.0, + "step": 5101 + }, + { + "epoch": 0.6490268413687826, + "grad_norm": 2.426858425140381, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8869717717170715, + "num_tokens": 198317379.0, + "step": 5102 + }, + { + "epoch": 0.649154051647373, + "grad_norm": 2.642775535583496, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8877325654029846, + "num_tokens": 198352352.0, + "step": 5103 + }, + { + "epoch": 0.6492812619259636, + "grad_norm": 2.6724705696105957, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8690107464790344, + "num_tokens": 198390463.0, + "step": 5104 + }, + { + "epoch": 0.6494084722045541, + "grad_norm": 2.513166666030884, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8823301196098328, + "num_tokens": 198429705.0, + "step": 5105 + }, + { + "epoch": 0.6495356824831446, + "grad_norm": 2.431241512298584, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8823018074035645, + "num_tokens": 198472723.0, + "step": 5106 + }, + { + "epoch": 0.6496628927617352, + "grad_norm": 2.6362502574920654, + "learning_rate": 1e-06, + "loss": 0.3672, + "mean_token_accuracy": 0.875254213809967, + "num_tokens": 198514492.0, + "step": 5107 + }, + { + "epoch": 0.6497901030403257, + "grad_norm": 2.6550133228302, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8778833150863647, + "num_tokens": 198551736.0, + "step": 5108 + }, + { + "epoch": 0.6499173133189162, + "grad_norm": 2.2805137634277344, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8802415728569031, + "num_tokens": 198598062.0, + "step": 5109 + }, + { + "epoch": 0.6500445235975066, + "grad_norm": 2.641296863555908, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8856840133666992, + "num_tokens": 198633610.0, + "step": 5110 + }, + { + "epoch": 0.6501717338760972, + "grad_norm": 2.629140853881836, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8833445906639099, + "num_tokens": 198670596.0, + "step": 5111 + }, + { + "epoch": 0.6502989441546877, + "grad_norm": 2.4057698249816895, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8925983309745789, + "num_tokens": 198709678.0, + "step": 5112 + }, + { + "epoch": 0.6504261544332782, + "grad_norm": 2.651634693145752, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8899250030517578, + "num_tokens": 198745044.0, + "step": 5113 + }, + { + "epoch": 0.6505533647118688, + "grad_norm": 2.7415263652801514, + "learning_rate": 1e-06, + "loss": 0.4019, + "mean_token_accuracy": 0.8580284118652344, + "num_tokens": 198785719.0, + "step": 5114 + }, + { + "epoch": 0.6506805749904593, + "grad_norm": 2.6456148624420166, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8684154748916626, + "num_tokens": 198824499.0, + "step": 5115 + }, + { + "epoch": 0.6508077852690497, + "grad_norm": 2.607029438018799, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8733491897583008, + "num_tokens": 198868508.0, + "step": 5116 + }, + { + "epoch": 0.6509349955476402, + "grad_norm": 2.716487169265747, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8938422203063965, + "num_tokens": 198898097.0, + "step": 5117 + }, + { + "epoch": 0.6510622058262308, + "grad_norm": 2.743072032928467, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8731878995895386, + "num_tokens": 198935764.0, + "step": 5118 + }, + { + "epoch": 0.6511894161048213, + "grad_norm": 2.5690081119537354, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8770875334739685, + "num_tokens": 198973470.0, + "step": 5119 + }, + { + "epoch": 0.6513166263834118, + "grad_norm": 2.5010359287261963, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.896127462387085, + "num_tokens": 199015032.0, + "step": 5120 + }, + { + "epoch": 0.6514438366620023, + "grad_norm": 2.7924351692199707, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8753958940505981, + "num_tokens": 199051680.0, + "step": 5121 + }, + { + "epoch": 0.6515710469405928, + "grad_norm": 2.4964540004730225, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8800978660583496, + "num_tokens": 199091683.0, + "step": 5122 + }, + { + "epoch": 0.6516982572191833, + "grad_norm": 2.3668670654296875, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8908265233039856, + "num_tokens": 199133733.0, + "step": 5123 + }, + { + "epoch": 0.6518254674977738, + "grad_norm": 2.5332579612731934, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8726045489311218, + "num_tokens": 199176116.0, + "step": 5124 + }, + { + "epoch": 0.6519526777763643, + "grad_norm": 2.598409652709961, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.876718282699585, + "num_tokens": 199214570.0, + "step": 5125 + }, + { + "epoch": 0.6520798880549549, + "grad_norm": 2.536423921585083, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8851829767227173, + "num_tokens": 199255529.0, + "step": 5126 + }, + { + "epoch": 0.6522070983335454, + "grad_norm": 2.7501304149627686, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.871641218662262, + "num_tokens": 199293611.0, + "step": 5127 + }, + { + "epoch": 0.6523343086121358, + "grad_norm": 2.7873313426971436, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8829624652862549, + "num_tokens": 199330378.0, + "step": 5128 + }, + { + "epoch": 0.6524615188907263, + "grad_norm": 2.3738269805908203, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8984609842300415, + "num_tokens": 199368558.0, + "step": 5129 + }, + { + "epoch": 0.6525887291693169, + "grad_norm": 2.7793490886688232, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8753103613853455, + "num_tokens": 199406104.0, + "step": 5130 + }, + { + "epoch": 0.6527159394479074, + "grad_norm": 2.7327370643615723, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8737017512321472, + "num_tokens": 199444718.0, + "step": 5131 + }, + { + "epoch": 0.6528431497264979, + "grad_norm": 2.7255051136016846, + "learning_rate": 1e-06, + "loss": 0.4177, + "mean_token_accuracy": 0.8554002046585083, + "num_tokens": 199482057.0, + "step": 5132 + }, + { + "epoch": 0.6529703600050885, + "grad_norm": 2.5160539150238037, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8772128224372864, + "num_tokens": 199521270.0, + "step": 5133 + }, + { + "epoch": 0.6530975702836789, + "grad_norm": 2.545544147491455, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8733731508255005, + "num_tokens": 199561160.0, + "step": 5134 + }, + { + "epoch": 0.6532247805622694, + "grad_norm": 2.8839821815490723, + "learning_rate": 1e-06, + "loss": 0.4111, + "mean_token_accuracy": 0.8636505007743835, + "num_tokens": 199598134.0, + "step": 5135 + }, + { + "epoch": 0.6533519908408599, + "grad_norm": 2.759352684020996, + "learning_rate": 1e-06, + "loss": 0.3895, + "mean_token_accuracy": 0.8642095327377319, + "num_tokens": 199637418.0, + "step": 5136 + }, + { + "epoch": 0.6534792011194505, + "grad_norm": 2.5038633346557617, + "learning_rate": 1e-06, + "loss": 0.3798, + "mean_token_accuracy": 0.8721075057983398, + "num_tokens": 199679261.0, + "step": 5137 + }, + { + "epoch": 0.653606411398041, + "grad_norm": 2.8291749954223633, + "learning_rate": 1e-06, + "loss": 0.403, + "mean_token_accuracy": 0.8598675727844238, + "num_tokens": 199718069.0, + "step": 5138 + }, + { + "epoch": 0.6537336216766315, + "grad_norm": 2.760754108428955, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8760636448860168, + "num_tokens": 199753160.0, + "step": 5139 + }, + { + "epoch": 0.6538608319552219, + "grad_norm": 2.593660593032837, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8757567405700684, + "num_tokens": 199790373.0, + "step": 5140 + }, + { + "epoch": 0.6539880422338125, + "grad_norm": 2.618624687194824, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8796247243881226, + "num_tokens": 199828789.0, + "step": 5141 + }, + { + "epoch": 0.654115252512403, + "grad_norm": 2.5089099407196045, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8758140206336975, + "num_tokens": 199868852.0, + "step": 5142 + }, + { + "epoch": 0.6542424627909935, + "grad_norm": 2.5791571140289307, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8715060353279114, + "num_tokens": 199912906.0, + "step": 5143 + }, + { + "epoch": 0.654369673069584, + "grad_norm": 2.7108266353607178, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.883952260017395, + "num_tokens": 199946625.0, + "step": 5144 + }, + { + "epoch": 0.6544968833481746, + "grad_norm": 2.695866107940674, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8758294582366943, + "num_tokens": 199982309.0, + "step": 5145 + }, + { + "epoch": 0.654624093626765, + "grad_norm": 2.4714176654815674, + "learning_rate": 1e-06, + "loss": 0.393, + "mean_token_accuracy": 0.8654809594154358, + "num_tokens": 200029681.0, + "step": 5146 + }, + { + "epoch": 0.6547513039053555, + "grad_norm": 2.6001110076904297, + "learning_rate": 1e-06, + "loss": 0.3969, + "mean_token_accuracy": 0.8666884899139404, + "num_tokens": 200070998.0, + "step": 5147 + }, + { + "epoch": 0.654878514183946, + "grad_norm": 2.6728250980377197, + "learning_rate": 1e-06, + "loss": 0.391, + "mean_token_accuracy": 0.8647119402885437, + "num_tokens": 200110287.0, + "step": 5148 + }, + { + "epoch": 0.6550057244625366, + "grad_norm": 2.9217894077301025, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8596925139427185, + "num_tokens": 200143240.0, + "step": 5149 + }, + { + "epoch": 0.6551329347411271, + "grad_norm": 2.9717230796813965, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8766252398490906, + "num_tokens": 200181322.0, + "step": 5150 + }, + { + "epoch": 0.6552601450197176, + "grad_norm": 2.5107297897338867, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8825446963310242, + "num_tokens": 200221176.0, + "step": 5151 + }, + { + "epoch": 0.655387355298308, + "grad_norm": 2.794776439666748, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8840792775154114, + "num_tokens": 200252261.0, + "step": 5152 + }, + { + "epoch": 0.6555145655768986, + "grad_norm": 2.4101243019104004, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8825270533561707, + "num_tokens": 200294197.0, + "step": 5153 + }, + { + "epoch": 0.6556417758554891, + "grad_norm": 2.605119228363037, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8727208971977234, + "num_tokens": 200332966.0, + "step": 5154 + }, + { + "epoch": 0.6557689861340796, + "grad_norm": 2.480388641357422, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8768423795700073, + "num_tokens": 200374768.0, + "step": 5155 + }, + { + "epoch": 0.6558961964126702, + "grad_norm": 2.5603232383728027, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.8675413131713867, + "num_tokens": 200417694.0, + "step": 5156 + }, + { + "epoch": 0.6560234066912607, + "grad_norm": 2.5814743041992188, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8769166469573975, + "num_tokens": 200459220.0, + "step": 5157 + }, + { + "epoch": 0.6561506169698512, + "grad_norm": 2.5324127674102783, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8890846967697144, + "num_tokens": 200495403.0, + "step": 5158 + }, + { + "epoch": 0.6562778272484416, + "grad_norm": 2.6485795974731445, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8802813291549683, + "num_tokens": 200533787.0, + "step": 5159 + }, + { + "epoch": 0.6564050375270322, + "grad_norm": 3.1045329570770264, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8687058091163635, + "num_tokens": 200567002.0, + "step": 5160 + }, + { + "epoch": 0.6565322478056227, + "grad_norm": 2.577744960784912, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8830323219299316, + "num_tokens": 200608300.0, + "step": 5161 + }, + { + "epoch": 0.6566594580842132, + "grad_norm": 2.750993490219116, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8672964572906494, + "num_tokens": 200648545.0, + "step": 5162 + }, + { + "epoch": 0.6567866683628037, + "grad_norm": 2.5811550617218018, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.8644075393676758, + "num_tokens": 200693430.0, + "step": 5163 + }, + { + "epoch": 0.6569138786413943, + "grad_norm": 2.7052595615386963, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.878410816192627, + "num_tokens": 200729308.0, + "step": 5164 + }, + { + "epoch": 0.6570410889199847, + "grad_norm": 2.75976824760437, + "learning_rate": 1e-06, + "loss": 0.3944, + "mean_token_accuracy": 0.8600022792816162, + "num_tokens": 200766292.0, + "step": 5165 + }, + { + "epoch": 0.6571682991985752, + "grad_norm": 2.563563585281372, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8835306167602539, + "num_tokens": 200805890.0, + "step": 5166 + }, + { + "epoch": 0.6572955094771658, + "grad_norm": 2.6982948780059814, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8777517080307007, + "num_tokens": 200840262.0, + "step": 5167 + }, + { + "epoch": 0.6574227197557563, + "grad_norm": 2.472933769226074, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8791204690933228, + "num_tokens": 200880092.0, + "step": 5168 + }, + { + "epoch": 0.6575499300343468, + "grad_norm": 3.0792717933654785, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8724914789199829, + "num_tokens": 200909409.0, + "step": 5169 + }, + { + "epoch": 0.6576771403129373, + "grad_norm": 2.6277546882629395, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8928084373474121, + "num_tokens": 200942694.0, + "step": 5170 + }, + { + "epoch": 0.6578043505915278, + "grad_norm": 2.3974690437316895, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8909938335418701, + "num_tokens": 200983701.0, + "step": 5171 + }, + { + "epoch": 0.6579315608701183, + "grad_norm": 2.455641984939575, + "learning_rate": 1e-06, + "loss": 0.3839, + "mean_token_accuracy": 0.8665590286254883, + "num_tokens": 201025945.0, + "step": 5172 + }, + { + "epoch": 0.6580587711487088, + "grad_norm": 2.634211301803589, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.873559832572937, + "num_tokens": 201065636.0, + "step": 5173 + }, + { + "epoch": 0.6581859814272993, + "grad_norm": 2.5907318592071533, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8767827153205872, + "num_tokens": 201109587.0, + "step": 5174 + }, + { + "epoch": 0.6583131917058899, + "grad_norm": 2.572866439819336, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8813008666038513, + "num_tokens": 201146423.0, + "step": 5175 + }, + { + "epoch": 0.6584404019844804, + "grad_norm": 2.826650381088257, + "learning_rate": 1e-06, + "loss": 0.4058, + "mean_token_accuracy": 0.857650637626648, + "num_tokens": 201183125.0, + "step": 5176 + }, + { + "epoch": 0.6585676122630708, + "grad_norm": 2.5366568565368652, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8803039789199829, + "num_tokens": 201223642.0, + "step": 5177 + }, + { + "epoch": 0.6586948225416613, + "grad_norm": 2.567971706390381, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8786681890487671, + "num_tokens": 201264784.0, + "step": 5178 + }, + { + "epoch": 0.6588220328202519, + "grad_norm": 2.393733024597168, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8842529058456421, + "num_tokens": 201305544.0, + "step": 5179 + }, + { + "epoch": 0.6589492430988424, + "grad_norm": 2.594230890274048, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.8671218156814575, + "num_tokens": 201348726.0, + "step": 5180 + }, + { + "epoch": 0.6590764533774329, + "grad_norm": 2.5368642807006836, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8751173615455627, + "num_tokens": 201389329.0, + "step": 5181 + }, + { + "epoch": 0.6592036636560235, + "grad_norm": 2.7675869464874268, + "learning_rate": 1e-06, + "loss": 0.4108, + "mean_token_accuracy": 0.8595792055130005, + "num_tokens": 201431624.0, + "step": 5182 + }, + { + "epoch": 0.6593308739346139, + "grad_norm": 2.7340097427368164, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8732849359512329, + "num_tokens": 201470688.0, + "step": 5183 + }, + { + "epoch": 0.6594580842132044, + "grad_norm": 2.826800584793091, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8719305992126465, + "num_tokens": 201504394.0, + "step": 5184 + }, + { + "epoch": 0.6595852944917949, + "grad_norm": 2.765442371368408, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8773062825202942, + "num_tokens": 201541394.0, + "step": 5185 + }, + { + "epoch": 0.6597125047703855, + "grad_norm": 2.7846014499664307, + "learning_rate": 1e-06, + "loss": 0.4031, + "mean_token_accuracy": 0.8581476211547852, + "num_tokens": 201579784.0, + "step": 5186 + }, + { + "epoch": 0.659839715048976, + "grad_norm": 7.746676445007324, + "learning_rate": 1e-06, + "loss": 0.4062, + "mean_token_accuracy": 0.8652814626693726, + "num_tokens": 201618386.0, + "step": 5187 + }, + { + "epoch": 0.6599669253275665, + "grad_norm": 2.595860719680786, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8763982653617859, + "num_tokens": 201654376.0, + "step": 5188 + }, + { + "epoch": 0.6600941356061569, + "grad_norm": 2.860743522644043, + "learning_rate": 1e-06, + "loss": 0.3947, + "mean_token_accuracy": 0.8614335656166077, + "num_tokens": 201691264.0, + "step": 5189 + }, + { + "epoch": 0.6602213458847475, + "grad_norm": 2.691751003265381, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.86134934425354, + "num_tokens": 201737651.0, + "step": 5190 + }, + { + "epoch": 0.660348556163338, + "grad_norm": 2.676570415496826, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8897098898887634, + "num_tokens": 201771594.0, + "step": 5191 + }, + { + "epoch": 0.6604757664419285, + "grad_norm": 2.3966753482818604, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8914289474487305, + "num_tokens": 201809975.0, + "step": 5192 + }, + { + "epoch": 0.660602976720519, + "grad_norm": 2.3050429821014404, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8859142065048218, + "num_tokens": 201856631.0, + "step": 5193 + }, + { + "epoch": 0.6607301869991096, + "grad_norm": 2.458984136581421, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8913559913635254, + "num_tokens": 201897242.0, + "step": 5194 + }, + { + "epoch": 0.6608573972777, + "grad_norm": 2.683100700378418, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8712396621704102, + "num_tokens": 201933907.0, + "step": 5195 + }, + { + "epoch": 0.6609846075562905, + "grad_norm": 2.4013946056365967, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8867015242576599, + "num_tokens": 201976228.0, + "step": 5196 + }, + { + "epoch": 0.661111817834881, + "grad_norm": 2.7544517517089844, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8788745403289795, + "num_tokens": 202008770.0, + "step": 5197 + }, + { + "epoch": 0.6612390281134716, + "grad_norm": 2.526026964187622, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.894516110420227, + "num_tokens": 202044408.0, + "step": 5198 + }, + { + "epoch": 0.6613662383920621, + "grad_norm": 2.8496639728546143, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8693915605545044, + "num_tokens": 202081548.0, + "step": 5199 + }, + { + "epoch": 0.6614934486706526, + "grad_norm": 2.613766670227051, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8945293426513672, + "num_tokens": 202115476.0, + "step": 5200 + }, + { + "epoch": 0.661620658949243, + "grad_norm": 2.4658877849578857, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8745098114013672, + "num_tokens": 202157555.0, + "step": 5201 + }, + { + "epoch": 0.6617478692278336, + "grad_norm": 2.6910202503204346, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8841261267662048, + "num_tokens": 202195148.0, + "step": 5202 + }, + { + "epoch": 0.6618750795064241, + "grad_norm": 2.7481744289398193, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8780002593994141, + "num_tokens": 202231971.0, + "step": 5203 + }, + { + "epoch": 0.6620022897850146, + "grad_norm": 2.76702880859375, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.8769619464874268, + "num_tokens": 202269616.0, + "step": 5204 + }, + { + "epoch": 0.6621295000636052, + "grad_norm": 2.679349184036255, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8689144849777222, + "num_tokens": 202310337.0, + "step": 5205 + }, + { + "epoch": 0.6622567103421957, + "grad_norm": 2.451828718185425, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8845144510269165, + "num_tokens": 202351218.0, + "step": 5206 + }, + { + "epoch": 0.6623839206207861, + "grad_norm": 2.6730620861053467, + "learning_rate": 1e-06, + "loss": 0.3972, + "mean_token_accuracy": 0.8644044399261475, + "num_tokens": 202395082.0, + "step": 5207 + }, + { + "epoch": 0.6625111308993766, + "grad_norm": 2.3866193294525146, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8786800503730774, + "num_tokens": 202438583.0, + "step": 5208 + }, + { + "epoch": 0.6626383411779672, + "grad_norm": 2.5497422218322754, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8828094005584717, + "num_tokens": 202476017.0, + "step": 5209 + }, + { + "epoch": 0.6627655514565577, + "grad_norm": 2.6829710006713867, + "learning_rate": 1e-06, + "loss": 0.3676, + "mean_token_accuracy": 0.8725306987762451, + "num_tokens": 202511998.0, + "step": 5210 + }, + { + "epoch": 0.6628927617351482, + "grad_norm": 2.784564256668091, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8770965337753296, + "num_tokens": 202549657.0, + "step": 5211 + }, + { + "epoch": 0.6630199720137387, + "grad_norm": 2.6959776878356934, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8708535432815552, + "num_tokens": 202590989.0, + "step": 5212 + }, + { + "epoch": 0.6631471822923293, + "grad_norm": 2.533567190170288, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8741375207901001, + "num_tokens": 202632959.0, + "step": 5213 + }, + { + "epoch": 0.6632743925709197, + "grad_norm": 2.6926023960113525, + "learning_rate": 1e-06, + "loss": 0.3821, + "mean_token_accuracy": 0.8646982908248901, + "num_tokens": 202670377.0, + "step": 5214 + }, + { + "epoch": 0.6634016028495102, + "grad_norm": 2.752315044403076, + "learning_rate": 1e-06, + "loss": 0.3557, + "mean_token_accuracy": 0.87773197889328, + "num_tokens": 202705091.0, + "step": 5215 + }, + { + "epoch": 0.6635288131281007, + "grad_norm": 2.3570351600646973, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8861619234085083, + "num_tokens": 202749152.0, + "step": 5216 + }, + { + "epoch": 0.6636560234066913, + "grad_norm": 2.5372159481048584, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8795458078384399, + "num_tokens": 202788160.0, + "step": 5217 + }, + { + "epoch": 0.6637832336852818, + "grad_norm": 2.612300395965576, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.884860098361969, + "num_tokens": 202826495.0, + "step": 5218 + }, + { + "epoch": 0.6639104439638723, + "grad_norm": 2.6305058002471924, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8742674589157104, + "num_tokens": 202867329.0, + "step": 5219 + }, + { + "epoch": 0.6640376542424627, + "grad_norm": 2.5655179023742676, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8753951787948608, + "num_tokens": 202909452.0, + "step": 5220 + }, + { + "epoch": 0.6641648645210533, + "grad_norm": 2.6680281162261963, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8738750219345093, + "num_tokens": 202945525.0, + "step": 5221 + }, + { + "epoch": 0.6642920747996438, + "grad_norm": 2.519109010696411, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8644187450408936, + "num_tokens": 202991034.0, + "step": 5222 + }, + { + "epoch": 0.6644192850782343, + "grad_norm": 2.2720370292663574, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.882583737373352, + "num_tokens": 203037060.0, + "step": 5223 + }, + { + "epoch": 0.6645464953568249, + "grad_norm": 2.81022047996521, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.878261387348175, + "num_tokens": 203070573.0, + "step": 5224 + }, + { + "epoch": 0.6646737056354154, + "grad_norm": 2.4768033027648926, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8744283318519592, + "num_tokens": 203113814.0, + "step": 5225 + }, + { + "epoch": 0.6648009159140058, + "grad_norm": 2.624513626098633, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.87587571144104, + "num_tokens": 203155765.0, + "step": 5226 + }, + { + "epoch": 0.6649281261925963, + "grad_norm": 2.835932731628418, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8727447390556335, + "num_tokens": 203189331.0, + "step": 5227 + }, + { + "epoch": 0.6650553364711869, + "grad_norm": 2.7358169555664062, + "learning_rate": 1e-06, + "loss": 0.3746, + "mean_token_accuracy": 0.8705732822418213, + "num_tokens": 203225658.0, + "step": 5228 + }, + { + "epoch": 0.6651825467497774, + "grad_norm": 2.7073330879211426, + "learning_rate": 1e-06, + "loss": 0.3986, + "mean_token_accuracy": 0.8651140928268433, + "num_tokens": 203265328.0, + "step": 5229 + }, + { + "epoch": 0.6653097570283679, + "grad_norm": 2.714139461517334, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8789454102516174, + "num_tokens": 203301866.0, + "step": 5230 + }, + { + "epoch": 0.6654369673069584, + "grad_norm": 2.5654985904693604, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.8708466291427612, + "num_tokens": 203344328.0, + "step": 5231 + }, + { + "epoch": 0.6655641775855489, + "grad_norm": 2.500474691390991, + "learning_rate": 1e-06, + "loss": 0.3912, + "mean_token_accuracy": 0.8628338575363159, + "num_tokens": 203387522.0, + "step": 5232 + }, + { + "epoch": 0.6656913878641394, + "grad_norm": 2.367166042327881, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8866317272186279, + "num_tokens": 203430858.0, + "step": 5233 + }, + { + "epoch": 0.6658185981427299, + "grad_norm": 2.629028797149658, + "learning_rate": 1e-06, + "loss": 0.4091, + "mean_token_accuracy": 0.8602215051651001, + "num_tokens": 203470188.0, + "step": 5234 + }, + { + "epoch": 0.6659458084213205, + "grad_norm": 2.6629300117492676, + "learning_rate": 1e-06, + "loss": 0.3903, + "mean_token_accuracy": 0.8674062490463257, + "num_tokens": 203510966.0, + "step": 5235 + }, + { + "epoch": 0.666073018699911, + "grad_norm": 2.7539265155792236, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8746214509010315, + "num_tokens": 203551270.0, + "step": 5236 + }, + { + "epoch": 0.6662002289785015, + "grad_norm": 2.495120048522949, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8827034831047058, + "num_tokens": 203589292.0, + "step": 5237 + }, + { + "epoch": 0.6663274392570919, + "grad_norm": 2.8720128536224365, + "learning_rate": 1e-06, + "loss": 0.3974, + "mean_token_accuracy": 0.8570730090141296, + "num_tokens": 203625495.0, + "step": 5238 + }, + { + "epoch": 0.6664546495356825, + "grad_norm": 2.52681040763855, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8746013641357422, + "num_tokens": 203667977.0, + "step": 5239 + }, + { + "epoch": 0.666581859814273, + "grad_norm": 2.5801069736480713, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8749452829360962, + "num_tokens": 203710498.0, + "step": 5240 + }, + { + "epoch": 0.6667090700928635, + "grad_norm": 2.3405587673187256, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8910659551620483, + "num_tokens": 203750890.0, + "step": 5241 + }, + { + "epoch": 0.666836280371454, + "grad_norm": 2.689234972000122, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8850481510162354, + "num_tokens": 203785270.0, + "step": 5242 + }, + { + "epoch": 0.6669634906500446, + "grad_norm": 2.427676200866699, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8822472095489502, + "num_tokens": 203828928.0, + "step": 5243 + }, + { + "epoch": 0.667090700928635, + "grad_norm": 2.851158380508423, + "learning_rate": 1e-06, + "loss": 0.3938, + "mean_token_accuracy": 0.8603343963623047, + "num_tokens": 203863726.0, + "step": 5244 + }, + { + "epoch": 0.6672179112072255, + "grad_norm": 2.7154898643493652, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8783267736434937, + "num_tokens": 203897959.0, + "step": 5245 + }, + { + "epoch": 0.667345121485816, + "grad_norm": 2.318310022354126, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8944361209869385, + "num_tokens": 203941028.0, + "step": 5246 + }, + { + "epoch": 0.6674723317644066, + "grad_norm": 2.7994658946990967, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8878651261329651, + "num_tokens": 203970525.0, + "step": 5247 + }, + { + "epoch": 0.6675995420429971, + "grad_norm": 2.7015883922576904, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8755230903625488, + "num_tokens": 204006844.0, + "step": 5248 + }, + { + "epoch": 0.6677267523215876, + "grad_norm": 2.6724352836608887, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8839320540428162, + "num_tokens": 204044880.0, + "step": 5249 + }, + { + "epoch": 0.667853962600178, + "grad_norm": 2.5669612884521484, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8852595686912537, + "num_tokens": 204085052.0, + "step": 5250 + }, + { + "epoch": 0.6679811728787686, + "grad_norm": 2.667020082473755, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.879310667514801, + "num_tokens": 204124579.0, + "step": 5251 + }, + { + "epoch": 0.6681083831573591, + "grad_norm": 2.56107234954834, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8862388134002686, + "num_tokens": 204162712.0, + "step": 5252 + }, + { + "epoch": 0.6682355934359496, + "grad_norm": 2.4009850025177, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8762460350990295, + "num_tokens": 204210620.0, + "step": 5253 + }, + { + "epoch": 0.6683628037145402, + "grad_norm": 2.8350040912628174, + "learning_rate": 1e-06, + "loss": 0.4197, + "mean_token_accuracy": 0.8574920296669006, + "num_tokens": 204247591.0, + "step": 5254 + }, + { + "epoch": 0.6684900139931307, + "grad_norm": 2.472825288772583, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8850763440132141, + "num_tokens": 204285193.0, + "step": 5255 + }, + { + "epoch": 0.6686172242717211, + "grad_norm": 2.714755058288574, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8762169480323792, + "num_tokens": 204324130.0, + "step": 5256 + }, + { + "epoch": 0.6687444345503116, + "grad_norm": 2.523560047149658, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.888623833656311, + "num_tokens": 204361486.0, + "step": 5257 + }, + { + "epoch": 0.6688716448289022, + "grad_norm": 2.4606130123138428, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.8709378242492676, + "num_tokens": 204405791.0, + "step": 5258 + }, + { + "epoch": 0.6689988551074927, + "grad_norm": 2.675065040588379, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8829390406608582, + "num_tokens": 204445480.0, + "step": 5259 + }, + { + "epoch": 0.6691260653860832, + "grad_norm": 2.8679661750793457, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8732776641845703, + "num_tokens": 204479375.0, + "step": 5260 + }, + { + "epoch": 0.6692532756646737, + "grad_norm": 2.6186277866363525, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8702036142349243, + "num_tokens": 204519388.0, + "step": 5261 + }, + { + "epoch": 0.6693804859432643, + "grad_norm": 2.663923978805542, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8752959966659546, + "num_tokens": 204554200.0, + "step": 5262 + }, + { + "epoch": 0.6695076962218547, + "grad_norm": 2.5396525859832764, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8725383281707764, + "num_tokens": 204595452.0, + "step": 5263 + }, + { + "epoch": 0.6696349065004452, + "grad_norm": 2.4851653575897217, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8860240578651428, + "num_tokens": 204634780.0, + "step": 5264 + }, + { + "epoch": 0.6697621167790357, + "grad_norm": 2.657318592071533, + "learning_rate": 1e-06, + "loss": 0.4179, + "mean_token_accuracy": 0.8542764782905579, + "num_tokens": 204679992.0, + "step": 5265 + }, + { + "epoch": 0.6698893270576263, + "grad_norm": 2.5734055042266846, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8745848536491394, + "num_tokens": 204722324.0, + "step": 5266 + }, + { + "epoch": 0.6700165373362168, + "grad_norm": 2.607081413269043, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.882933497428894, + "num_tokens": 204760495.0, + "step": 5267 + }, + { + "epoch": 0.6701437476148073, + "grad_norm": 2.6502013206481934, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8839249014854431, + "num_tokens": 204797917.0, + "step": 5268 + }, + { + "epoch": 0.6702709578933977, + "grad_norm": 2.875239372253418, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.863903820514679, + "num_tokens": 204833319.0, + "step": 5269 + }, + { + "epoch": 0.6703981681719883, + "grad_norm": 2.866276264190674, + "learning_rate": 1e-06, + "loss": 0.4128, + "mean_token_accuracy": 0.8560901880264282, + "num_tokens": 204869712.0, + "step": 5270 + }, + { + "epoch": 0.6705253784505788, + "grad_norm": 2.7243494987487793, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8798203468322754, + "num_tokens": 204904587.0, + "step": 5271 + }, + { + "epoch": 0.6706525887291693, + "grad_norm": 2.62471866607666, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8796815872192383, + "num_tokens": 204943306.0, + "step": 5272 + }, + { + "epoch": 0.6707797990077599, + "grad_norm": 2.529458522796631, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8811969757080078, + "num_tokens": 204982260.0, + "step": 5273 + }, + { + "epoch": 0.6709070092863504, + "grad_norm": 2.683522939682007, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8987893462181091, + "num_tokens": 205014189.0, + "step": 5274 + }, + { + "epoch": 0.6710342195649408, + "grad_norm": 2.8535826206207275, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8738127946853638, + "num_tokens": 205050025.0, + "step": 5275 + }, + { + "epoch": 0.6711614298435313, + "grad_norm": 2.464891195297241, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8886287212371826, + "num_tokens": 205089150.0, + "step": 5276 + }, + { + "epoch": 0.6712886401221219, + "grad_norm": 2.391761064529419, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8816738724708557, + "num_tokens": 205133664.0, + "step": 5277 + }, + { + "epoch": 0.6714158504007124, + "grad_norm": 2.512521266937256, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8946425914764404, + "num_tokens": 205170982.0, + "step": 5278 + }, + { + "epoch": 0.6715430606793029, + "grad_norm": 2.9025838375091553, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8700237274169922, + "num_tokens": 205203263.0, + "step": 5279 + }, + { + "epoch": 0.6716702709578934, + "grad_norm": 2.7253172397613525, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8706777095794678, + "num_tokens": 205242885.0, + "step": 5280 + }, + { + "epoch": 0.6717974812364839, + "grad_norm": 2.4339959621429443, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8883440494537354, + "num_tokens": 205282823.0, + "step": 5281 + }, + { + "epoch": 0.6719246915150744, + "grad_norm": 2.794090509414673, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8842101097106934, + "num_tokens": 205317739.0, + "step": 5282 + }, + { + "epoch": 0.6720519017936649, + "grad_norm": 2.788513422012329, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8756104707717896, + "num_tokens": 205359402.0, + "step": 5283 + }, + { + "epoch": 0.6721791120722554, + "grad_norm": 2.4562103748321533, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8796131610870361, + "num_tokens": 205401039.0, + "step": 5284 + }, + { + "epoch": 0.672306322350846, + "grad_norm": 2.3927667140960693, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8733103275299072, + "num_tokens": 205446185.0, + "step": 5285 + }, + { + "epoch": 0.6724335326294365, + "grad_norm": 2.588266611099243, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8729947805404663, + "num_tokens": 205487189.0, + "step": 5286 + }, + { + "epoch": 0.6725607429080269, + "grad_norm": 2.743515968322754, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8672718405723572, + "num_tokens": 205526862.0, + "step": 5287 + }, + { + "epoch": 0.6726879531866174, + "grad_norm": 2.8024795055389404, + "learning_rate": 1e-06, + "loss": 0.3752, + "mean_token_accuracy": 0.871055006980896, + "num_tokens": 205565154.0, + "step": 5288 + }, + { + "epoch": 0.672815163465208, + "grad_norm": 2.641482353210449, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8734016418457031, + "num_tokens": 205607589.0, + "step": 5289 + }, + { + "epoch": 0.6729423737437985, + "grad_norm": 2.950331687927246, + "learning_rate": 1e-06, + "loss": 0.4194, + "mean_token_accuracy": 0.8556573390960693, + "num_tokens": 205644769.0, + "step": 5290 + }, + { + "epoch": 0.673069584022389, + "grad_norm": 2.6619293689727783, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8681612610816956, + "num_tokens": 205684682.0, + "step": 5291 + }, + { + "epoch": 0.6731967943009796, + "grad_norm": 2.535954475402832, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8888631463050842, + "num_tokens": 205720387.0, + "step": 5292 + }, + { + "epoch": 0.67332400457957, + "grad_norm": 2.6762008666992188, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8786813020706177, + "num_tokens": 205756351.0, + "step": 5293 + }, + { + "epoch": 0.6734512148581605, + "grad_norm": 2.917856454849243, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8781609535217285, + "num_tokens": 205789617.0, + "step": 5294 + }, + { + "epoch": 0.673578425136751, + "grad_norm": 2.77288556098938, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.880352258682251, + "num_tokens": 205824058.0, + "step": 5295 + }, + { + "epoch": 0.6737056354153416, + "grad_norm": 2.562701940536499, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8759049773216248, + "num_tokens": 205865451.0, + "step": 5296 + }, + { + "epoch": 0.6738328456939321, + "grad_norm": 2.488203287124634, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8790514469146729, + "num_tokens": 205907977.0, + "step": 5297 + }, + { + "epoch": 0.6739600559725226, + "grad_norm": 2.4058403968811035, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8822470903396606, + "num_tokens": 205949711.0, + "step": 5298 + }, + { + "epoch": 0.674087266251113, + "grad_norm": 2.650373935699463, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8759234547615051, + "num_tokens": 205987693.0, + "step": 5299 + }, + { + "epoch": 0.6742144765297036, + "grad_norm": 2.7284326553344727, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8866153955459595, + "num_tokens": 206025277.0, + "step": 5300 + }, + { + "epoch": 0.6743416868082941, + "grad_norm": 2.7547998428344727, + "learning_rate": 1e-06, + "loss": 0.4151, + "mean_token_accuracy": 0.8580584526062012, + "num_tokens": 206065671.0, + "step": 5301 + }, + { + "epoch": 0.6744688970868846, + "grad_norm": 2.4301888942718506, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8855640888214111, + "num_tokens": 206105234.0, + "step": 5302 + }, + { + "epoch": 0.6745961073654752, + "grad_norm": 2.434020757675171, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8726028800010681, + "num_tokens": 206151480.0, + "step": 5303 + }, + { + "epoch": 0.6747233176440657, + "grad_norm": 2.354417562484741, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8834071159362793, + "num_tokens": 206196826.0, + "step": 5304 + }, + { + "epoch": 0.6748505279226561, + "grad_norm": 2.441004514694214, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8813516497612, + "num_tokens": 206239082.0, + "step": 5305 + }, + { + "epoch": 0.6749777382012466, + "grad_norm": 2.4059364795684814, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8845899105072021, + "num_tokens": 206278053.0, + "step": 5306 + }, + { + "epoch": 0.6751049484798372, + "grad_norm": 2.738429069519043, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8704837560653687, + "num_tokens": 206314474.0, + "step": 5307 + }, + { + "epoch": 0.6752321587584277, + "grad_norm": 2.480577230453491, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8822907209396362, + "num_tokens": 206358008.0, + "step": 5308 + }, + { + "epoch": 0.6753593690370182, + "grad_norm": 2.500833511352539, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8793374300003052, + "num_tokens": 206400745.0, + "step": 5309 + }, + { + "epoch": 0.6754865793156087, + "grad_norm": 2.488430976867676, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8813338279724121, + "num_tokens": 206436267.0, + "step": 5310 + }, + { + "epoch": 0.6756137895941993, + "grad_norm": 2.498122453689575, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8798499703407288, + "num_tokens": 206478285.0, + "step": 5311 + }, + { + "epoch": 0.6757409998727897, + "grad_norm": 2.5847856998443604, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8818342685699463, + "num_tokens": 206517931.0, + "step": 5312 + }, + { + "epoch": 0.6758682101513802, + "grad_norm": 2.502403497695923, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8954002857208252, + "num_tokens": 206556352.0, + "step": 5313 + }, + { + "epoch": 0.6759954204299707, + "grad_norm": 2.5762577056884766, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8908407688140869, + "num_tokens": 206590769.0, + "step": 5314 + }, + { + "epoch": 0.6761226307085613, + "grad_norm": 2.5664639472961426, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8790045380592346, + "num_tokens": 206628386.0, + "step": 5315 + }, + { + "epoch": 0.6762498409871518, + "grad_norm": 2.70269775390625, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8721152544021606, + "num_tokens": 206667007.0, + "step": 5316 + }, + { + "epoch": 0.6763770512657423, + "grad_norm": 2.502213478088379, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.882922351360321, + "num_tokens": 206706821.0, + "step": 5317 + }, + { + "epoch": 0.6765042615443327, + "grad_norm": 2.2303788661956787, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8929628133773804, + "num_tokens": 206752019.0, + "step": 5318 + }, + { + "epoch": 0.6766314718229233, + "grad_norm": 2.2735507488250732, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8877239227294922, + "num_tokens": 206795534.0, + "step": 5319 + }, + { + "epoch": 0.6767586821015138, + "grad_norm": 2.4348621368408203, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8886528611183167, + "num_tokens": 206834277.0, + "step": 5320 + }, + { + "epoch": 0.6768858923801043, + "grad_norm": 2.5868046283721924, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8711491823196411, + "num_tokens": 206873142.0, + "step": 5321 + }, + { + "epoch": 0.6770131026586949, + "grad_norm": 2.5956311225891113, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8800652623176575, + "num_tokens": 206911849.0, + "step": 5322 + }, + { + "epoch": 0.6771403129372854, + "grad_norm": 2.4824414253234863, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8832046985626221, + "num_tokens": 206951931.0, + "step": 5323 + }, + { + "epoch": 0.6772675232158758, + "grad_norm": 2.5248544216156006, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8872197866439819, + "num_tokens": 206988659.0, + "step": 5324 + }, + { + "epoch": 0.6773947334944663, + "grad_norm": 2.836137056350708, + "learning_rate": 1e-06, + "loss": 0.3781, + "mean_token_accuracy": 0.8692238330841064, + "num_tokens": 207028260.0, + "step": 5325 + }, + { + "epoch": 0.6775219437730569, + "grad_norm": 3.01857590675354, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8629248142242432, + "num_tokens": 207059082.0, + "step": 5326 + }, + { + "epoch": 0.6776491540516474, + "grad_norm": 2.8051815032958984, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8759132623672485, + "num_tokens": 207093232.0, + "step": 5327 + }, + { + "epoch": 0.6777763643302379, + "grad_norm": 2.3922817707061768, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8824739456176758, + "num_tokens": 207133832.0, + "step": 5328 + }, + { + "epoch": 0.6779035746088284, + "grad_norm": 2.6774187088012695, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8834081888198853, + "num_tokens": 207168025.0, + "step": 5329 + }, + { + "epoch": 0.6780307848874189, + "grad_norm": 2.5829646587371826, + "learning_rate": 1e-06, + "loss": 0.3866, + "mean_token_accuracy": 0.8649485111236572, + "num_tokens": 207209971.0, + "step": 5330 + }, + { + "epoch": 0.6781579951660094, + "grad_norm": 2.516062021255493, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8812448382377625, + "num_tokens": 207251833.0, + "step": 5331 + }, + { + "epoch": 0.6782852054445999, + "grad_norm": 2.7886998653411865, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8827924728393555, + "num_tokens": 207284254.0, + "step": 5332 + }, + { + "epoch": 0.6784124157231904, + "grad_norm": 2.570573568344116, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8837937116622925, + "num_tokens": 207321269.0, + "step": 5333 + }, + { + "epoch": 0.678539626001781, + "grad_norm": 2.6047966480255127, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8925807476043701, + "num_tokens": 207359311.0, + "step": 5334 + }, + { + "epoch": 0.6786668362803715, + "grad_norm": 2.356518030166626, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8783509731292725, + "num_tokens": 207406101.0, + "step": 5335 + }, + { + "epoch": 0.6787940465589619, + "grad_norm": 2.552442789077759, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8814549446105957, + "num_tokens": 207445638.0, + "step": 5336 + }, + { + "epoch": 0.6789212568375524, + "grad_norm": 2.721581220626831, + "learning_rate": 1e-06, + "loss": 0.3805, + "mean_token_accuracy": 0.8662925958633423, + "num_tokens": 207482300.0, + "step": 5337 + }, + { + "epoch": 0.679048467116143, + "grad_norm": 2.515000820159912, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8865950107574463, + "num_tokens": 207520707.0, + "step": 5338 + }, + { + "epoch": 0.6791756773947335, + "grad_norm": 2.5412325859069824, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8865498900413513, + "num_tokens": 207561949.0, + "step": 5339 + }, + { + "epoch": 0.679302887673324, + "grad_norm": 2.781550645828247, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8661231994628906, + "num_tokens": 207597062.0, + "step": 5340 + }, + { + "epoch": 0.6794300979519146, + "grad_norm": 2.7577033042907715, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8703826665878296, + "num_tokens": 207633839.0, + "step": 5341 + }, + { + "epoch": 0.679557308230505, + "grad_norm": 2.5246212482452393, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.8730107545852661, + "num_tokens": 207676530.0, + "step": 5342 + }, + { + "epoch": 0.6796845185090955, + "grad_norm": 2.785889148712158, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8760161995887756, + "num_tokens": 207714975.0, + "step": 5343 + }, + { + "epoch": 0.679811728787686, + "grad_norm": 2.6323373317718506, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8923119902610779, + "num_tokens": 207748975.0, + "step": 5344 + }, + { + "epoch": 0.6799389390662766, + "grad_norm": 2.6242244243621826, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8800187110900879, + "num_tokens": 207784778.0, + "step": 5345 + }, + { + "epoch": 0.6800661493448671, + "grad_norm": 2.490926742553711, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8735994100570679, + "num_tokens": 207825892.0, + "step": 5346 + }, + { + "epoch": 0.6801933596234576, + "grad_norm": 2.47344970703125, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8768524527549744, + "num_tokens": 207868308.0, + "step": 5347 + }, + { + "epoch": 0.680320569902048, + "grad_norm": 2.57702374458313, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.870294988155365, + "num_tokens": 207907584.0, + "step": 5348 + }, + { + "epoch": 0.6804477801806386, + "grad_norm": 2.528477668762207, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8825470805168152, + "num_tokens": 207949499.0, + "step": 5349 + }, + { + "epoch": 0.6805749904592291, + "grad_norm": 2.72347092628479, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8776906728744507, + "num_tokens": 207990654.0, + "step": 5350 + }, + { + "epoch": 0.6807022007378196, + "grad_norm": 2.844442129135132, + "learning_rate": 1e-06, + "loss": 0.3619, + "mean_token_accuracy": 0.8719086050987244, + "num_tokens": 208027624.0, + "step": 5351 + }, + { + "epoch": 0.6808294110164101, + "grad_norm": 2.6149919033050537, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8708003163337708, + "num_tokens": 208068488.0, + "step": 5352 + }, + { + "epoch": 0.6809566212950007, + "grad_norm": 2.9308974742889404, + "learning_rate": 1e-06, + "loss": 0.4068, + "mean_token_accuracy": 0.8572033643722534, + "num_tokens": 208101815.0, + "step": 5353 + }, + { + "epoch": 0.6810838315735911, + "grad_norm": 2.6716108322143555, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8775544166564941, + "num_tokens": 208141105.0, + "step": 5354 + }, + { + "epoch": 0.6812110418521816, + "grad_norm": 2.9886221885681152, + "learning_rate": 1e-06, + "loss": 0.3563, + "mean_token_accuracy": 0.873425304889679, + "num_tokens": 208178346.0, + "step": 5355 + }, + { + "epoch": 0.6813382521307721, + "grad_norm": 2.687135696411133, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.882500171661377, + "num_tokens": 208214110.0, + "step": 5356 + }, + { + "epoch": 0.6814654624093627, + "grad_norm": 2.288949489593506, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8791507482528687, + "num_tokens": 208261997.0, + "step": 5357 + }, + { + "epoch": 0.6815926726879532, + "grad_norm": 2.6349873542785645, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8808873891830444, + "num_tokens": 208295596.0, + "step": 5358 + }, + { + "epoch": 0.6817198829665437, + "grad_norm": 2.8058178424835205, + "learning_rate": 1e-06, + "loss": 0.4181, + "mean_token_accuracy": 0.8602930307388306, + "num_tokens": 208336616.0, + "step": 5359 + }, + { + "epoch": 0.6818470932451343, + "grad_norm": 2.6939659118652344, + "learning_rate": 1e-06, + "loss": 0.4075, + "mean_token_accuracy": 0.857924222946167, + "num_tokens": 208374147.0, + "step": 5360 + }, + { + "epoch": 0.6819743035237247, + "grad_norm": 2.4337596893310547, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8818225264549255, + "num_tokens": 208416105.0, + "step": 5361 + }, + { + "epoch": 0.6821015138023152, + "grad_norm": 2.4946184158325195, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8755192756652832, + "num_tokens": 208453922.0, + "step": 5362 + }, + { + "epoch": 0.6822287240809057, + "grad_norm": 2.6508121490478516, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8781641125679016, + "num_tokens": 208489242.0, + "step": 5363 + }, + { + "epoch": 0.6823559343594963, + "grad_norm": 2.5005533695220947, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8868670463562012, + "num_tokens": 208528660.0, + "step": 5364 + }, + { + "epoch": 0.6824831446380868, + "grad_norm": 2.600267171859741, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8833575248718262, + "num_tokens": 208568741.0, + "step": 5365 + }, + { + "epoch": 0.6826103549166773, + "grad_norm": 2.62288236618042, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8873060941696167, + "num_tokens": 208605962.0, + "step": 5366 + }, + { + "epoch": 0.6827375651952677, + "grad_norm": 2.4987807273864746, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8739729523658752, + "num_tokens": 208650544.0, + "step": 5367 + }, + { + "epoch": 0.6828647754738583, + "grad_norm": 2.8884875774383545, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8767940402030945, + "num_tokens": 208684056.0, + "step": 5368 + }, + { + "epoch": 0.6829919857524488, + "grad_norm": 2.6269609928131104, + "learning_rate": 1e-06, + "loss": 0.3934, + "mean_token_accuracy": 0.8644484281539917, + "num_tokens": 208725247.0, + "step": 5369 + }, + { + "epoch": 0.6831191960310393, + "grad_norm": 3.049882411956787, + "learning_rate": 1e-06, + "loss": 0.3662, + "mean_token_accuracy": 0.8726574778556824, + "num_tokens": 208755321.0, + "step": 5370 + }, + { + "epoch": 0.6832464063096299, + "grad_norm": 2.8354883193969727, + "learning_rate": 1e-06, + "loss": 0.3784, + "mean_token_accuracy": 0.8690787553787231, + "num_tokens": 208791759.0, + "step": 5371 + }, + { + "epoch": 0.6833736165882204, + "grad_norm": 2.7043418884277344, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8760703802108765, + "num_tokens": 208828669.0, + "step": 5372 + }, + { + "epoch": 0.6835008268668108, + "grad_norm": 2.9511911869049072, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.870204746723175, + "num_tokens": 208864991.0, + "step": 5373 + }, + { + "epoch": 0.6836280371454013, + "grad_norm": 2.522735834121704, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8947781324386597, + "num_tokens": 208900908.0, + "step": 5374 + }, + { + "epoch": 0.6837552474239919, + "grad_norm": 2.649653434753418, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8821734189987183, + "num_tokens": 208939299.0, + "step": 5375 + }, + { + "epoch": 0.6838824577025824, + "grad_norm": 3.0908007621765137, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8657856583595276, + "num_tokens": 208970721.0, + "step": 5376 + }, + { + "epoch": 0.6840096679811729, + "grad_norm": 2.5276362895965576, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8735280632972717, + "num_tokens": 209010760.0, + "step": 5377 + }, + { + "epoch": 0.6841368782597634, + "grad_norm": 2.8973419666290283, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8810532093048096, + "num_tokens": 209043767.0, + "step": 5378 + }, + { + "epoch": 0.6842640885383539, + "grad_norm": 2.6178975105285645, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.870862603187561, + "num_tokens": 209083528.0, + "step": 5379 + }, + { + "epoch": 0.6843912988169444, + "grad_norm": 2.4503397941589355, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8774902820587158, + "num_tokens": 209123849.0, + "step": 5380 + }, + { + "epoch": 0.6845185090955349, + "grad_norm": 2.584500312805176, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8719521760940552, + "num_tokens": 209164164.0, + "step": 5381 + }, + { + "epoch": 0.6846457193741254, + "grad_norm": 2.6356570720672607, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8740696907043457, + "num_tokens": 209200832.0, + "step": 5382 + }, + { + "epoch": 0.684772929652716, + "grad_norm": 2.4153687953948975, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8884010314941406, + "num_tokens": 209239867.0, + "step": 5383 + }, + { + "epoch": 0.6849001399313065, + "grad_norm": 2.602551221847534, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8821281790733337, + "num_tokens": 209276102.0, + "step": 5384 + }, + { + "epoch": 0.6850273502098969, + "grad_norm": 2.4403297901153564, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8792246580123901, + "num_tokens": 209319187.0, + "step": 5385 + }, + { + "epoch": 0.6851545604884874, + "grad_norm": 2.7614803314208984, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8834381103515625, + "num_tokens": 209353238.0, + "step": 5386 + }, + { + "epoch": 0.685281770767078, + "grad_norm": 2.786405324935913, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8664355278015137, + "num_tokens": 209389892.0, + "step": 5387 + }, + { + "epoch": 0.6854089810456685, + "grad_norm": 2.872269630432129, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.873593270778656, + "num_tokens": 209423525.0, + "step": 5388 + }, + { + "epoch": 0.685536191324259, + "grad_norm": 2.4819416999816895, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8768172264099121, + "num_tokens": 209466029.0, + "step": 5389 + }, + { + "epoch": 0.6856634016028496, + "grad_norm": 2.7297112941741943, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.8688892722129822, + "num_tokens": 209503326.0, + "step": 5390 + }, + { + "epoch": 0.68579061188144, + "grad_norm": 2.6691579818725586, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8796882629394531, + "num_tokens": 209539019.0, + "step": 5391 + }, + { + "epoch": 0.6859178221600305, + "grad_norm": 2.6101324558258057, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.889901876449585, + "num_tokens": 209575332.0, + "step": 5392 + }, + { + "epoch": 0.686045032438621, + "grad_norm": 2.593109369277954, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8727931976318359, + "num_tokens": 209614415.0, + "step": 5393 + }, + { + "epoch": 0.6861722427172116, + "grad_norm": 2.5529651641845703, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8860983848571777, + "num_tokens": 209652820.0, + "step": 5394 + }, + { + "epoch": 0.6862994529958021, + "grad_norm": 2.6022331714630127, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8827482461929321, + "num_tokens": 209689264.0, + "step": 5395 + }, + { + "epoch": 0.6864266632743926, + "grad_norm": 2.510895252227783, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8770706653594971, + "num_tokens": 209727070.0, + "step": 5396 + }, + { + "epoch": 0.686553873552983, + "grad_norm": 2.562404155731201, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8762229084968567, + "num_tokens": 209768178.0, + "step": 5397 + }, + { + "epoch": 0.6866810838315736, + "grad_norm": 2.9331157207489014, + "learning_rate": 1e-06, + "loss": 0.447, + "mean_token_accuracy": 0.8443901538848877, + "num_tokens": 209809271.0, + "step": 5398 + }, + { + "epoch": 0.6868082941101641, + "grad_norm": 2.7391304969787598, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8845363259315491, + "num_tokens": 209842631.0, + "step": 5399 + }, + { + "epoch": 0.6869355043887546, + "grad_norm": 2.5523014068603516, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.875556230545044, + "num_tokens": 209884364.0, + "step": 5400 + }, + { + "epoch": 0.6870627146673451, + "grad_norm": 2.5945417881011963, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8895279169082642, + "num_tokens": 209918736.0, + "step": 5401 + }, + { + "epoch": 0.6871899249459357, + "grad_norm": 2.514551877975464, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8727761507034302, + "num_tokens": 209959976.0, + "step": 5402 + }, + { + "epoch": 0.6873171352245261, + "grad_norm": 2.509464740753174, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8700396418571472, + "num_tokens": 210000775.0, + "step": 5403 + }, + { + "epoch": 0.6874443455031166, + "grad_norm": 2.545198678970337, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8767763376235962, + "num_tokens": 210042292.0, + "step": 5404 + }, + { + "epoch": 0.6875715557817071, + "grad_norm": 2.7715375423431396, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.875947117805481, + "num_tokens": 210075484.0, + "step": 5405 + }, + { + "epoch": 0.6876987660602977, + "grad_norm": 2.5768771171569824, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8746808767318726, + "num_tokens": 210115749.0, + "step": 5406 + }, + { + "epoch": 0.6878259763388882, + "grad_norm": 2.724167585372925, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8782517910003662, + "num_tokens": 210152028.0, + "step": 5407 + }, + { + "epoch": 0.6879531866174787, + "grad_norm": 2.7702016830444336, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.860095202922821, + "num_tokens": 210187193.0, + "step": 5408 + }, + { + "epoch": 0.6880803968960693, + "grad_norm": 2.5083119869232178, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8785240650177002, + "num_tokens": 210226991.0, + "step": 5409 + }, + { + "epoch": 0.6882076071746597, + "grad_norm": 3.014503002166748, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8761763572692871, + "num_tokens": 210258955.0, + "step": 5410 + }, + { + "epoch": 0.6883348174532502, + "grad_norm": 2.7479803562164307, + "learning_rate": 1e-06, + "loss": 0.3823, + "mean_token_accuracy": 0.87191241979599, + "num_tokens": 210297449.0, + "step": 5411 + }, + { + "epoch": 0.6884620277318407, + "grad_norm": 2.5372419357299805, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8735538125038147, + "num_tokens": 210338700.0, + "step": 5412 + }, + { + "epoch": 0.6885892380104313, + "grad_norm": 2.637690782546997, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8698370456695557, + "num_tokens": 210377391.0, + "step": 5413 + }, + { + "epoch": 0.6887164482890218, + "grad_norm": 2.5810022354125977, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8776243925094604, + "num_tokens": 210415625.0, + "step": 5414 + }, + { + "epoch": 0.6888436585676123, + "grad_norm": 2.458205223083496, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8943603038787842, + "num_tokens": 210449813.0, + "step": 5415 + }, + { + "epoch": 0.6889708688462027, + "grad_norm": 2.561681032180786, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8857847452163696, + "num_tokens": 210486577.0, + "step": 5416 + }, + { + "epoch": 0.6890980791247933, + "grad_norm": 2.4681317806243896, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.885765552520752, + "num_tokens": 210525398.0, + "step": 5417 + }, + { + "epoch": 0.6892252894033838, + "grad_norm": 2.559962034225464, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8800468444824219, + "num_tokens": 210562971.0, + "step": 5418 + }, + { + "epoch": 0.6893524996819743, + "grad_norm": 2.450671911239624, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8820741176605225, + "num_tokens": 210604530.0, + "step": 5419 + }, + { + "epoch": 0.6894797099605648, + "grad_norm": 2.723682165145874, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8759913444519043, + "num_tokens": 210639363.0, + "step": 5420 + }, + { + "epoch": 0.6896069202391554, + "grad_norm": 2.565544605255127, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8788708448410034, + "num_tokens": 210678821.0, + "step": 5421 + }, + { + "epoch": 0.6897341305177458, + "grad_norm": 2.552626371383667, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8730088472366333, + "num_tokens": 210722840.0, + "step": 5422 + }, + { + "epoch": 0.6898613407963363, + "grad_norm": 2.5623693466186523, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8859224915504456, + "num_tokens": 210755639.0, + "step": 5423 + }, + { + "epoch": 0.6899885510749268, + "grad_norm": 2.530815839767456, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8838305473327637, + "num_tokens": 210795877.0, + "step": 5424 + }, + { + "epoch": 0.6901157613535174, + "grad_norm": 2.521592855453491, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.88133305311203, + "num_tokens": 210835730.0, + "step": 5425 + }, + { + "epoch": 0.6902429716321079, + "grad_norm": 2.6691598892211914, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8738951086997986, + "num_tokens": 210871259.0, + "step": 5426 + }, + { + "epoch": 0.6903701819106984, + "grad_norm": 2.4676613807678223, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8797877430915833, + "num_tokens": 210912900.0, + "step": 5427 + }, + { + "epoch": 0.6904973921892888, + "grad_norm": 2.901115894317627, + "learning_rate": 1e-06, + "loss": 0.3983, + "mean_token_accuracy": 0.8612310886383057, + "num_tokens": 210947499.0, + "step": 5428 + }, + { + "epoch": 0.6906246024678794, + "grad_norm": 2.4658279418945312, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8742673397064209, + "num_tokens": 210988766.0, + "step": 5429 + }, + { + "epoch": 0.6907518127464699, + "grad_norm": 2.8787753582000732, + "learning_rate": 1e-06, + "loss": 0.42, + "mean_token_accuracy": 0.8579698801040649, + "num_tokens": 211027145.0, + "step": 5430 + }, + { + "epoch": 0.6908790230250604, + "grad_norm": 2.991002082824707, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8711633682250977, + "num_tokens": 211056910.0, + "step": 5431 + }, + { + "epoch": 0.691006233303651, + "grad_norm": 2.7363364696502686, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.87930828332901, + "num_tokens": 211091404.0, + "step": 5432 + }, + { + "epoch": 0.6911334435822415, + "grad_norm": 2.963602066040039, + "learning_rate": 1e-06, + "loss": 0.3964, + "mean_token_accuracy": 0.8598699569702148, + "num_tokens": 211125848.0, + "step": 5433 + }, + { + "epoch": 0.6912606538608319, + "grad_norm": 2.5809900760650635, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8886573910713196, + "num_tokens": 211161600.0, + "step": 5434 + }, + { + "epoch": 0.6913878641394224, + "grad_norm": 2.3841493129730225, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8867115378379822, + "num_tokens": 211202035.0, + "step": 5435 + }, + { + "epoch": 0.691515074418013, + "grad_norm": 2.5585873126983643, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8682332634925842, + "num_tokens": 211245551.0, + "step": 5436 + }, + { + "epoch": 0.6916422846966035, + "grad_norm": 2.4200687408447266, + "learning_rate": 1e-06, + "loss": 0.3599, + "mean_token_accuracy": 0.8737274408340454, + "num_tokens": 211290430.0, + "step": 5437 + }, + { + "epoch": 0.691769494975194, + "grad_norm": 2.6120636463165283, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8780863285064697, + "num_tokens": 211330527.0, + "step": 5438 + }, + { + "epoch": 0.6918967052537845, + "grad_norm": 2.5635592937469482, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8840291500091553, + "num_tokens": 211367204.0, + "step": 5439 + }, + { + "epoch": 0.692023915532375, + "grad_norm": 2.217190742492676, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8867683410644531, + "num_tokens": 211413100.0, + "step": 5440 + }, + { + "epoch": 0.6921511258109655, + "grad_norm": 2.4387316703796387, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8860780000686646, + "num_tokens": 211453782.0, + "step": 5441 + }, + { + "epoch": 0.692278336089556, + "grad_norm": 2.4882214069366455, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8773577809333801, + "num_tokens": 211497771.0, + "step": 5442 + }, + { + "epoch": 0.6924055463681466, + "grad_norm": 2.763117551803589, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.881186842918396, + "num_tokens": 211530816.0, + "step": 5443 + }, + { + "epoch": 0.6925327566467371, + "grad_norm": 2.621422290802002, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8683842420578003, + "num_tokens": 211571332.0, + "step": 5444 + }, + { + "epoch": 0.6926599669253276, + "grad_norm": 2.8088033199310303, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8825966715812683, + "num_tokens": 211602498.0, + "step": 5445 + }, + { + "epoch": 0.692787177203918, + "grad_norm": 2.9392893314361572, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8693589568138123, + "num_tokens": 211633814.0, + "step": 5446 + }, + { + "epoch": 0.6929143874825086, + "grad_norm": 2.5776026248931885, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8684484362602234, + "num_tokens": 211670747.0, + "step": 5447 + }, + { + "epoch": 0.6930415977610991, + "grad_norm": 2.418476104736328, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8832113742828369, + "num_tokens": 211710849.0, + "step": 5448 + }, + { + "epoch": 0.6931688080396896, + "grad_norm": 2.832875967025757, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8646257519721985, + "num_tokens": 211750692.0, + "step": 5449 + }, + { + "epoch": 0.6932960183182801, + "grad_norm": 2.539900302886963, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8786633610725403, + "num_tokens": 211794329.0, + "step": 5450 + }, + { + "epoch": 0.6934232285968707, + "grad_norm": 2.472160577774048, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8773300647735596, + "num_tokens": 211838674.0, + "step": 5451 + }, + { + "epoch": 0.6935504388754611, + "grad_norm": 2.4573593139648438, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8965247869491577, + "num_tokens": 211874760.0, + "step": 5452 + }, + { + "epoch": 0.6936776491540516, + "grad_norm": 2.4226746559143066, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8883774280548096, + "num_tokens": 211916584.0, + "step": 5453 + }, + { + "epoch": 0.6938048594326421, + "grad_norm": 2.5273690223693848, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.887294590473175, + "num_tokens": 211951671.0, + "step": 5454 + }, + { + "epoch": 0.6939320697112327, + "grad_norm": 2.40494966506958, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.871410608291626, + "num_tokens": 211995053.0, + "step": 5455 + }, + { + "epoch": 0.6940592799898232, + "grad_norm": 2.5932629108428955, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8757338523864746, + "num_tokens": 212032590.0, + "step": 5456 + }, + { + "epoch": 0.6941864902684137, + "grad_norm": 2.4191806316375732, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.881911039352417, + "num_tokens": 212071085.0, + "step": 5457 + }, + { + "epoch": 0.6943137005470043, + "grad_norm": 2.8717923164367676, + "learning_rate": 1e-06, + "loss": 0.387, + "mean_token_accuracy": 0.8663067817687988, + "num_tokens": 212104654.0, + "step": 5458 + }, + { + "epoch": 0.6944409108255947, + "grad_norm": 2.638747453689575, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8759740591049194, + "num_tokens": 212144597.0, + "step": 5459 + }, + { + "epoch": 0.6945681211041852, + "grad_norm": 2.5269718170166016, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8725252747535706, + "num_tokens": 212186365.0, + "step": 5460 + }, + { + "epoch": 0.6946953313827757, + "grad_norm": 2.6285789012908936, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8846070766448975, + "num_tokens": 212222364.0, + "step": 5461 + }, + { + "epoch": 0.6948225416613663, + "grad_norm": 2.532931089401245, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8717854619026184, + "num_tokens": 212265239.0, + "step": 5462 + }, + { + "epoch": 0.6949497519399568, + "grad_norm": 2.9943954944610596, + "learning_rate": 1e-06, + "loss": 0.4612, + "mean_token_accuracy": 0.8451386094093323, + "num_tokens": 212303543.0, + "step": 5463 + }, + { + "epoch": 0.6950769622185473, + "grad_norm": 2.66200852394104, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8860063552856445, + "num_tokens": 212341499.0, + "step": 5464 + }, + { + "epoch": 0.6952041724971377, + "grad_norm": 2.6234381198883057, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.885672926902771, + "num_tokens": 212377595.0, + "step": 5465 + }, + { + "epoch": 0.6953313827757283, + "grad_norm": 2.6799545288085938, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8920875787734985, + "num_tokens": 212411618.0, + "step": 5466 + }, + { + "epoch": 0.6954585930543188, + "grad_norm": 2.8288707733154297, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8857908248901367, + "num_tokens": 212444196.0, + "step": 5467 + }, + { + "epoch": 0.6955858033329093, + "grad_norm": 2.5699288845062256, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8914744853973389, + "num_tokens": 212478967.0, + "step": 5468 + }, + { + "epoch": 0.6957130136114998, + "grad_norm": 2.7764859199523926, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8859577775001526, + "num_tokens": 212511867.0, + "step": 5469 + }, + { + "epoch": 0.6958402238900904, + "grad_norm": 2.8609840869903564, + "learning_rate": 1e-06, + "loss": 0.4166, + "mean_token_accuracy": 0.8541275858879089, + "num_tokens": 212550232.0, + "step": 5470 + }, + { + "epoch": 0.6959674341686808, + "grad_norm": 2.6884944438934326, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8704425096511841, + "num_tokens": 212591163.0, + "step": 5471 + }, + { + "epoch": 0.6960946444472713, + "grad_norm": 2.5686540603637695, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8847400546073914, + "num_tokens": 212631839.0, + "step": 5472 + }, + { + "epoch": 0.6962218547258618, + "grad_norm": 2.901834011077881, + "learning_rate": 1e-06, + "loss": 0.3929, + "mean_token_accuracy": 0.8655562400817871, + "num_tokens": 212668608.0, + "step": 5473 + }, + { + "epoch": 0.6963490650044524, + "grad_norm": 2.6591200828552246, + "learning_rate": 1e-06, + "loss": 0.3995, + "mean_token_accuracy": 0.8634642958641052, + "num_tokens": 212711924.0, + "step": 5474 + }, + { + "epoch": 0.6964762752830429, + "grad_norm": 2.846543550491333, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.867255449295044, + "num_tokens": 212747396.0, + "step": 5475 + }, + { + "epoch": 0.6966034855616334, + "grad_norm": 2.596229314804077, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8786544799804688, + "num_tokens": 212786505.0, + "step": 5476 + }, + { + "epoch": 0.6967306958402238, + "grad_norm": 2.4846551418304443, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8838707208633423, + "num_tokens": 212825093.0, + "step": 5477 + }, + { + "epoch": 0.6968579061188144, + "grad_norm": 2.437714099884033, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8838785290718079, + "num_tokens": 212864871.0, + "step": 5478 + }, + { + "epoch": 0.6969851163974049, + "grad_norm": 2.2780680656433105, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8840229511260986, + "num_tokens": 212909161.0, + "step": 5479 + }, + { + "epoch": 0.6971123266759954, + "grad_norm": 2.5013015270233154, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8767566680908203, + "num_tokens": 212951057.0, + "step": 5480 + }, + { + "epoch": 0.697239536954586, + "grad_norm": 2.629215955734253, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8823049664497375, + "num_tokens": 212988505.0, + "step": 5481 + }, + { + "epoch": 0.6973667472331765, + "grad_norm": 2.309087038040161, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8789989352226257, + "num_tokens": 213033828.0, + "step": 5482 + }, + { + "epoch": 0.6974939575117669, + "grad_norm": 2.6454968452453613, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8878879547119141, + "num_tokens": 213067654.0, + "step": 5483 + }, + { + "epoch": 0.6976211677903574, + "grad_norm": 2.5965330600738525, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8759055137634277, + "num_tokens": 213106988.0, + "step": 5484 + }, + { + "epoch": 0.697748378068948, + "grad_norm": 2.601975440979004, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8806143999099731, + "num_tokens": 213145620.0, + "step": 5485 + }, + { + "epoch": 0.6978755883475385, + "grad_norm": 2.6643636226654053, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8771276473999023, + "num_tokens": 213183112.0, + "step": 5486 + }, + { + "epoch": 0.698002798626129, + "grad_norm": 2.7949209213256836, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8790687322616577, + "num_tokens": 213216208.0, + "step": 5487 + }, + { + "epoch": 0.6981300089047195, + "grad_norm": 2.580564022064209, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8863461017608643, + "num_tokens": 213256627.0, + "step": 5488 + }, + { + "epoch": 0.69825721918331, + "grad_norm": 2.6436879634857178, + "learning_rate": 1e-06, + "loss": 0.3941, + "mean_token_accuracy": 0.8620227575302124, + "num_tokens": 213297246.0, + "step": 5489 + }, + { + "epoch": 0.6983844294619005, + "grad_norm": 2.280543804168701, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9007211327552795, + "num_tokens": 213337178.0, + "step": 5490 + }, + { + "epoch": 0.698511639740491, + "grad_norm": 2.539616584777832, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8913910388946533, + "num_tokens": 213371703.0, + "step": 5491 + }, + { + "epoch": 0.6986388500190815, + "grad_norm": 2.591409206390381, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.88057541847229, + "num_tokens": 213408133.0, + "step": 5492 + }, + { + "epoch": 0.6987660602976721, + "grad_norm": 2.576383113861084, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.878549337387085, + "num_tokens": 213444824.0, + "step": 5493 + }, + { + "epoch": 0.6988932705762626, + "grad_norm": 2.530008554458618, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8753660917282104, + "num_tokens": 213486434.0, + "step": 5494 + }, + { + "epoch": 0.699020480854853, + "grad_norm": 2.8170676231384277, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8685781955718994, + "num_tokens": 213521539.0, + "step": 5495 + }, + { + "epoch": 0.6991476911334435, + "grad_norm": 2.5146234035491943, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8794702887535095, + "num_tokens": 213562325.0, + "step": 5496 + }, + { + "epoch": 0.6992749014120341, + "grad_norm": 2.7257142066955566, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.867169976234436, + "num_tokens": 213600778.0, + "step": 5497 + }, + { + "epoch": 0.6994021116906246, + "grad_norm": 2.313138723373413, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8938207626342773, + "num_tokens": 213643996.0, + "step": 5498 + }, + { + "epoch": 0.6995293219692151, + "grad_norm": 2.5030369758605957, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8873127698898315, + "num_tokens": 213684499.0, + "step": 5499 + }, + { + "epoch": 0.6996565322478057, + "grad_norm": 2.5734360218048096, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8773307204246521, + "num_tokens": 213723288.0, + "step": 5500 + }, + { + "epoch": 0.6997837425263961, + "grad_norm": 2.4769651889801025, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8856663703918457, + "num_tokens": 213766004.0, + "step": 5501 + }, + { + "epoch": 0.6999109528049866, + "grad_norm": 2.6873390674591064, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8749305605888367, + "num_tokens": 213804009.0, + "step": 5502 + }, + { + "epoch": 0.7000381630835771, + "grad_norm": 2.559535503387451, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8781211376190186, + "num_tokens": 213843968.0, + "step": 5503 + }, + { + "epoch": 0.7001653733621677, + "grad_norm": 2.782968521118164, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8816792964935303, + "num_tokens": 213879636.0, + "step": 5504 + }, + { + "epoch": 0.7002925836407582, + "grad_norm": 2.6711039543151855, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8704327344894409, + "num_tokens": 213920190.0, + "step": 5505 + }, + { + "epoch": 0.7004197939193487, + "grad_norm": 2.704782485961914, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8732904195785522, + "num_tokens": 213960454.0, + "step": 5506 + }, + { + "epoch": 0.7005470041979391, + "grad_norm": 2.651357650756836, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.878424882888794, + "num_tokens": 213997244.0, + "step": 5507 + }, + { + "epoch": 0.7006742144765297, + "grad_norm": 2.5430984497070312, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8866612911224365, + "num_tokens": 214031761.0, + "step": 5508 + }, + { + "epoch": 0.7008014247551202, + "grad_norm": 2.6301040649414062, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8684656023979187, + "num_tokens": 214070630.0, + "step": 5509 + }, + { + "epoch": 0.7009286350337107, + "grad_norm": 2.5965945720672607, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.879772424697876, + "num_tokens": 214105810.0, + "step": 5510 + }, + { + "epoch": 0.7010558453123013, + "grad_norm": 2.843897581100464, + "learning_rate": 1e-06, + "loss": 0.3917, + "mean_token_accuracy": 0.8662515878677368, + "num_tokens": 214141869.0, + "step": 5511 + }, + { + "epoch": 0.7011830555908918, + "grad_norm": 2.563307285308838, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.873926043510437, + "num_tokens": 214180004.0, + "step": 5512 + }, + { + "epoch": 0.7013102658694823, + "grad_norm": 2.4666554927825928, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.868800163269043, + "num_tokens": 214224777.0, + "step": 5513 + }, + { + "epoch": 0.7014374761480727, + "grad_norm": 2.3872146606445312, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8809161186218262, + "num_tokens": 214270533.0, + "step": 5514 + }, + { + "epoch": 0.7015646864266633, + "grad_norm": 2.4762356281280518, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8883487582206726, + "num_tokens": 214309175.0, + "step": 5515 + }, + { + "epoch": 0.7016918967052538, + "grad_norm": 2.6421992778778076, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8710833787918091, + "num_tokens": 214350631.0, + "step": 5516 + }, + { + "epoch": 0.7018191069838443, + "grad_norm": 2.635990858078003, + "learning_rate": 1e-06, + "loss": 0.3609, + "mean_token_accuracy": 0.8714762330055237, + "num_tokens": 214391137.0, + "step": 5517 + }, + { + "epoch": 0.7019463172624348, + "grad_norm": 2.5477638244628906, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8705098032951355, + "num_tokens": 214435783.0, + "step": 5518 + }, + { + "epoch": 0.7020735275410254, + "grad_norm": 2.6409265995025635, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8865728378295898, + "num_tokens": 214467891.0, + "step": 5519 + }, + { + "epoch": 0.7022007378196158, + "grad_norm": 2.6196792125701904, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8844047784805298, + "num_tokens": 214503826.0, + "step": 5520 + }, + { + "epoch": 0.7023279480982063, + "grad_norm": 2.563978672027588, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8841965794563293, + "num_tokens": 214541622.0, + "step": 5521 + }, + { + "epoch": 0.7024551583767968, + "grad_norm": 2.356149911880493, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8964682221412659, + "num_tokens": 214583072.0, + "step": 5522 + }, + { + "epoch": 0.7025823686553874, + "grad_norm": 2.4268171787261963, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8763910531997681, + "num_tokens": 214625440.0, + "step": 5523 + }, + { + "epoch": 0.7027095789339779, + "grad_norm": 2.6886484622955322, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.88466477394104, + "num_tokens": 214662000.0, + "step": 5524 + }, + { + "epoch": 0.7028367892125684, + "grad_norm": 2.854538679122925, + "learning_rate": 1e-06, + "loss": 0.3906, + "mean_token_accuracy": 0.8663492202758789, + "num_tokens": 214696021.0, + "step": 5525 + }, + { + "epoch": 0.7029639994911588, + "grad_norm": 2.61311674118042, + "learning_rate": 1e-06, + "loss": 0.3866, + "mean_token_accuracy": 0.869331955909729, + "num_tokens": 214736291.0, + "step": 5526 + }, + { + "epoch": 0.7030912097697494, + "grad_norm": 2.61482572555542, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8800097107887268, + "num_tokens": 214776334.0, + "step": 5527 + }, + { + "epoch": 0.7032184200483399, + "grad_norm": 2.5454823970794678, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.893841028213501, + "num_tokens": 214814716.0, + "step": 5528 + }, + { + "epoch": 0.7033456303269304, + "grad_norm": 2.70548939704895, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8710348606109619, + "num_tokens": 214852043.0, + "step": 5529 + }, + { + "epoch": 0.703472840605521, + "grad_norm": 2.881415367126465, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8791289925575256, + "num_tokens": 214889552.0, + "step": 5530 + }, + { + "epoch": 0.7036000508841115, + "grad_norm": 2.4647951126098633, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8764572143554688, + "num_tokens": 214932097.0, + "step": 5531 + }, + { + "epoch": 0.7037272611627019, + "grad_norm": 2.478776693344116, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8807863593101501, + "num_tokens": 214971277.0, + "step": 5532 + }, + { + "epoch": 0.7038544714412924, + "grad_norm": 2.6224074363708496, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8831937313079834, + "num_tokens": 215008459.0, + "step": 5533 + }, + { + "epoch": 0.703981681719883, + "grad_norm": 2.913008451461792, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8679949045181274, + "num_tokens": 215044238.0, + "step": 5534 + }, + { + "epoch": 0.7041088919984735, + "grad_norm": 2.419438123703003, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8906053900718689, + "num_tokens": 215088550.0, + "step": 5535 + }, + { + "epoch": 0.704236102277064, + "grad_norm": 2.770279884338379, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8762686252593994, + "num_tokens": 215125452.0, + "step": 5536 + }, + { + "epoch": 0.7043633125556545, + "grad_norm": 2.6673619747161865, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8839271068572998, + "num_tokens": 215160465.0, + "step": 5537 + }, + { + "epoch": 0.704490522834245, + "grad_norm": 2.5237114429473877, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8846451640129089, + "num_tokens": 215202531.0, + "step": 5538 + }, + { + "epoch": 0.7046177331128355, + "grad_norm": 2.5406270027160645, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8854956030845642, + "num_tokens": 215241244.0, + "step": 5539 + }, + { + "epoch": 0.704744943391426, + "grad_norm": 2.885392904281616, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8687943816184998, + "num_tokens": 215276006.0, + "step": 5540 + }, + { + "epoch": 0.7048721536700165, + "grad_norm": 2.573805570602417, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8905097246170044, + "num_tokens": 215311094.0, + "step": 5541 + }, + { + "epoch": 0.7049993639486071, + "grad_norm": 2.7861428260803223, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8691322803497314, + "num_tokens": 215345066.0, + "step": 5542 + }, + { + "epoch": 0.7051265742271976, + "grad_norm": 2.5494744777679443, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8859580755233765, + "num_tokens": 215383742.0, + "step": 5543 + }, + { + "epoch": 0.705253784505788, + "grad_norm": 2.6876800060272217, + "learning_rate": 1e-06, + "loss": 0.3901, + "mean_token_accuracy": 0.8649688363075256, + "num_tokens": 215424618.0, + "step": 5544 + }, + { + "epoch": 0.7053809947843785, + "grad_norm": 2.4342000484466553, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.890132486820221, + "num_tokens": 215464615.0, + "step": 5545 + }, + { + "epoch": 0.7055082050629691, + "grad_norm": 2.47739315032959, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8718032836914062, + "num_tokens": 215507335.0, + "step": 5546 + }, + { + "epoch": 0.7056354153415596, + "grad_norm": 2.490914821624756, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8685831427574158, + "num_tokens": 215549662.0, + "step": 5547 + }, + { + "epoch": 0.7057626256201501, + "grad_norm": 3.0351271629333496, + "learning_rate": 1e-06, + "loss": 0.4238, + "mean_token_accuracy": 0.8523520231246948, + "num_tokens": 215582285.0, + "step": 5548 + }, + { + "epoch": 0.7058898358987407, + "grad_norm": 2.6392524242401123, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8801078796386719, + "num_tokens": 215618192.0, + "step": 5549 + }, + { + "epoch": 0.7060170461773311, + "grad_norm": 2.746134042739868, + "learning_rate": 1e-06, + "loss": 0.3729, + "mean_token_accuracy": 0.8680570721626282, + "num_tokens": 215654311.0, + "step": 5550 + }, + { + "epoch": 0.7061442564559216, + "grad_norm": 2.703714370727539, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8798150420188904, + "num_tokens": 215689934.0, + "step": 5551 + }, + { + "epoch": 0.7062714667345121, + "grad_norm": 2.5849668979644775, + "learning_rate": 1e-06, + "loss": 0.3826, + "mean_token_accuracy": 0.8682944774627686, + "num_tokens": 215729967.0, + "step": 5552 + }, + { + "epoch": 0.7063986770131027, + "grad_norm": 2.6544580459594727, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8880157470703125, + "num_tokens": 215766193.0, + "step": 5553 + }, + { + "epoch": 0.7065258872916932, + "grad_norm": 2.7103700637817383, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8835611343383789, + "num_tokens": 215800286.0, + "step": 5554 + }, + { + "epoch": 0.7066530975702837, + "grad_norm": 2.699608087539673, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8956049680709839, + "num_tokens": 215831643.0, + "step": 5555 + }, + { + "epoch": 0.7067803078488741, + "grad_norm": 2.763993978500366, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8754031658172607, + "num_tokens": 215867882.0, + "step": 5556 + }, + { + "epoch": 0.7069075181274647, + "grad_norm": 2.806988477706909, + "learning_rate": 1e-06, + "loss": 0.3988, + "mean_token_accuracy": 0.8646465539932251, + "num_tokens": 215905131.0, + "step": 5557 + }, + { + "epoch": 0.7070347284060552, + "grad_norm": 2.861269235610962, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8733212351799011, + "num_tokens": 215935176.0, + "step": 5558 + }, + { + "epoch": 0.7071619386846457, + "grad_norm": 2.579608201980591, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8714706897735596, + "num_tokens": 215975454.0, + "step": 5559 + }, + { + "epoch": 0.7072891489632362, + "grad_norm": 2.4916951656341553, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8791909217834473, + "num_tokens": 216016406.0, + "step": 5560 + }, + { + "epoch": 0.7074163592418268, + "grad_norm": 2.6954290866851807, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8766049146652222, + "num_tokens": 216052654.0, + "step": 5561 + }, + { + "epoch": 0.7075435695204173, + "grad_norm": 2.64312744140625, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8795824646949768, + "num_tokens": 216088549.0, + "step": 5562 + }, + { + "epoch": 0.7076707797990077, + "grad_norm": 2.7290871143341064, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8758406639099121, + "num_tokens": 216124848.0, + "step": 5563 + }, + { + "epoch": 0.7077979900775982, + "grad_norm": 2.5162253379821777, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8777585029602051, + "num_tokens": 216163522.0, + "step": 5564 + }, + { + "epoch": 0.7079252003561888, + "grad_norm": 2.6695635318756104, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8843350410461426, + "num_tokens": 216199004.0, + "step": 5565 + }, + { + "epoch": 0.7080524106347793, + "grad_norm": 2.537946939468384, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8753498792648315, + "num_tokens": 216240540.0, + "step": 5566 + }, + { + "epoch": 0.7081796209133698, + "grad_norm": 2.4976108074188232, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8864890336990356, + "num_tokens": 216280898.0, + "step": 5567 + }, + { + "epoch": 0.7083068311919604, + "grad_norm": 2.522089719772339, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8890480399131775, + "num_tokens": 216320826.0, + "step": 5568 + }, + { + "epoch": 0.7084340414705508, + "grad_norm": 2.9896023273468018, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8741869926452637, + "num_tokens": 216351320.0, + "step": 5569 + }, + { + "epoch": 0.7085612517491413, + "grad_norm": 2.8036623001098633, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8924129605293274, + "num_tokens": 216383028.0, + "step": 5570 + }, + { + "epoch": 0.7086884620277318, + "grad_norm": 2.5337984561920166, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8813472390174866, + "num_tokens": 216427680.0, + "step": 5571 + }, + { + "epoch": 0.7088156723063224, + "grad_norm": 2.737396001815796, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8771711587905884, + "num_tokens": 216465666.0, + "step": 5572 + }, + { + "epoch": 0.7089428825849129, + "grad_norm": 2.6753973960876465, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8768319487571716, + "num_tokens": 216506969.0, + "step": 5573 + }, + { + "epoch": 0.7090700928635034, + "grad_norm": 2.430417776107788, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8899879455566406, + "num_tokens": 216545903.0, + "step": 5574 + }, + { + "epoch": 0.7091973031420938, + "grad_norm": 2.5495455265045166, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8662041425704956, + "num_tokens": 216589607.0, + "step": 5575 + }, + { + "epoch": 0.7093245134206844, + "grad_norm": 2.6359212398529053, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8754987716674805, + "num_tokens": 216627289.0, + "step": 5576 + }, + { + "epoch": 0.7094517236992749, + "grad_norm": 2.3762223720550537, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8777422904968262, + "num_tokens": 216670691.0, + "step": 5577 + }, + { + "epoch": 0.7095789339778654, + "grad_norm": 2.613389015197754, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8860467672348022, + "num_tokens": 216707598.0, + "step": 5578 + }, + { + "epoch": 0.709706144256456, + "grad_norm": 2.6989917755126953, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8729188442230225, + "num_tokens": 216747797.0, + "step": 5579 + }, + { + "epoch": 0.7098333545350465, + "grad_norm": 2.5827693939208984, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8761682510375977, + "num_tokens": 216787497.0, + "step": 5580 + }, + { + "epoch": 0.7099605648136369, + "grad_norm": 2.685664176940918, + "learning_rate": 1e-06, + "loss": 0.3777, + "mean_token_accuracy": 0.868247926235199, + "num_tokens": 216826922.0, + "step": 5581 + }, + { + "epoch": 0.7100877750922274, + "grad_norm": 2.4987921714782715, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8800048828125, + "num_tokens": 216869840.0, + "step": 5582 + }, + { + "epoch": 0.710214985370818, + "grad_norm": 2.74277925491333, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.8641879558563232, + "num_tokens": 216905615.0, + "step": 5583 + }, + { + "epoch": 0.7103421956494085, + "grad_norm": 2.806934118270874, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8957861661911011, + "num_tokens": 216936187.0, + "step": 5584 + }, + { + "epoch": 0.710469405927999, + "grad_norm": 2.8113691806793213, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8716139197349548, + "num_tokens": 216972536.0, + "step": 5585 + }, + { + "epoch": 0.7105966162065895, + "grad_norm": 3.040220022201538, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8840648531913757, + "num_tokens": 217014418.0, + "step": 5586 + }, + { + "epoch": 0.71072382648518, + "grad_norm": 2.497688055038452, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8975605964660645, + "num_tokens": 217049422.0, + "step": 5587 + }, + { + "epoch": 0.7108510367637705, + "grad_norm": 2.3619143962860107, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8851158618927002, + "num_tokens": 217091826.0, + "step": 5588 + }, + { + "epoch": 0.710978247042361, + "grad_norm": 2.275916337966919, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8820478916168213, + "num_tokens": 217140915.0, + "step": 5589 + }, + { + "epoch": 0.7111054573209515, + "grad_norm": 2.5752017498016357, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8845053911209106, + "num_tokens": 217180559.0, + "step": 5590 + }, + { + "epoch": 0.7112326675995421, + "grad_norm": 2.724646806716919, + "learning_rate": 1e-06, + "loss": 0.3796, + "mean_token_accuracy": 0.866068959236145, + "num_tokens": 217218212.0, + "step": 5591 + }, + { + "epoch": 0.7113598778781326, + "grad_norm": 2.4088242053985596, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8892713785171509, + "num_tokens": 217258447.0, + "step": 5592 + }, + { + "epoch": 0.711487088156723, + "grad_norm": 2.6585216522216797, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8840824961662292, + "num_tokens": 217293804.0, + "step": 5593 + }, + { + "epoch": 0.7116142984353135, + "grad_norm": 2.399453639984131, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8795099258422852, + "num_tokens": 217334758.0, + "step": 5594 + }, + { + "epoch": 0.7117415087139041, + "grad_norm": 2.6187968254089355, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8814808130264282, + "num_tokens": 217371093.0, + "step": 5595 + }, + { + "epoch": 0.7118687189924946, + "grad_norm": 2.598412036895752, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.874670147895813, + "num_tokens": 217411135.0, + "step": 5596 + }, + { + "epoch": 0.7119959292710851, + "grad_norm": 2.7048792839050293, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8746957778930664, + "num_tokens": 217448672.0, + "step": 5597 + }, + { + "epoch": 0.7121231395496757, + "grad_norm": 2.6337785720825195, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.8640149831771851, + "num_tokens": 217493558.0, + "step": 5598 + }, + { + "epoch": 0.7122503498282661, + "grad_norm": 2.5161173343658447, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8819714784622192, + "num_tokens": 217539449.0, + "step": 5599 + }, + { + "epoch": 0.7123775601068566, + "grad_norm": 2.672374725341797, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8739726543426514, + "num_tokens": 217576621.0, + "step": 5600 + }, + { + "epoch": 0.7125047703854471, + "grad_norm": 2.614994764328003, + "learning_rate": 1e-06, + "loss": 0.3872, + "mean_token_accuracy": 0.8635931015014648, + "num_tokens": 217617034.0, + "step": 5601 + }, + { + "epoch": 0.7126319806640377, + "grad_norm": 2.450566530227661, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8714370131492615, + "num_tokens": 217661573.0, + "step": 5602 + }, + { + "epoch": 0.7127591909426282, + "grad_norm": 2.7140700817108154, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8839021921157837, + "num_tokens": 217694280.0, + "step": 5603 + }, + { + "epoch": 0.7128864012212187, + "grad_norm": 2.7171101570129395, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.88075852394104, + "num_tokens": 217729781.0, + "step": 5604 + }, + { + "epoch": 0.7130136114998091, + "grad_norm": 2.804081678390503, + "learning_rate": 1e-06, + "loss": 0.3932, + "mean_token_accuracy": 0.8640110492706299, + "num_tokens": 217768009.0, + "step": 5605 + }, + { + "epoch": 0.7131408217783997, + "grad_norm": 2.6832029819488525, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8817904591560364, + "num_tokens": 217803693.0, + "step": 5606 + }, + { + "epoch": 0.7132680320569902, + "grad_norm": 2.6658575534820557, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8781890273094177, + "num_tokens": 217839782.0, + "step": 5607 + }, + { + "epoch": 0.7133952423355807, + "grad_norm": 2.5224249362945557, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8731255531311035, + "num_tokens": 217880892.0, + "step": 5608 + }, + { + "epoch": 0.7135224526141712, + "grad_norm": 2.6536595821380615, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8745715618133545, + "num_tokens": 217921429.0, + "step": 5609 + }, + { + "epoch": 0.7136496628927618, + "grad_norm": 2.6854376792907715, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8800216317176819, + "num_tokens": 217961413.0, + "step": 5610 + }, + { + "epoch": 0.7137768731713523, + "grad_norm": 2.4708948135375977, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8926019668579102, + "num_tokens": 217999723.0, + "step": 5611 + }, + { + "epoch": 0.7139040834499427, + "grad_norm": 2.5609376430511475, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8694679737091064, + "num_tokens": 218040803.0, + "step": 5612 + }, + { + "epoch": 0.7140312937285332, + "grad_norm": 2.5471391677856445, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8915212154388428, + "num_tokens": 218076951.0, + "step": 5613 + }, + { + "epoch": 0.7141585040071238, + "grad_norm": 2.7612617015838623, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8774193525314331, + "num_tokens": 218112422.0, + "step": 5614 + }, + { + "epoch": 0.7142857142857143, + "grad_norm": 2.7816061973571777, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8728909492492676, + "num_tokens": 218150255.0, + "step": 5615 + }, + { + "epoch": 0.7144129245643048, + "grad_norm": 2.614121198654175, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8828613758087158, + "num_tokens": 218185547.0, + "step": 5616 + }, + { + "epoch": 0.7145401348428954, + "grad_norm": 3.2920870780944824, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8837482929229736, + "num_tokens": 218222522.0, + "step": 5617 + }, + { + "epoch": 0.7146673451214858, + "grad_norm": 2.5280392169952393, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8824854493141174, + "num_tokens": 218260015.0, + "step": 5618 + }, + { + "epoch": 0.7147945554000763, + "grad_norm": 2.5882461071014404, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8822957277297974, + "num_tokens": 218297509.0, + "step": 5619 + }, + { + "epoch": 0.7149217656786668, + "grad_norm": 2.4789934158325195, + "learning_rate": 1e-06, + "loss": 0.4042, + "mean_token_accuracy": 0.8627111911773682, + "num_tokens": 218342948.0, + "step": 5620 + }, + { + "epoch": 0.7150489759572574, + "grad_norm": 2.6076903343200684, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8691271543502808, + "num_tokens": 218383590.0, + "step": 5621 + }, + { + "epoch": 0.7151761862358479, + "grad_norm": 2.7283425331115723, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8722445964813232, + "num_tokens": 218422878.0, + "step": 5622 + }, + { + "epoch": 0.7153033965144384, + "grad_norm": 2.524822235107422, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8773289918899536, + "num_tokens": 218465749.0, + "step": 5623 + }, + { + "epoch": 0.7154306067930288, + "grad_norm": 2.467426300048828, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.873620331287384, + "num_tokens": 218507447.0, + "step": 5624 + }, + { + "epoch": 0.7155578170716194, + "grad_norm": 2.575237274169922, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8832969665527344, + "num_tokens": 218542709.0, + "step": 5625 + }, + { + "epoch": 0.7156850273502099, + "grad_norm": 2.7155814170837402, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8814858198165894, + "num_tokens": 218577110.0, + "step": 5626 + }, + { + "epoch": 0.7158122376288004, + "grad_norm": 2.532017707824707, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8787329196929932, + "num_tokens": 218616624.0, + "step": 5627 + }, + { + "epoch": 0.715939447907391, + "grad_norm": 2.617422342300415, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.889062762260437, + "num_tokens": 218652699.0, + "step": 5628 + }, + { + "epoch": 0.7160666581859815, + "grad_norm": 2.6817424297332764, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8842557668685913, + "num_tokens": 218686147.0, + "step": 5629 + }, + { + "epoch": 0.7161938684645719, + "grad_norm": 2.8775856494903564, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8792355060577393, + "num_tokens": 218717251.0, + "step": 5630 + }, + { + "epoch": 0.7163210787431624, + "grad_norm": 2.6724295616149902, + "learning_rate": 1e-06, + "loss": 0.4093, + "mean_token_accuracy": 0.8554565906524658, + "num_tokens": 218758201.0, + "step": 5631 + }, + { + "epoch": 0.716448289021753, + "grad_norm": 2.5201547145843506, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8818296194076538, + "num_tokens": 218799423.0, + "step": 5632 + }, + { + "epoch": 0.7165754993003435, + "grad_norm": 2.6437511444091797, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8702467679977417, + "num_tokens": 218836294.0, + "step": 5633 + }, + { + "epoch": 0.716702709578934, + "grad_norm": 2.491896390914917, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8807178735733032, + "num_tokens": 218877868.0, + "step": 5634 + }, + { + "epoch": 0.7168299198575245, + "grad_norm": 2.721052646636963, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8807130455970764, + "num_tokens": 218910094.0, + "step": 5635 + }, + { + "epoch": 0.716957130136115, + "grad_norm": 2.4624431133270264, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8856526613235474, + "num_tokens": 218948221.0, + "step": 5636 + }, + { + "epoch": 0.7170843404147055, + "grad_norm": 2.4951367378234863, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8771572113037109, + "num_tokens": 218988047.0, + "step": 5637 + }, + { + "epoch": 0.717211550693296, + "grad_norm": 2.4663562774658203, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8833805322647095, + "num_tokens": 219027942.0, + "step": 5638 + }, + { + "epoch": 0.7173387609718865, + "grad_norm": 2.6695690155029297, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.8646464943885803, + "num_tokens": 219067819.0, + "step": 5639 + }, + { + "epoch": 0.7174659712504771, + "grad_norm": 2.5503487586975098, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.877763569355011, + "num_tokens": 219103737.0, + "step": 5640 + }, + { + "epoch": 0.7175931815290676, + "grad_norm": 2.538689136505127, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8948967456817627, + "num_tokens": 219138962.0, + "step": 5641 + }, + { + "epoch": 0.717720391807658, + "grad_norm": 2.5213348865509033, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8811744451522827, + "num_tokens": 219177434.0, + "step": 5642 + }, + { + "epoch": 0.7178476020862485, + "grad_norm": 2.4307634830474854, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8777220845222473, + "num_tokens": 219218602.0, + "step": 5643 + }, + { + "epoch": 0.7179748123648391, + "grad_norm": 2.885913133621216, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8790201544761658, + "num_tokens": 219251637.0, + "step": 5644 + }, + { + "epoch": 0.7181020226434296, + "grad_norm": 2.679046154022217, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8964376449584961, + "num_tokens": 219283689.0, + "step": 5645 + }, + { + "epoch": 0.7182292329220201, + "grad_norm": 2.43267560005188, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.882757306098938, + "num_tokens": 219322614.0, + "step": 5646 + }, + { + "epoch": 0.7183564432006107, + "grad_norm": 2.5963470935821533, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8808109164237976, + "num_tokens": 219363090.0, + "step": 5647 + }, + { + "epoch": 0.7184836534792011, + "grad_norm": 3.0544238090515137, + "learning_rate": 1e-06, + "loss": 0.4214, + "mean_token_accuracy": 0.8515936136245728, + "num_tokens": 219401674.0, + "step": 5648 + }, + { + "epoch": 0.7186108637577916, + "grad_norm": 2.726332187652588, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8685276508331299, + "num_tokens": 219438867.0, + "step": 5649 + }, + { + "epoch": 0.7187380740363821, + "grad_norm": 2.6969382762908936, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8796263933181763, + "num_tokens": 219477372.0, + "step": 5650 + }, + { + "epoch": 0.7188652843149727, + "grad_norm": 2.77211856842041, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8793137073516846, + "num_tokens": 219514086.0, + "step": 5651 + }, + { + "epoch": 0.7189924945935632, + "grad_norm": 2.673354148864746, + "learning_rate": 1e-06, + "loss": 0.395, + "mean_token_accuracy": 0.8614516854286194, + "num_tokens": 219553349.0, + "step": 5652 + }, + { + "epoch": 0.7191197048721537, + "grad_norm": 2.735884189605713, + "learning_rate": 1e-06, + "loss": 0.4104, + "mean_token_accuracy": 0.855933666229248, + "num_tokens": 219592103.0, + "step": 5653 + }, + { + "epoch": 0.7192469151507441, + "grad_norm": 2.5024197101593018, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8653881549835205, + "num_tokens": 219636591.0, + "step": 5654 + }, + { + "epoch": 0.7193741254293347, + "grad_norm": 3.0220930576324463, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8700804710388184, + "num_tokens": 219669808.0, + "step": 5655 + }, + { + "epoch": 0.7195013357079252, + "grad_norm": 2.5426156520843506, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8940364122390747, + "num_tokens": 219705892.0, + "step": 5656 + }, + { + "epoch": 0.7196285459865157, + "grad_norm": 2.6006877422332764, + "learning_rate": 1e-06, + "loss": 0.3811, + "mean_token_accuracy": 0.8672774434089661, + "num_tokens": 219749811.0, + "step": 5657 + }, + { + "epoch": 0.7197557562651062, + "grad_norm": 2.7651724815368652, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8838750123977661, + "num_tokens": 219782369.0, + "step": 5658 + }, + { + "epoch": 0.7198829665436968, + "grad_norm": 2.3175747394561768, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8806390166282654, + "num_tokens": 219829051.0, + "step": 5659 + }, + { + "epoch": 0.7200101768222873, + "grad_norm": 2.436135768890381, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8809605836868286, + "num_tokens": 219866221.0, + "step": 5660 + }, + { + "epoch": 0.7201373871008777, + "grad_norm": 2.4676992893218994, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8799816370010376, + "num_tokens": 219910519.0, + "step": 5661 + }, + { + "epoch": 0.7202645973794682, + "grad_norm": 2.7316184043884277, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8735581040382385, + "num_tokens": 219949074.0, + "step": 5662 + }, + { + "epoch": 0.7203918076580588, + "grad_norm": 2.8091373443603516, + "learning_rate": 1e-06, + "loss": 0.4012, + "mean_token_accuracy": 0.8625487685203552, + "num_tokens": 219985926.0, + "step": 5663 + }, + { + "epoch": 0.7205190179366493, + "grad_norm": 2.5302734375, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8888602256774902, + "num_tokens": 220020782.0, + "step": 5664 + }, + { + "epoch": 0.7206462282152398, + "grad_norm": 2.707158327102661, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8823003172874451, + "num_tokens": 220055521.0, + "step": 5665 + }, + { + "epoch": 0.7207734384938304, + "grad_norm": 2.8622281551361084, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8718315362930298, + "num_tokens": 220093650.0, + "step": 5666 + }, + { + "epoch": 0.7209006487724208, + "grad_norm": 2.724348306655884, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8896409273147583, + "num_tokens": 220126263.0, + "step": 5667 + }, + { + "epoch": 0.7210278590510113, + "grad_norm": 3.20367431640625, + "learning_rate": 1e-06, + "loss": 0.4074, + "mean_token_accuracy": 0.8599926829338074, + "num_tokens": 220154857.0, + "step": 5668 + }, + { + "epoch": 0.7211550693296018, + "grad_norm": 2.4017648696899414, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.888375461101532, + "num_tokens": 220197775.0, + "step": 5669 + }, + { + "epoch": 0.7212822796081924, + "grad_norm": 2.7346577644348145, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8756762742996216, + "num_tokens": 220233125.0, + "step": 5670 + }, + { + "epoch": 0.7214094898867829, + "grad_norm": 2.546679973602295, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8904304504394531, + "num_tokens": 220269820.0, + "step": 5671 + }, + { + "epoch": 0.7215367001653734, + "grad_norm": 2.480001449584961, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8815852403640747, + "num_tokens": 220308613.0, + "step": 5672 + }, + { + "epoch": 0.7216639104439638, + "grad_norm": 2.7529942989349365, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8711563348770142, + "num_tokens": 220345203.0, + "step": 5673 + }, + { + "epoch": 0.7217911207225544, + "grad_norm": 2.7455978393554688, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.881676197052002, + "num_tokens": 220377612.0, + "step": 5674 + }, + { + "epoch": 0.7219183310011449, + "grad_norm": 2.5664420127868652, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8930081129074097, + "num_tokens": 220412776.0, + "step": 5675 + }, + { + "epoch": 0.7220455412797354, + "grad_norm": 2.6241114139556885, + "learning_rate": 1e-06, + "loss": 0.3837, + "mean_token_accuracy": 0.8660586476325989, + "num_tokens": 220453909.0, + "step": 5676 + }, + { + "epoch": 0.7221727515583259, + "grad_norm": 2.4883503913879395, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8832166194915771, + "num_tokens": 220491850.0, + "step": 5677 + }, + { + "epoch": 0.7222999618369165, + "grad_norm": 2.679291009902954, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8848893642425537, + "num_tokens": 220529104.0, + "step": 5678 + }, + { + "epoch": 0.7224271721155069, + "grad_norm": 2.5697951316833496, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8807289004325867, + "num_tokens": 220569543.0, + "step": 5679 + }, + { + "epoch": 0.7225543823940974, + "grad_norm": 2.746026039123535, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8845751881599426, + "num_tokens": 220602531.0, + "step": 5680 + }, + { + "epoch": 0.722681592672688, + "grad_norm": 2.430790424346924, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8863438367843628, + "num_tokens": 220642805.0, + "step": 5681 + }, + { + "epoch": 0.7228088029512785, + "grad_norm": 2.5653164386749268, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8795919418334961, + "num_tokens": 220681272.0, + "step": 5682 + }, + { + "epoch": 0.722936013229869, + "grad_norm": 2.323078155517578, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8927229642868042, + "num_tokens": 220723722.0, + "step": 5683 + }, + { + "epoch": 0.7230632235084595, + "grad_norm": 2.446500778198242, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8765019774436951, + "num_tokens": 220766435.0, + "step": 5684 + }, + { + "epoch": 0.72319043378705, + "grad_norm": 2.401740550994873, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8717045187950134, + "num_tokens": 220812247.0, + "step": 5685 + }, + { + "epoch": 0.7233176440656405, + "grad_norm": 2.786034345626831, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8855591416358948, + "num_tokens": 220842661.0, + "step": 5686 + }, + { + "epoch": 0.723444854344231, + "grad_norm": 2.606834888458252, + "learning_rate": 1e-06, + "loss": 0.3835, + "mean_token_accuracy": 0.8681585788726807, + "num_tokens": 220883001.0, + "step": 5687 + }, + { + "epoch": 0.7235720646228215, + "grad_norm": 2.6520702838897705, + "learning_rate": 1e-06, + "loss": 0.3813, + "mean_token_accuracy": 0.8690348863601685, + "num_tokens": 220925129.0, + "step": 5688 + }, + { + "epoch": 0.7236992749014121, + "grad_norm": 2.530479907989502, + "learning_rate": 1e-06, + "loss": 0.3975, + "mean_token_accuracy": 0.8621718883514404, + "num_tokens": 220973032.0, + "step": 5689 + }, + { + "epoch": 0.7238264851800026, + "grad_norm": 2.6719930171966553, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8757798671722412, + "num_tokens": 221010294.0, + "step": 5690 + }, + { + "epoch": 0.723953695458593, + "grad_norm": 2.8429858684539795, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8680406808853149, + "num_tokens": 221044204.0, + "step": 5691 + }, + { + "epoch": 0.7240809057371835, + "grad_norm": 2.619436264038086, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8663226366043091, + "num_tokens": 221085286.0, + "step": 5692 + }, + { + "epoch": 0.7242081160157741, + "grad_norm": 2.475398063659668, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8801361322402954, + "num_tokens": 221127000.0, + "step": 5693 + }, + { + "epoch": 0.7243353262943646, + "grad_norm": 2.7974610328674316, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8708289861679077, + "num_tokens": 221160310.0, + "step": 5694 + }, + { + "epoch": 0.7244625365729551, + "grad_norm": 2.5368101596832275, + "learning_rate": 1e-06, + "loss": 0.3816, + "mean_token_accuracy": 0.8691737055778503, + "num_tokens": 221200494.0, + "step": 5695 + }, + { + "epoch": 0.7245897468515456, + "grad_norm": 2.3561389446258545, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8994811177253723, + "num_tokens": 221240714.0, + "step": 5696 + }, + { + "epoch": 0.7247169571301361, + "grad_norm": 3.003718376159668, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8720972537994385, + "num_tokens": 221271895.0, + "step": 5697 + }, + { + "epoch": 0.7248441674087266, + "grad_norm": 2.5969855785369873, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.87298583984375, + "num_tokens": 221316071.0, + "step": 5698 + }, + { + "epoch": 0.7249713776873171, + "grad_norm": 2.694044589996338, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8753310441970825, + "num_tokens": 221351776.0, + "step": 5699 + }, + { + "epoch": 0.7250985879659076, + "grad_norm": 2.474977731704712, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8793707489967346, + "num_tokens": 221393774.0, + "step": 5700 + }, + { + "epoch": 0.7252257982444982, + "grad_norm": 2.7881569862365723, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8804861903190613, + "num_tokens": 221428642.0, + "step": 5701 + }, + { + "epoch": 0.7253530085230887, + "grad_norm": 2.2880754470825195, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8983508348464966, + "num_tokens": 221469925.0, + "step": 5702 + }, + { + "epoch": 0.7254802188016791, + "grad_norm": 2.7787272930145264, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8841349482536316, + "num_tokens": 221503070.0, + "step": 5703 + }, + { + "epoch": 0.7256074290802697, + "grad_norm": 2.4468791484832764, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8813730478286743, + "num_tokens": 221544917.0, + "step": 5704 + }, + { + "epoch": 0.7257346393588602, + "grad_norm": 2.50675630569458, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8866638541221619, + "num_tokens": 221583486.0, + "step": 5705 + }, + { + "epoch": 0.7258618496374507, + "grad_norm": 2.912971258163452, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8721318244934082, + "num_tokens": 221617131.0, + "step": 5706 + }, + { + "epoch": 0.7259890599160412, + "grad_norm": 2.8277366161346436, + "learning_rate": 1e-06, + "loss": 0.4012, + "mean_token_accuracy": 0.861477255821228, + "num_tokens": 221652418.0, + "step": 5707 + }, + { + "epoch": 0.7261162701946318, + "grad_norm": 2.3773319721221924, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8881651759147644, + "num_tokens": 221692665.0, + "step": 5708 + }, + { + "epoch": 0.7262434804732223, + "grad_norm": 2.5501129627227783, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8680155277252197, + "num_tokens": 221735260.0, + "step": 5709 + }, + { + "epoch": 0.7263706907518127, + "grad_norm": 2.580749988555908, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8831173181533813, + "num_tokens": 221772800.0, + "step": 5710 + }, + { + "epoch": 0.7264979010304032, + "grad_norm": 2.393047571182251, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8788053393363953, + "num_tokens": 221815735.0, + "step": 5711 + }, + { + "epoch": 0.7266251113089938, + "grad_norm": 2.7023675441741943, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8800122141838074, + "num_tokens": 221849632.0, + "step": 5712 + }, + { + "epoch": 0.7267523215875843, + "grad_norm": 2.536325216293335, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8840861320495605, + "num_tokens": 221891211.0, + "step": 5713 + }, + { + "epoch": 0.7268795318661748, + "grad_norm": 2.454838991165161, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8915339112281799, + "num_tokens": 221929276.0, + "step": 5714 + }, + { + "epoch": 0.7270067421447653, + "grad_norm": 2.4689550399780273, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8876878023147583, + "num_tokens": 221964733.0, + "step": 5715 + }, + { + "epoch": 0.7271339524233558, + "grad_norm": 2.5281879901885986, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8816930055618286, + "num_tokens": 222003565.0, + "step": 5716 + }, + { + "epoch": 0.7272611627019463, + "grad_norm": 2.866811752319336, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8774293661117554, + "num_tokens": 222035716.0, + "step": 5717 + }, + { + "epoch": 0.7273883729805368, + "grad_norm": 2.448320150375366, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8823255300521851, + "num_tokens": 222077848.0, + "step": 5718 + }, + { + "epoch": 0.7275155832591274, + "grad_norm": 2.4265453815460205, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8838764429092407, + "num_tokens": 222123347.0, + "step": 5719 + }, + { + "epoch": 0.7276427935377179, + "grad_norm": 2.6802144050598145, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8733087778091431, + "num_tokens": 222158521.0, + "step": 5720 + }, + { + "epoch": 0.7277700038163084, + "grad_norm": 2.603752851486206, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8810195326805115, + "num_tokens": 222194789.0, + "step": 5721 + }, + { + "epoch": 0.7278972140948988, + "grad_norm": 2.421327829360962, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8792964220046997, + "num_tokens": 222239917.0, + "step": 5722 + }, + { + "epoch": 0.7280244243734894, + "grad_norm": 2.818903923034668, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8788436651229858, + "num_tokens": 222273509.0, + "step": 5723 + }, + { + "epoch": 0.7281516346520799, + "grad_norm": 2.408268928527832, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8840571641921997, + "num_tokens": 222313717.0, + "step": 5724 + }, + { + "epoch": 0.7282788449306704, + "grad_norm": 2.6548702716827393, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8806555271148682, + "num_tokens": 222349308.0, + "step": 5725 + }, + { + "epoch": 0.7284060552092609, + "grad_norm": 2.582261085510254, + "learning_rate": 1e-06, + "loss": 0.3911, + "mean_token_accuracy": 0.8640516400337219, + "num_tokens": 222390366.0, + "step": 5726 + }, + { + "epoch": 0.7285332654878515, + "grad_norm": 2.7458086013793945, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.880972146987915, + "num_tokens": 222425053.0, + "step": 5727 + }, + { + "epoch": 0.7286604757664419, + "grad_norm": 2.740326166152954, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8823481798171997, + "num_tokens": 222463045.0, + "step": 5728 + }, + { + "epoch": 0.7287876860450324, + "grad_norm": 2.4832231998443604, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8778758645057678, + "num_tokens": 222505125.0, + "step": 5729 + }, + { + "epoch": 0.7289148963236229, + "grad_norm": 2.637242555618286, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.883887529373169, + "num_tokens": 222538931.0, + "step": 5730 + }, + { + "epoch": 0.7290421066022135, + "grad_norm": 2.527465581893921, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8753288984298706, + "num_tokens": 222584544.0, + "step": 5731 + }, + { + "epoch": 0.729169316880804, + "grad_norm": 2.4144837856292725, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8832862377166748, + "num_tokens": 222626736.0, + "step": 5732 + }, + { + "epoch": 0.7292965271593945, + "grad_norm": 2.775561809539795, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.874443531036377, + "num_tokens": 222663726.0, + "step": 5733 + }, + { + "epoch": 0.7294237374379849, + "grad_norm": 2.670908212661743, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8769111633300781, + "num_tokens": 222697452.0, + "step": 5734 + }, + { + "epoch": 0.7295509477165755, + "grad_norm": 2.4781439304351807, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8847916126251221, + "num_tokens": 222736064.0, + "step": 5735 + }, + { + "epoch": 0.729678157995166, + "grad_norm": 2.3915798664093018, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8846204280853271, + "num_tokens": 222776814.0, + "step": 5736 + }, + { + "epoch": 0.7298053682737565, + "grad_norm": 2.4983296394348145, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8782464861869812, + "num_tokens": 222817677.0, + "step": 5737 + }, + { + "epoch": 0.7299325785523471, + "grad_norm": 2.6083567142486572, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8797982931137085, + "num_tokens": 222856445.0, + "step": 5738 + }, + { + "epoch": 0.7300597888309376, + "grad_norm": 2.691788911819458, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8757594227790833, + "num_tokens": 222891691.0, + "step": 5739 + }, + { + "epoch": 0.730186999109528, + "grad_norm": 2.3441295623779297, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8879342079162598, + "num_tokens": 222931791.0, + "step": 5740 + }, + { + "epoch": 0.7303142093881185, + "grad_norm": 2.507357358932495, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8835857510566711, + "num_tokens": 222972247.0, + "step": 5741 + }, + { + "epoch": 0.7304414196667091, + "grad_norm": 2.6208548545837402, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8851057291030884, + "num_tokens": 223010398.0, + "step": 5742 + }, + { + "epoch": 0.7305686299452996, + "grad_norm": 2.454479932785034, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8825171589851379, + "num_tokens": 223049077.0, + "step": 5743 + }, + { + "epoch": 0.7306958402238901, + "grad_norm": 2.4158742427825928, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8788359761238098, + "num_tokens": 223091179.0, + "step": 5744 + }, + { + "epoch": 0.7308230505024806, + "grad_norm": 2.4149692058563232, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8705341815948486, + "num_tokens": 223134989.0, + "step": 5745 + }, + { + "epoch": 0.7309502607810711, + "grad_norm": 2.570279359817505, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8779481649398804, + "num_tokens": 223173964.0, + "step": 5746 + }, + { + "epoch": 0.7310774710596616, + "grad_norm": 2.2699055671691895, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8885912895202637, + "num_tokens": 223214857.0, + "step": 5747 + }, + { + "epoch": 0.7312046813382521, + "grad_norm": 2.51065731048584, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8764828443527222, + "num_tokens": 223254925.0, + "step": 5748 + }, + { + "epoch": 0.7313318916168426, + "grad_norm": 2.516714572906494, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8770607113838196, + "num_tokens": 223298338.0, + "step": 5749 + }, + { + "epoch": 0.7314591018954332, + "grad_norm": 2.6057565212249756, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8742069005966187, + "num_tokens": 223336840.0, + "step": 5750 + }, + { + "epoch": 0.7315863121740237, + "grad_norm": 2.480698347091675, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8802381753921509, + "num_tokens": 223377740.0, + "step": 5751 + }, + { + "epoch": 0.7317135224526141, + "grad_norm": 2.8231334686279297, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8826197981834412, + "num_tokens": 223411838.0, + "step": 5752 + }, + { + "epoch": 0.7318407327312046, + "grad_norm": 2.5349600315093994, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8796562552452087, + "num_tokens": 223453546.0, + "step": 5753 + }, + { + "epoch": 0.7319679430097952, + "grad_norm": 2.47318959236145, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8863269090652466, + "num_tokens": 223489129.0, + "step": 5754 + }, + { + "epoch": 0.7320951532883857, + "grad_norm": 2.7775778770446777, + "learning_rate": 1e-06, + "loss": 0.3675, + "mean_token_accuracy": 0.8727226853370667, + "num_tokens": 223529802.0, + "step": 5755 + }, + { + "epoch": 0.7322223635669762, + "grad_norm": 2.6332406997680664, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8875141143798828, + "num_tokens": 223566235.0, + "step": 5756 + }, + { + "epoch": 0.7323495738455668, + "grad_norm": 2.5663397312164307, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8697192668914795, + "num_tokens": 223610209.0, + "step": 5757 + }, + { + "epoch": 0.7324767841241572, + "grad_norm": 2.5849626064300537, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8730679750442505, + "num_tokens": 223650392.0, + "step": 5758 + }, + { + "epoch": 0.7326039944027477, + "grad_norm": 2.7183949947357178, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8699718713760376, + "num_tokens": 223688792.0, + "step": 5759 + }, + { + "epoch": 0.7327312046813382, + "grad_norm": 2.405841827392578, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.874187171459198, + "num_tokens": 223732496.0, + "step": 5760 + }, + { + "epoch": 0.7328584149599288, + "grad_norm": 2.804715156555176, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8711661100387573, + "num_tokens": 223767936.0, + "step": 5761 + }, + { + "epoch": 0.7329856252385193, + "grad_norm": 2.447669744491577, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8771209716796875, + "num_tokens": 223810424.0, + "step": 5762 + }, + { + "epoch": 0.7331128355171098, + "grad_norm": 2.701317548751831, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8708947896957397, + "num_tokens": 223847012.0, + "step": 5763 + }, + { + "epoch": 0.7332400457957003, + "grad_norm": 2.6241352558135986, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8881952166557312, + "num_tokens": 223880685.0, + "step": 5764 + }, + { + "epoch": 0.7333672560742908, + "grad_norm": 2.512585401535034, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8865741491317749, + "num_tokens": 223918366.0, + "step": 5765 + }, + { + "epoch": 0.7334944663528813, + "grad_norm": 2.5206351280212402, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8735781908035278, + "num_tokens": 223959102.0, + "step": 5766 + }, + { + "epoch": 0.7336216766314718, + "grad_norm": 2.7025561332702637, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8800489902496338, + "num_tokens": 223995170.0, + "step": 5767 + }, + { + "epoch": 0.7337488869100623, + "grad_norm": 2.7011730670928955, + "learning_rate": 1e-06, + "loss": 0.3968, + "mean_token_accuracy": 0.8652330636978149, + "num_tokens": 224037287.0, + "step": 5768 + }, + { + "epoch": 0.7338760971886529, + "grad_norm": 2.496399402618408, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8870178461074829, + "num_tokens": 224074789.0, + "step": 5769 + }, + { + "epoch": 0.7340033074672434, + "grad_norm": 2.5186219215393066, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8761309385299683, + "num_tokens": 224116070.0, + "step": 5770 + }, + { + "epoch": 0.7341305177458338, + "grad_norm": 2.785823106765747, + "learning_rate": 1e-06, + "loss": 0.4253, + "mean_token_accuracy": 0.8552700281143188, + "num_tokens": 224152620.0, + "step": 5771 + }, + { + "epoch": 0.7342577280244243, + "grad_norm": 2.8824899196624756, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.868024468421936, + "num_tokens": 224188877.0, + "step": 5772 + }, + { + "epoch": 0.7343849383030149, + "grad_norm": 2.6145637035369873, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8739707469940186, + "num_tokens": 224229298.0, + "step": 5773 + }, + { + "epoch": 0.7345121485816054, + "grad_norm": 2.548731565475464, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8843086361885071, + "num_tokens": 224267276.0, + "step": 5774 + }, + { + "epoch": 0.7346393588601959, + "grad_norm": 2.4036505222320557, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8808377385139465, + "num_tokens": 224310759.0, + "step": 5775 + }, + { + "epoch": 0.7347665691387865, + "grad_norm": 2.730273962020874, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8766762018203735, + "num_tokens": 224347968.0, + "step": 5776 + }, + { + "epoch": 0.7348937794173769, + "grad_norm": 2.396871328353882, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.899890661239624, + "num_tokens": 224383488.0, + "step": 5777 + }, + { + "epoch": 0.7350209896959674, + "grad_norm": 2.343428373336792, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8778570890426636, + "num_tokens": 224430064.0, + "step": 5778 + }, + { + "epoch": 0.7351481999745579, + "grad_norm": 2.8973662853240967, + "learning_rate": 1e-06, + "loss": 0.392, + "mean_token_accuracy": 0.865816593170166, + "num_tokens": 224466989.0, + "step": 5779 + }, + { + "epoch": 0.7352754102531485, + "grad_norm": 2.7175614833831787, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8738696575164795, + "num_tokens": 224503408.0, + "step": 5780 + }, + { + "epoch": 0.735402620531739, + "grad_norm": 2.762066125869751, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8730465173721313, + "num_tokens": 224537200.0, + "step": 5781 + }, + { + "epoch": 0.7355298308103295, + "grad_norm": 2.6408908367156982, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8740797638893127, + "num_tokens": 224575604.0, + "step": 5782 + }, + { + "epoch": 0.7356570410889199, + "grad_norm": 2.627242088317871, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8870534896850586, + "num_tokens": 224612303.0, + "step": 5783 + }, + { + "epoch": 0.7357842513675105, + "grad_norm": 2.7331066131591797, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.8724415302276611, + "num_tokens": 224648073.0, + "step": 5784 + }, + { + "epoch": 0.735911461646101, + "grad_norm": 2.6261467933654785, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8775284886360168, + "num_tokens": 224686659.0, + "step": 5785 + }, + { + "epoch": 0.7360386719246915, + "grad_norm": 2.6911497116088867, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8672752380371094, + "num_tokens": 224725350.0, + "step": 5786 + }, + { + "epoch": 0.736165882203282, + "grad_norm": 2.5244367122650146, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8940675854682922, + "num_tokens": 224763027.0, + "step": 5787 + }, + { + "epoch": 0.7362930924818726, + "grad_norm": 2.5008585453033447, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8869838118553162, + "num_tokens": 224803634.0, + "step": 5788 + }, + { + "epoch": 0.736420302760463, + "grad_norm": 2.5322744846343994, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.887785792350769, + "num_tokens": 224842404.0, + "step": 5789 + }, + { + "epoch": 0.7365475130390535, + "grad_norm": 2.525921583175659, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8862524032592773, + "num_tokens": 224880166.0, + "step": 5790 + }, + { + "epoch": 0.736674723317644, + "grad_norm": 2.6779534816741943, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8632299304008484, + "num_tokens": 224919527.0, + "step": 5791 + }, + { + "epoch": 0.7368019335962346, + "grad_norm": 2.652052402496338, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8726176619529724, + "num_tokens": 224960213.0, + "step": 5792 + }, + { + "epoch": 0.7369291438748251, + "grad_norm": 2.6821558475494385, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8858979940414429, + "num_tokens": 225001256.0, + "step": 5793 + }, + { + "epoch": 0.7370563541534156, + "grad_norm": 2.6609299182891846, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8828352689743042, + "num_tokens": 225037941.0, + "step": 5794 + }, + { + "epoch": 0.737183564432006, + "grad_norm": 2.733891487121582, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.8758497834205627, + "num_tokens": 225071446.0, + "step": 5795 + }, + { + "epoch": 0.7373107747105966, + "grad_norm": 2.6748170852661133, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8802145719528198, + "num_tokens": 225108884.0, + "step": 5796 + }, + { + "epoch": 0.7374379849891871, + "grad_norm": 2.6467764377593994, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.873160719871521, + "num_tokens": 225148565.0, + "step": 5797 + }, + { + "epoch": 0.7375651952677776, + "grad_norm": 2.601203203201294, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8714880347251892, + "num_tokens": 225190767.0, + "step": 5798 + }, + { + "epoch": 0.7376924055463682, + "grad_norm": 2.43115496635437, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8835145235061646, + "num_tokens": 225235588.0, + "step": 5799 + }, + { + "epoch": 0.7378196158249587, + "grad_norm": 2.4321670532226562, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8839330077171326, + "num_tokens": 225276365.0, + "step": 5800 + }, + { + "epoch": 0.7379468261035491, + "grad_norm": 2.6021523475646973, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8835981488227844, + "num_tokens": 225311228.0, + "step": 5801 + }, + { + "epoch": 0.7380740363821396, + "grad_norm": 2.6347970962524414, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.880443811416626, + "num_tokens": 225350456.0, + "step": 5802 + }, + { + "epoch": 0.7382012466607302, + "grad_norm": 2.5368616580963135, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8854979872703552, + "num_tokens": 225390362.0, + "step": 5803 + }, + { + "epoch": 0.7383284569393207, + "grad_norm": 2.622382879257202, + "learning_rate": 1e-06, + "loss": 0.3871, + "mean_token_accuracy": 0.866153359413147, + "num_tokens": 225431191.0, + "step": 5804 + }, + { + "epoch": 0.7384556672179112, + "grad_norm": 2.6323065757751465, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.876949667930603, + "num_tokens": 225467048.0, + "step": 5805 + }, + { + "epoch": 0.7385828774965018, + "grad_norm": 2.523925542831421, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8747131824493408, + "num_tokens": 225505988.0, + "step": 5806 + }, + { + "epoch": 0.7387100877750922, + "grad_norm": 2.4184939861297607, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8889285922050476, + "num_tokens": 225545859.0, + "step": 5807 + }, + { + "epoch": 0.7388372980536827, + "grad_norm": 2.32326340675354, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.875625729560852, + "num_tokens": 225592789.0, + "step": 5808 + }, + { + "epoch": 0.7389645083322732, + "grad_norm": 2.390052556991577, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8735204935073853, + "num_tokens": 225638931.0, + "step": 5809 + }, + { + "epoch": 0.7390917186108638, + "grad_norm": 2.4153590202331543, + "learning_rate": 1e-06, + "loss": 0.3804, + "mean_token_accuracy": 0.8700479865074158, + "num_tokens": 225688269.0, + "step": 5810 + }, + { + "epoch": 0.7392189288894543, + "grad_norm": 2.736299991607666, + "learning_rate": 1e-06, + "loss": 0.3861, + "mean_token_accuracy": 0.8682085275650024, + "num_tokens": 225728505.0, + "step": 5811 + }, + { + "epoch": 0.7393461391680448, + "grad_norm": 2.3595755100250244, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8916420340538025, + "num_tokens": 225769465.0, + "step": 5812 + }, + { + "epoch": 0.7394733494466353, + "grad_norm": 2.4821906089782715, + "learning_rate": 1e-06, + "loss": 0.4056, + "mean_token_accuracy": 0.8603793382644653, + "num_tokens": 225816613.0, + "step": 5813 + }, + { + "epoch": 0.7396005597252258, + "grad_norm": 2.9835715293884277, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8796734809875488, + "num_tokens": 225846968.0, + "step": 5814 + }, + { + "epoch": 0.7397277700038163, + "grad_norm": 2.516908645629883, + "learning_rate": 1e-06, + "loss": 0.3785, + "mean_token_accuracy": 0.8702629804611206, + "num_tokens": 225887053.0, + "step": 5815 + }, + { + "epoch": 0.7398549802824068, + "grad_norm": 2.703460454940796, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8826377391815186, + "num_tokens": 225921282.0, + "step": 5816 + }, + { + "epoch": 0.7399821905609973, + "grad_norm": 2.5351362228393555, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8757708072662354, + "num_tokens": 225959438.0, + "step": 5817 + }, + { + "epoch": 0.7401094008395879, + "grad_norm": 2.621264934539795, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8698189854621887, + "num_tokens": 225999643.0, + "step": 5818 + }, + { + "epoch": 0.7402366111181784, + "grad_norm": 2.459637403488159, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8855746388435364, + "num_tokens": 226039634.0, + "step": 5819 + }, + { + "epoch": 0.7403638213967688, + "grad_norm": 2.7963807582855225, + "learning_rate": 1e-06, + "loss": 0.3807, + "mean_token_accuracy": 0.8679685592651367, + "num_tokens": 226079181.0, + "step": 5820 + }, + { + "epoch": 0.7404910316753593, + "grad_norm": 2.3391828536987305, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8872107267379761, + "num_tokens": 226124773.0, + "step": 5821 + }, + { + "epoch": 0.7406182419539499, + "grad_norm": 2.729060411453247, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8788836002349854, + "num_tokens": 226161427.0, + "step": 5822 + }, + { + "epoch": 0.7407454522325404, + "grad_norm": 2.599951982498169, + "learning_rate": 1e-06, + "loss": 0.3922, + "mean_token_accuracy": 0.865925669670105, + "num_tokens": 226204315.0, + "step": 5823 + }, + { + "epoch": 0.7408726625111309, + "grad_norm": 2.7739248275756836, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8724454641342163, + "num_tokens": 226240421.0, + "step": 5824 + }, + { + "epoch": 0.7409998727897215, + "grad_norm": 2.5895841121673584, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.871153712272644, + "num_tokens": 226281010.0, + "step": 5825 + }, + { + "epoch": 0.7411270830683119, + "grad_norm": 2.7146363258361816, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8724017143249512, + "num_tokens": 226319129.0, + "step": 5826 + }, + { + "epoch": 0.7412542933469024, + "grad_norm": 2.6102681159973145, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8886772990226746, + "num_tokens": 226354334.0, + "step": 5827 + }, + { + "epoch": 0.7413815036254929, + "grad_norm": 2.493870496749878, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8677652478218079, + "num_tokens": 226398586.0, + "step": 5828 + }, + { + "epoch": 0.7415087139040835, + "grad_norm": 2.779750108718872, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8720890283584595, + "num_tokens": 226435015.0, + "step": 5829 + }, + { + "epoch": 0.741635924182674, + "grad_norm": 2.505819320678711, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8819717764854431, + "num_tokens": 226476459.0, + "step": 5830 + }, + { + "epoch": 0.7417631344612645, + "grad_norm": 2.468958854675293, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8849203586578369, + "num_tokens": 226517632.0, + "step": 5831 + }, + { + "epoch": 0.7418903447398549, + "grad_norm": 2.338712215423584, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8777194023132324, + "num_tokens": 226562235.0, + "step": 5832 + }, + { + "epoch": 0.7420175550184455, + "grad_norm": 2.6863620281219482, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8744239807128906, + "num_tokens": 226599116.0, + "step": 5833 + }, + { + "epoch": 0.742144765297036, + "grad_norm": 2.482882499694824, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8911348581314087, + "num_tokens": 226635447.0, + "step": 5834 + }, + { + "epoch": 0.7422719755756265, + "grad_norm": 2.64786958694458, + "learning_rate": 1e-06, + "loss": 0.4041, + "mean_token_accuracy": 0.8594221472740173, + "num_tokens": 226678052.0, + "step": 5835 + }, + { + "epoch": 0.742399185854217, + "grad_norm": 2.5884463787078857, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8860929012298584, + "num_tokens": 226712072.0, + "step": 5836 + }, + { + "epoch": 0.7425263961328076, + "grad_norm": 2.491340160369873, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8842002749443054, + "num_tokens": 226752944.0, + "step": 5837 + }, + { + "epoch": 0.742653606411398, + "grad_norm": 2.602167844772339, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8784226179122925, + "num_tokens": 226792870.0, + "step": 5838 + }, + { + "epoch": 0.7427808166899885, + "grad_norm": 2.6305367946624756, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8706960678100586, + "num_tokens": 226830966.0, + "step": 5839 + }, + { + "epoch": 0.742908026968579, + "grad_norm": 2.789278984069824, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8791134357452393, + "num_tokens": 226865084.0, + "step": 5840 + }, + { + "epoch": 0.7430352372471696, + "grad_norm": 2.5850720405578613, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8678723573684692, + "num_tokens": 226905652.0, + "step": 5841 + }, + { + "epoch": 0.7431624475257601, + "grad_norm": 2.438807964324951, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8843857049942017, + "num_tokens": 226946009.0, + "step": 5842 + }, + { + "epoch": 0.7432896578043506, + "grad_norm": 2.613410711288452, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8788771033287048, + "num_tokens": 226986821.0, + "step": 5843 + }, + { + "epoch": 0.743416868082941, + "grad_norm": 2.684460401535034, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8739466071128845, + "num_tokens": 227021919.0, + "step": 5844 + }, + { + "epoch": 0.7435440783615316, + "grad_norm": 2.9424753189086914, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8752076625823975, + "num_tokens": 227053566.0, + "step": 5845 + }, + { + "epoch": 0.7436712886401221, + "grad_norm": 2.7144317626953125, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8743799924850464, + "num_tokens": 227090639.0, + "step": 5846 + }, + { + "epoch": 0.7437984989187126, + "grad_norm": 2.442582845687866, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8727183938026428, + "num_tokens": 227133096.0, + "step": 5847 + }, + { + "epoch": 0.7439257091973032, + "grad_norm": 2.666379690170288, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8738017082214355, + "num_tokens": 227173361.0, + "step": 5848 + }, + { + "epoch": 0.7440529194758937, + "grad_norm": 2.4312076568603516, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8776265978813171, + "num_tokens": 227214537.0, + "step": 5849 + }, + { + "epoch": 0.7441801297544841, + "grad_norm": 2.2626211643218994, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8770944476127625, + "num_tokens": 227259863.0, + "step": 5850 + }, + { + "epoch": 0.7443073400330746, + "grad_norm": 2.592526435852051, + "learning_rate": 1e-06, + "loss": 0.4032, + "mean_token_accuracy": 0.86162269115448, + "num_tokens": 227302500.0, + "step": 5851 + }, + { + "epoch": 0.7444345503116652, + "grad_norm": 2.6194796562194824, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8946315050125122, + "num_tokens": 227338214.0, + "step": 5852 + }, + { + "epoch": 0.7445617605902557, + "grad_norm": 2.3330955505371094, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8822142481803894, + "num_tokens": 227384415.0, + "step": 5853 + }, + { + "epoch": 0.7446889708688462, + "grad_norm": 2.494450807571411, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8729235529899597, + "num_tokens": 227426672.0, + "step": 5854 + }, + { + "epoch": 0.7448161811474368, + "grad_norm": 2.3843655586242676, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8872426748275757, + "num_tokens": 227468364.0, + "step": 5855 + }, + { + "epoch": 0.7449433914260272, + "grad_norm": 2.643411636352539, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.873386561870575, + "num_tokens": 227505548.0, + "step": 5856 + }, + { + "epoch": 0.7450706017046177, + "grad_norm": 3.0086984634399414, + "learning_rate": 1e-06, + "loss": 0.3855, + "mean_token_accuracy": 0.8702918887138367, + "num_tokens": 227538870.0, + "step": 5857 + }, + { + "epoch": 0.7451978119832082, + "grad_norm": 2.6151950359344482, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.8687183260917664, + "num_tokens": 227584212.0, + "step": 5858 + }, + { + "epoch": 0.7453250222617988, + "grad_norm": 2.8503565788269043, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.865395188331604, + "num_tokens": 227622714.0, + "step": 5859 + }, + { + "epoch": 0.7454522325403893, + "grad_norm": 2.6659114360809326, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8814635872840881, + "num_tokens": 227662350.0, + "step": 5860 + }, + { + "epoch": 0.7455794428189798, + "grad_norm": 2.560596227645874, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8795123100280762, + "num_tokens": 227700896.0, + "step": 5861 + }, + { + "epoch": 0.7457066530975703, + "grad_norm": 2.6635279655456543, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8874322175979614, + "num_tokens": 227735770.0, + "step": 5862 + }, + { + "epoch": 0.7458338633761608, + "grad_norm": 2.522848606109619, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8852200508117676, + "num_tokens": 227774502.0, + "step": 5863 + }, + { + "epoch": 0.7459610736547513, + "grad_norm": 2.8487842082977295, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8786948323249817, + "num_tokens": 227808761.0, + "step": 5864 + }, + { + "epoch": 0.7460882839333418, + "grad_norm": 2.621748208999634, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8832751512527466, + "num_tokens": 227847400.0, + "step": 5865 + }, + { + "epoch": 0.7462154942119323, + "grad_norm": 2.421858549118042, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.886833906173706, + "num_tokens": 227888423.0, + "step": 5866 + }, + { + "epoch": 0.7463427044905229, + "grad_norm": 2.5706210136413574, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8741024732589722, + "num_tokens": 227927256.0, + "step": 5867 + }, + { + "epoch": 0.7464699147691134, + "grad_norm": 2.541438102722168, + "learning_rate": 1e-06, + "loss": 0.3967, + "mean_token_accuracy": 0.866889476776123, + "num_tokens": 227967703.0, + "step": 5868 + }, + { + "epoch": 0.7465971250477038, + "grad_norm": 2.3581058979034424, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8901268243789673, + "num_tokens": 228011738.0, + "step": 5869 + }, + { + "epoch": 0.7467243353262943, + "grad_norm": 2.5644314289093018, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8712893724441528, + "num_tokens": 228049066.0, + "step": 5870 + }, + { + "epoch": 0.7468515456048849, + "grad_norm": 2.5286407470703125, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8795803189277649, + "num_tokens": 228093435.0, + "step": 5871 + }, + { + "epoch": 0.7469787558834754, + "grad_norm": 2.685816526412964, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8772644996643066, + "num_tokens": 228127498.0, + "step": 5872 + }, + { + "epoch": 0.7471059661620659, + "grad_norm": 2.5057756900787354, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8706425428390503, + "num_tokens": 228169515.0, + "step": 5873 + }, + { + "epoch": 0.7472331764406565, + "grad_norm": 2.4071459770202637, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8876858353614807, + "num_tokens": 228207034.0, + "step": 5874 + }, + { + "epoch": 0.7473603867192469, + "grad_norm": 2.8832430839538574, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.888556957244873, + "num_tokens": 228238006.0, + "step": 5875 + }, + { + "epoch": 0.7474875969978374, + "grad_norm": 2.520026922225952, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8749077916145325, + "num_tokens": 228282163.0, + "step": 5876 + }, + { + "epoch": 0.7476148072764279, + "grad_norm": 2.5895137786865234, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8877121210098267, + "num_tokens": 228319558.0, + "step": 5877 + }, + { + "epoch": 0.7477420175550185, + "grad_norm": 2.765089511871338, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8720450401306152, + "num_tokens": 228357471.0, + "step": 5878 + }, + { + "epoch": 0.747869227833609, + "grad_norm": 2.6970410346984863, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8743958473205566, + "num_tokens": 228395458.0, + "step": 5879 + }, + { + "epoch": 0.7479964381121995, + "grad_norm": 2.4375486373901367, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8781204223632812, + "num_tokens": 228437243.0, + "step": 5880 + }, + { + "epoch": 0.7481236483907899, + "grad_norm": 2.569592237472534, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8836398124694824, + "num_tokens": 228475301.0, + "step": 5881 + }, + { + "epoch": 0.7482508586693805, + "grad_norm": 2.4588568210601807, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8816419839859009, + "num_tokens": 228519008.0, + "step": 5882 + }, + { + "epoch": 0.748378068947971, + "grad_norm": 2.4768946170806885, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8831278085708618, + "num_tokens": 228560977.0, + "step": 5883 + }, + { + "epoch": 0.7485052792265615, + "grad_norm": 2.2301759719848633, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.903573215007782, + "num_tokens": 228601019.0, + "step": 5884 + }, + { + "epoch": 0.748632489505152, + "grad_norm": 2.2109639644622803, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8946313858032227, + "num_tokens": 228645273.0, + "step": 5885 + }, + { + "epoch": 0.7487596997837426, + "grad_norm": 2.3509902954101562, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8829959630966187, + "num_tokens": 228689414.0, + "step": 5886 + }, + { + "epoch": 0.748886910062333, + "grad_norm": 2.268350839614868, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8895353078842163, + "num_tokens": 228732782.0, + "step": 5887 + }, + { + "epoch": 0.7490141203409235, + "grad_norm": 2.5916059017181396, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8717055916786194, + "num_tokens": 228774660.0, + "step": 5888 + }, + { + "epoch": 0.749141330619514, + "grad_norm": 2.700673818588257, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8700576424598694, + "num_tokens": 228813541.0, + "step": 5889 + }, + { + "epoch": 0.7492685408981046, + "grad_norm": 2.4164350032806396, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8825667500495911, + "num_tokens": 228854324.0, + "step": 5890 + }, + { + "epoch": 0.7493957511766951, + "grad_norm": 2.4712047576904297, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8725125789642334, + "num_tokens": 228893707.0, + "step": 5891 + }, + { + "epoch": 0.7495229614552856, + "grad_norm": 2.3201396465301514, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8893353939056396, + "num_tokens": 228936092.0, + "step": 5892 + }, + { + "epoch": 0.749650171733876, + "grad_norm": 2.5721724033355713, + "learning_rate": 1e-06, + "loss": 0.401, + "mean_token_accuracy": 0.8590481281280518, + "num_tokens": 228978159.0, + "step": 5893 + }, + { + "epoch": 0.7497773820124666, + "grad_norm": 2.6362922191619873, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8732947111129761, + "num_tokens": 229014407.0, + "step": 5894 + }, + { + "epoch": 0.7499045922910571, + "grad_norm": 2.554306983947754, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.8730207085609436, + "num_tokens": 229054929.0, + "step": 5895 + }, + { + "epoch": 0.7500318025696476, + "grad_norm": 2.298924207687378, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8908814191818237, + "num_tokens": 229096997.0, + "step": 5896 + }, + { + "epoch": 0.7501590128482382, + "grad_norm": 2.581536054611206, + "learning_rate": 1e-06, + "loss": 0.4013, + "mean_token_accuracy": 0.8611523509025574, + "num_tokens": 229141532.0, + "step": 5897 + }, + { + "epoch": 0.7502862231268287, + "grad_norm": 2.542189121246338, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.884105384349823, + "num_tokens": 229178494.0, + "step": 5898 + }, + { + "epoch": 0.7504134334054191, + "grad_norm": 2.6417651176452637, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8875725865364075, + "num_tokens": 229211727.0, + "step": 5899 + }, + { + "epoch": 0.7505406436840096, + "grad_norm": 2.748345136642456, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8717738389968872, + "num_tokens": 229247831.0, + "step": 5900 + }, + { + "epoch": 0.7506678539626002, + "grad_norm": 2.815960645675659, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.868975043296814, + "num_tokens": 229283863.0, + "step": 5901 + }, + { + "epoch": 0.7507950642411907, + "grad_norm": 3.114027500152588, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8767918348312378, + "num_tokens": 229311932.0, + "step": 5902 + }, + { + "epoch": 0.7509222745197812, + "grad_norm": 2.4862020015716553, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.8660281300544739, + "num_tokens": 229358655.0, + "step": 5903 + }, + { + "epoch": 0.7510494847983717, + "grad_norm": 2.4041826725006104, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8789108991622925, + "num_tokens": 229403089.0, + "step": 5904 + }, + { + "epoch": 0.7511766950769622, + "grad_norm": 2.7228453159332275, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8842953443527222, + "num_tokens": 229437259.0, + "step": 5905 + }, + { + "epoch": 0.7513039053555527, + "grad_norm": 2.6477198600769043, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8783808946609497, + "num_tokens": 229473774.0, + "step": 5906 + }, + { + "epoch": 0.7514311156341432, + "grad_norm": 2.731135845184326, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8767828941345215, + "num_tokens": 229510656.0, + "step": 5907 + }, + { + "epoch": 0.7515583259127337, + "grad_norm": 2.5927913188934326, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.877601146697998, + "num_tokens": 229550495.0, + "step": 5908 + }, + { + "epoch": 0.7516855361913243, + "grad_norm": 2.687554121017456, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8935003280639648, + "num_tokens": 229587111.0, + "step": 5909 + }, + { + "epoch": 0.7518127464699148, + "grad_norm": 2.5982327461242676, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8859958648681641, + "num_tokens": 229625587.0, + "step": 5910 + }, + { + "epoch": 0.7519399567485053, + "grad_norm": 2.482774019241333, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8778843283653259, + "num_tokens": 229669971.0, + "step": 5911 + }, + { + "epoch": 0.7520671670270958, + "grad_norm": 2.7543506622314453, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8824889659881592, + "num_tokens": 229704197.0, + "step": 5912 + }, + { + "epoch": 0.7521943773056863, + "grad_norm": 2.738589286804199, + "learning_rate": 1e-06, + "loss": 0.391, + "mean_token_accuracy": 0.8642268180847168, + "num_tokens": 229742942.0, + "step": 5913 + }, + { + "epoch": 0.7523215875842768, + "grad_norm": 2.721597194671631, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.885614812374115, + "num_tokens": 229781826.0, + "step": 5914 + }, + { + "epoch": 0.7524487978628673, + "grad_norm": 2.4794795513153076, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8998938798904419, + "num_tokens": 229820210.0, + "step": 5915 + }, + { + "epoch": 0.7525760081414579, + "grad_norm": 2.5221574306488037, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8740314841270447, + "num_tokens": 229864195.0, + "step": 5916 + }, + { + "epoch": 0.7527032184200484, + "grad_norm": 2.6929407119750977, + "learning_rate": 1e-06, + "loss": 0.3946, + "mean_token_accuracy": 0.8639087677001953, + "num_tokens": 229907103.0, + "step": 5917 + }, + { + "epoch": 0.7528304286986388, + "grad_norm": 2.5556724071502686, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8791136741638184, + "num_tokens": 229945185.0, + "step": 5918 + }, + { + "epoch": 0.7529576389772293, + "grad_norm": 2.530317783355713, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8849904537200928, + "num_tokens": 229983197.0, + "step": 5919 + }, + { + "epoch": 0.7530848492558199, + "grad_norm": 2.915379524230957, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8780770301818848, + "num_tokens": 230014924.0, + "step": 5920 + }, + { + "epoch": 0.7532120595344104, + "grad_norm": 2.56872820854187, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8754205703735352, + "num_tokens": 230053555.0, + "step": 5921 + }, + { + "epoch": 0.7533392698130009, + "grad_norm": 2.546964645385742, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.887961745262146, + "num_tokens": 230090683.0, + "step": 5922 + }, + { + "epoch": 0.7534664800915915, + "grad_norm": 2.8235199451446533, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.8562965393066406, + "num_tokens": 230131982.0, + "step": 5923 + }, + { + "epoch": 0.7535936903701819, + "grad_norm": 2.5533061027526855, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8787676095962524, + "num_tokens": 230175317.0, + "step": 5924 + }, + { + "epoch": 0.7537209006487724, + "grad_norm": 2.6690316200256348, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8818439841270447, + "num_tokens": 230213410.0, + "step": 5925 + }, + { + "epoch": 0.7538481109273629, + "grad_norm": 2.6572105884552, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8780272006988525, + "num_tokens": 230252222.0, + "step": 5926 + }, + { + "epoch": 0.7539753212059535, + "grad_norm": 2.700679063796997, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8727856874465942, + "num_tokens": 230291958.0, + "step": 5927 + }, + { + "epoch": 0.754102531484544, + "grad_norm": 2.618194341659546, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8744431138038635, + "num_tokens": 230328802.0, + "step": 5928 + }, + { + "epoch": 0.7542297417631345, + "grad_norm": 2.5914411544799805, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8784328699111938, + "num_tokens": 230368702.0, + "step": 5929 + }, + { + "epoch": 0.7543569520417249, + "grad_norm": 2.6309378147125244, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.876660168170929, + "num_tokens": 230407812.0, + "step": 5930 + }, + { + "epoch": 0.7544841623203155, + "grad_norm": 2.6438255310058594, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8735326528549194, + "num_tokens": 230447956.0, + "step": 5931 + }, + { + "epoch": 0.754611372598906, + "grad_norm": 2.6609814167022705, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8864339590072632, + "num_tokens": 230485464.0, + "step": 5932 + }, + { + "epoch": 0.7547385828774965, + "grad_norm": 2.687615394592285, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8849083185195923, + "num_tokens": 230519950.0, + "step": 5933 + }, + { + "epoch": 0.754865793156087, + "grad_norm": 2.579772710800171, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8786389827728271, + "num_tokens": 230560586.0, + "step": 5934 + }, + { + "epoch": 0.7549930034346776, + "grad_norm": 2.4960246086120605, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8764308094978333, + "num_tokens": 230603220.0, + "step": 5935 + }, + { + "epoch": 0.755120213713268, + "grad_norm": 2.5904946327209473, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8746554255485535, + "num_tokens": 230647208.0, + "step": 5936 + }, + { + "epoch": 0.7552474239918585, + "grad_norm": 2.6704158782958984, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8733357787132263, + "num_tokens": 230686680.0, + "step": 5937 + }, + { + "epoch": 0.755374634270449, + "grad_norm": 2.6542506217956543, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8851426839828491, + "num_tokens": 230724830.0, + "step": 5938 + }, + { + "epoch": 0.7555018445490396, + "grad_norm": 2.42869234085083, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8795937299728394, + "num_tokens": 230765515.0, + "step": 5939 + }, + { + "epoch": 0.7556290548276301, + "grad_norm": 2.363255023956299, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8839225769042969, + "num_tokens": 230804860.0, + "step": 5940 + }, + { + "epoch": 0.7557562651062206, + "grad_norm": 2.419235944747925, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8792070150375366, + "num_tokens": 230847882.0, + "step": 5941 + }, + { + "epoch": 0.755883475384811, + "grad_norm": 2.621500253677368, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8894928693771362, + "num_tokens": 230883846.0, + "step": 5942 + }, + { + "epoch": 0.7560106856634016, + "grad_norm": 2.6062514781951904, + "learning_rate": 1e-06, + "loss": 0.3987, + "mean_token_accuracy": 0.8644747734069824, + "num_tokens": 230926855.0, + "step": 5943 + }, + { + "epoch": 0.7561378959419921, + "grad_norm": 2.506070852279663, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8714065551757812, + "num_tokens": 230966131.0, + "step": 5944 + }, + { + "epoch": 0.7562651062205826, + "grad_norm": 2.673192262649536, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.8649797439575195, + "num_tokens": 231007266.0, + "step": 5945 + }, + { + "epoch": 0.7563923164991732, + "grad_norm": 2.4968957901000977, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8734952211380005, + "num_tokens": 231049677.0, + "step": 5946 + }, + { + "epoch": 0.7565195267777637, + "grad_norm": 2.681403160095215, + "learning_rate": 1e-06, + "loss": 0.3946, + "mean_token_accuracy": 0.8611940145492554, + "num_tokens": 231089183.0, + "step": 5947 + }, + { + "epoch": 0.7566467370563541, + "grad_norm": 2.438486099243164, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8807141780853271, + "num_tokens": 231130848.0, + "step": 5948 + }, + { + "epoch": 0.7567739473349446, + "grad_norm": 2.4790713787078857, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8804975748062134, + "num_tokens": 231168801.0, + "step": 5949 + }, + { + "epoch": 0.7569011576135352, + "grad_norm": 2.599618911743164, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.879604697227478, + "num_tokens": 231208664.0, + "step": 5950 + }, + { + "epoch": 0.7570283678921257, + "grad_norm": 2.664402961730957, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8730358481407166, + "num_tokens": 231245610.0, + "step": 5951 + }, + { + "epoch": 0.7571555781707162, + "grad_norm": 2.82953143119812, + "learning_rate": 1e-06, + "loss": 0.4009, + "mean_token_accuracy": 0.8655203580856323, + "num_tokens": 231282921.0, + "step": 5952 + }, + { + "epoch": 0.7572827884493067, + "grad_norm": 2.5099945068359375, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.8763664960861206, + "num_tokens": 231323548.0, + "step": 5953 + }, + { + "epoch": 0.7574099987278972, + "grad_norm": 2.907219409942627, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8717645406723022, + "num_tokens": 231355060.0, + "step": 5954 + }, + { + "epoch": 0.7575372090064877, + "grad_norm": 2.5850393772125244, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8759163022041321, + "num_tokens": 231392400.0, + "step": 5955 + }, + { + "epoch": 0.7576644192850782, + "grad_norm": 2.662015199661255, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8813809156417847, + "num_tokens": 231430148.0, + "step": 5956 + }, + { + "epoch": 0.7577916295636687, + "grad_norm": 2.77044939994812, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8726059198379517, + "num_tokens": 231466075.0, + "step": 5957 + }, + { + "epoch": 0.7579188398422593, + "grad_norm": 2.502047538757324, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8804806470870972, + "num_tokens": 231506703.0, + "step": 5958 + }, + { + "epoch": 0.7580460501208498, + "grad_norm": 2.7294833660125732, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8901346325874329, + "num_tokens": 231538949.0, + "step": 5959 + }, + { + "epoch": 0.7581732603994403, + "grad_norm": 2.549022674560547, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8744135499000549, + "num_tokens": 231580183.0, + "step": 5960 + }, + { + "epoch": 0.7583004706780307, + "grad_norm": 2.4403321743011475, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8935763835906982, + "num_tokens": 231617376.0, + "step": 5961 + }, + { + "epoch": 0.7584276809566213, + "grad_norm": 2.3985435962677, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8927690982818604, + "num_tokens": 231659674.0, + "step": 5962 + }, + { + "epoch": 0.7585548912352118, + "grad_norm": 2.37876033782959, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8862751126289368, + "num_tokens": 231702263.0, + "step": 5963 + }, + { + "epoch": 0.7586821015138023, + "grad_norm": 2.520139694213867, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8823907375335693, + "num_tokens": 231741299.0, + "step": 5964 + }, + { + "epoch": 0.7588093117923929, + "grad_norm": 2.6822760105133057, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8882395029067993, + "num_tokens": 231774108.0, + "step": 5965 + }, + { + "epoch": 0.7589365220709834, + "grad_norm": 2.51322865486145, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8957850933074951, + "num_tokens": 231809224.0, + "step": 5966 + }, + { + "epoch": 0.7590637323495738, + "grad_norm": 2.4753589630126953, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8738071322441101, + "num_tokens": 231851364.0, + "step": 5967 + }, + { + "epoch": 0.7591909426281643, + "grad_norm": 2.5236570835113525, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8753169775009155, + "num_tokens": 231889354.0, + "step": 5968 + }, + { + "epoch": 0.7593181529067549, + "grad_norm": 2.472604274749756, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8717430830001831, + "num_tokens": 231931609.0, + "step": 5969 + }, + { + "epoch": 0.7594453631853454, + "grad_norm": 2.6610639095306396, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8755660057067871, + "num_tokens": 231970302.0, + "step": 5970 + }, + { + "epoch": 0.7595725734639359, + "grad_norm": 2.79319429397583, + "learning_rate": 1e-06, + "loss": 0.3928, + "mean_token_accuracy": 0.8636817932128906, + "num_tokens": 232007886.0, + "step": 5971 + }, + { + "epoch": 0.7596997837425264, + "grad_norm": 2.4454574584960938, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8806993961334229, + "num_tokens": 232052161.0, + "step": 5972 + }, + { + "epoch": 0.7598269940211169, + "grad_norm": 2.403292417526245, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8753830194473267, + "num_tokens": 232093685.0, + "step": 5973 + }, + { + "epoch": 0.7599542042997074, + "grad_norm": 2.6297552585601807, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8749328851699829, + "num_tokens": 232131358.0, + "step": 5974 + }, + { + "epoch": 0.7600814145782979, + "grad_norm": 2.410292863845825, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8917403221130371, + "num_tokens": 232169848.0, + "step": 5975 + }, + { + "epoch": 0.7602086248568884, + "grad_norm": 2.175593376159668, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8874490857124329, + "num_tokens": 232219958.0, + "step": 5976 + }, + { + "epoch": 0.760335835135479, + "grad_norm": 2.5937917232513428, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.870470404624939, + "num_tokens": 232260525.0, + "step": 5977 + }, + { + "epoch": 0.7604630454140695, + "grad_norm": 2.5819077491760254, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8742637634277344, + "num_tokens": 232301037.0, + "step": 5978 + }, + { + "epoch": 0.7605902556926599, + "grad_norm": 2.58119797706604, + "learning_rate": 1e-06, + "loss": 0.3876, + "mean_token_accuracy": 0.8652471899986267, + "num_tokens": 232344771.0, + "step": 5979 + }, + { + "epoch": 0.7607174659712505, + "grad_norm": 2.6514229774475098, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8707921504974365, + "num_tokens": 232384168.0, + "step": 5980 + }, + { + "epoch": 0.760844676249841, + "grad_norm": 2.306654930114746, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8777655959129333, + "num_tokens": 232432211.0, + "step": 5981 + }, + { + "epoch": 0.7609718865284315, + "grad_norm": 3.5720982551574707, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8808574676513672, + "num_tokens": 232473605.0, + "step": 5982 + }, + { + "epoch": 0.761099096807022, + "grad_norm": 2.430271863937378, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8774479627609253, + "num_tokens": 232517051.0, + "step": 5983 + }, + { + "epoch": 0.7612263070856126, + "grad_norm": 2.516401529312134, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8722556829452515, + "num_tokens": 232555513.0, + "step": 5984 + }, + { + "epoch": 0.761353517364203, + "grad_norm": 2.663274049758911, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8788844347000122, + "num_tokens": 232589599.0, + "step": 5985 + }, + { + "epoch": 0.7614807276427935, + "grad_norm": 2.992009162902832, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8672523498535156, + "num_tokens": 232623160.0, + "step": 5986 + }, + { + "epoch": 0.761607937921384, + "grad_norm": 2.728348731994629, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.871680736541748, + "num_tokens": 232659241.0, + "step": 5987 + }, + { + "epoch": 0.7617351481999746, + "grad_norm": 2.4836442470550537, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8822660446166992, + "num_tokens": 232698980.0, + "step": 5988 + }, + { + "epoch": 0.7618623584785651, + "grad_norm": 2.312016010284424, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8892108201980591, + "num_tokens": 232740830.0, + "step": 5989 + }, + { + "epoch": 0.7619895687571556, + "grad_norm": 2.788990020751953, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8819398880004883, + "num_tokens": 232773175.0, + "step": 5990 + }, + { + "epoch": 0.762116779035746, + "grad_norm": 2.6248764991760254, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8803645372390747, + "num_tokens": 232810586.0, + "step": 5991 + }, + { + "epoch": 0.7622439893143366, + "grad_norm": 2.87768816947937, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8913713693618774, + "num_tokens": 232839942.0, + "step": 5992 + }, + { + "epoch": 0.7623711995929271, + "grad_norm": 2.6385579109191895, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8685781359672546, + "num_tokens": 232878197.0, + "step": 5993 + }, + { + "epoch": 0.7624984098715176, + "grad_norm": 2.4458725452423096, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8866859674453735, + "num_tokens": 232921635.0, + "step": 5994 + }, + { + "epoch": 0.7626256201501082, + "grad_norm": 2.8509175777435303, + "learning_rate": 1e-06, + "loss": 0.4033, + "mean_token_accuracy": 0.8601593971252441, + "num_tokens": 232959648.0, + "step": 5995 + }, + { + "epoch": 0.7627528304286987, + "grad_norm": 2.364452600479126, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8862239718437195, + "num_tokens": 233002891.0, + "step": 5996 + }, + { + "epoch": 0.7628800407072891, + "grad_norm": 2.4596071243286133, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8679450750350952, + "num_tokens": 233043490.0, + "step": 5997 + }, + { + "epoch": 0.7630072509858796, + "grad_norm": 3.1674306392669678, + "learning_rate": 1e-06, + "loss": 0.3776, + "mean_token_accuracy": 0.869922399520874, + "num_tokens": 233075612.0, + "step": 5998 + }, + { + "epoch": 0.7631344612644702, + "grad_norm": 2.824017286300659, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8729463815689087, + "num_tokens": 233111056.0, + "step": 5999 + }, + { + "epoch": 0.7632616715430607, + "grad_norm": 2.3475632667541504, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8918927907943726, + "num_tokens": 233154687.0, + "step": 6000 + }, + { + "epoch": 0.7633888818216512, + "grad_norm": 2.7792272567749023, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8827427625656128, + "num_tokens": 233189195.0, + "step": 6001 + }, + { + "epoch": 0.7635160921002417, + "grad_norm": 2.4823548793792725, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8947516679763794, + "num_tokens": 233224957.0, + "step": 6002 + }, + { + "epoch": 0.7636433023788322, + "grad_norm": 2.8324527740478516, + "learning_rate": 1e-06, + "loss": 0.4096, + "mean_token_accuracy": 0.8624194264411926, + "num_tokens": 233262236.0, + "step": 6003 + }, + { + "epoch": 0.7637705126574227, + "grad_norm": 2.6930317878723145, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.868087887763977, + "num_tokens": 233302394.0, + "step": 6004 + }, + { + "epoch": 0.7638977229360132, + "grad_norm": 2.5423452854156494, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8911443948745728, + "num_tokens": 233342269.0, + "step": 6005 + }, + { + "epoch": 0.7640249332146037, + "grad_norm": 2.7649896144866943, + "learning_rate": 1e-06, + "loss": 0.4107, + "mean_token_accuracy": 0.859764039516449, + "num_tokens": 233379172.0, + "step": 6006 + }, + { + "epoch": 0.7641521434931943, + "grad_norm": 2.4949114322662354, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.887657105922699, + "num_tokens": 233419068.0, + "step": 6007 + }, + { + "epoch": 0.7642793537717848, + "grad_norm": 2.4760797023773193, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8647805452346802, + "num_tokens": 233465334.0, + "step": 6008 + }, + { + "epoch": 0.7644065640503753, + "grad_norm": 2.5987350940704346, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8887941241264343, + "num_tokens": 233501367.0, + "step": 6009 + }, + { + "epoch": 0.7645337743289657, + "grad_norm": 2.65173602104187, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8904043436050415, + "num_tokens": 233538322.0, + "step": 6010 + }, + { + "epoch": 0.7646609846075563, + "grad_norm": 2.548499345779419, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8736574053764343, + "num_tokens": 233580686.0, + "step": 6011 + }, + { + "epoch": 0.7647881948861468, + "grad_norm": 2.645443916320801, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8886000514030457, + "num_tokens": 233619252.0, + "step": 6012 + }, + { + "epoch": 0.7649154051647373, + "grad_norm": 2.4603817462921143, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8801385164260864, + "num_tokens": 233661194.0, + "step": 6013 + }, + { + "epoch": 0.7650426154433279, + "grad_norm": 2.6729023456573486, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8661783933639526, + "num_tokens": 233702798.0, + "step": 6014 + }, + { + "epoch": 0.7651698257219184, + "grad_norm": 2.6784610748291016, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8689215779304504, + "num_tokens": 233742128.0, + "step": 6015 + }, + { + "epoch": 0.7652970360005088, + "grad_norm": 2.5807254314422607, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8827548027038574, + "num_tokens": 233779346.0, + "step": 6016 + }, + { + "epoch": 0.7654242462790993, + "grad_norm": 2.403582811355591, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8784521222114563, + "num_tokens": 233823909.0, + "step": 6017 + }, + { + "epoch": 0.7655514565576899, + "grad_norm": 2.5813028812408447, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8770159482955933, + "num_tokens": 233866493.0, + "step": 6018 + }, + { + "epoch": 0.7656786668362804, + "grad_norm": 2.5888679027557373, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8839784860610962, + "num_tokens": 233902056.0, + "step": 6019 + }, + { + "epoch": 0.7658058771148709, + "grad_norm": 2.6182639598846436, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8751307129859924, + "num_tokens": 233939230.0, + "step": 6020 + }, + { + "epoch": 0.7659330873934614, + "grad_norm": 2.4919826984405518, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8805736899375916, + "num_tokens": 233979940.0, + "step": 6021 + }, + { + "epoch": 0.7660602976720519, + "grad_norm": 2.46195387840271, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.881592869758606, + "num_tokens": 234024855.0, + "step": 6022 + }, + { + "epoch": 0.7661875079506424, + "grad_norm": 2.515397548675537, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8777603507041931, + "num_tokens": 234066894.0, + "step": 6023 + }, + { + "epoch": 0.7663147182292329, + "grad_norm": 2.5669400691986084, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8773094415664673, + "num_tokens": 234105779.0, + "step": 6024 + }, + { + "epoch": 0.7664419285078234, + "grad_norm": 2.5531203746795654, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8785184621810913, + "num_tokens": 234143534.0, + "step": 6025 + }, + { + "epoch": 0.766569138786414, + "grad_norm": 2.720114231109619, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.873675525188446, + "num_tokens": 234181199.0, + "step": 6026 + }, + { + "epoch": 0.7666963490650045, + "grad_norm": 2.5651583671569824, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8685414791107178, + "num_tokens": 234221882.0, + "step": 6027 + }, + { + "epoch": 0.7668235593435949, + "grad_norm": 2.4804766178131104, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8786922693252563, + "num_tokens": 234268361.0, + "step": 6028 + }, + { + "epoch": 0.7669507696221854, + "grad_norm": 2.420441150665283, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8851580023765564, + "num_tokens": 234309654.0, + "step": 6029 + }, + { + "epoch": 0.767077979900776, + "grad_norm": 2.491912364959717, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8792858123779297, + "num_tokens": 234350115.0, + "step": 6030 + }, + { + "epoch": 0.7672051901793665, + "grad_norm": 2.6432859897613525, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8769254684448242, + "num_tokens": 234389457.0, + "step": 6031 + }, + { + "epoch": 0.767332400457957, + "grad_norm": 2.549246311187744, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8736751079559326, + "num_tokens": 234429534.0, + "step": 6032 + }, + { + "epoch": 0.7674596107365476, + "grad_norm": 2.5603902339935303, + "learning_rate": 1e-06, + "loss": 0.3859, + "mean_token_accuracy": 0.8684277534484863, + "num_tokens": 234470796.0, + "step": 6033 + }, + { + "epoch": 0.767586821015138, + "grad_norm": 2.5389299392700195, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8830797672271729, + "num_tokens": 234507554.0, + "step": 6034 + }, + { + "epoch": 0.7677140312937285, + "grad_norm": 2.598869800567627, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8860551118850708, + "num_tokens": 234543146.0, + "step": 6035 + }, + { + "epoch": 0.767841241572319, + "grad_norm": 2.3723368644714355, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8835178017616272, + "num_tokens": 234588475.0, + "step": 6036 + }, + { + "epoch": 0.7679684518509096, + "grad_norm": 2.7214503288269043, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.8785581588745117, + "num_tokens": 234623472.0, + "step": 6037 + }, + { + "epoch": 0.7680956621295001, + "grad_norm": 2.5412697792053223, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8787757158279419, + "num_tokens": 234664300.0, + "step": 6038 + }, + { + "epoch": 0.7682228724080906, + "grad_norm": 2.991579055786133, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.8604142069816589, + "num_tokens": 234699142.0, + "step": 6039 + }, + { + "epoch": 0.768350082686681, + "grad_norm": 2.4725725650787354, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8812574744224548, + "num_tokens": 234741803.0, + "step": 6040 + }, + { + "epoch": 0.7684772929652716, + "grad_norm": 2.5817277431488037, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.877848744392395, + "num_tokens": 234780348.0, + "step": 6041 + }, + { + "epoch": 0.7686045032438621, + "grad_norm": 2.736004590988159, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8710466623306274, + "num_tokens": 234820318.0, + "step": 6042 + }, + { + "epoch": 0.7687317135224526, + "grad_norm": 2.6404993534088135, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8755244612693787, + "num_tokens": 234856705.0, + "step": 6043 + }, + { + "epoch": 0.7688589238010431, + "grad_norm": 2.6273434162139893, + "learning_rate": 1e-06, + "loss": 0.4104, + "mean_token_accuracy": 0.8590575456619263, + "num_tokens": 234898677.0, + "step": 6044 + }, + { + "epoch": 0.7689861340796337, + "grad_norm": 2.699540853500366, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.87720787525177, + "num_tokens": 234934986.0, + "step": 6045 + }, + { + "epoch": 0.7691133443582241, + "grad_norm": 2.7904763221740723, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.881787121295929, + "num_tokens": 234969403.0, + "step": 6046 + }, + { + "epoch": 0.7692405546368146, + "grad_norm": 3.1903765201568604, + "learning_rate": 1e-06, + "loss": 0.361, + "mean_token_accuracy": 0.8732425570487976, + "num_tokens": 235001728.0, + "step": 6047 + }, + { + "epoch": 0.7693677649154052, + "grad_norm": 2.722529649734497, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8691343665122986, + "num_tokens": 235036853.0, + "step": 6048 + }, + { + "epoch": 0.7694949751939957, + "grad_norm": 2.3470675945281982, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8847814798355103, + "num_tokens": 235080251.0, + "step": 6049 + }, + { + "epoch": 0.7696221854725862, + "grad_norm": 2.6018199920654297, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8754871487617493, + "num_tokens": 235118123.0, + "step": 6050 + }, + { + "epoch": 0.7697493957511767, + "grad_norm": 2.505326509475708, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8793795108795166, + "num_tokens": 235159732.0, + "step": 6051 + }, + { + "epoch": 0.7698766060297672, + "grad_norm": 2.439138174057007, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8762499690055847, + "num_tokens": 235204265.0, + "step": 6052 + }, + { + "epoch": 0.7700038163083577, + "grad_norm": 2.657407522201538, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8763360977172852, + "num_tokens": 235241426.0, + "step": 6053 + }, + { + "epoch": 0.7701310265869482, + "grad_norm": 2.459904909133911, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8836121559143066, + "num_tokens": 235282072.0, + "step": 6054 + }, + { + "epoch": 0.7702582368655387, + "grad_norm": 2.7124693393707275, + "learning_rate": 1e-06, + "loss": 0.3733, + "mean_token_accuracy": 0.870505690574646, + "num_tokens": 235318498.0, + "step": 6055 + }, + { + "epoch": 0.7703854471441293, + "grad_norm": 2.666691303253174, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8802095651626587, + "num_tokens": 235355990.0, + "step": 6056 + }, + { + "epoch": 0.7705126574227198, + "grad_norm": 2.6014974117279053, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8786346912384033, + "num_tokens": 235396543.0, + "step": 6057 + }, + { + "epoch": 0.7706398677013102, + "grad_norm": 2.732180595397949, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8858880400657654, + "num_tokens": 235428604.0, + "step": 6058 + }, + { + "epoch": 0.7707670779799007, + "grad_norm": 2.772690773010254, + "learning_rate": 1e-06, + "loss": 0.3841, + "mean_token_accuracy": 0.8651331663131714, + "num_tokens": 235465092.0, + "step": 6059 + }, + { + "epoch": 0.7708942882584913, + "grad_norm": 2.4892261028289795, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.870551586151123, + "num_tokens": 235508858.0, + "step": 6060 + }, + { + "epoch": 0.7710214985370818, + "grad_norm": 2.5560648441314697, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8714905977249146, + "num_tokens": 235547453.0, + "step": 6061 + }, + { + "epoch": 0.7711487088156723, + "grad_norm": 2.6111559867858887, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.88382887840271, + "num_tokens": 235583466.0, + "step": 6062 + }, + { + "epoch": 0.7712759190942629, + "grad_norm": 2.653855562210083, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8823083639144897, + "num_tokens": 235618896.0, + "step": 6063 + }, + { + "epoch": 0.7714031293728534, + "grad_norm": 2.597999095916748, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8730291128158569, + "num_tokens": 235658796.0, + "step": 6064 + }, + { + "epoch": 0.7715303396514438, + "grad_norm": 2.807908058166504, + "learning_rate": 1e-06, + "loss": 0.4021, + "mean_token_accuracy": 0.8614710569381714, + "num_tokens": 235694751.0, + "step": 6065 + }, + { + "epoch": 0.7716575499300343, + "grad_norm": 2.56968355178833, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8744595050811768, + "num_tokens": 235734103.0, + "step": 6066 + }, + { + "epoch": 0.7717847602086249, + "grad_norm": 2.5244598388671875, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.881848156452179, + "num_tokens": 235775792.0, + "step": 6067 + }, + { + "epoch": 0.7719119704872154, + "grad_norm": 2.7729997634887695, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8788198828697205, + "num_tokens": 235810655.0, + "step": 6068 + }, + { + "epoch": 0.7720391807658059, + "grad_norm": 2.536259651184082, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.873727560043335, + "num_tokens": 235852751.0, + "step": 6069 + }, + { + "epoch": 0.7721663910443964, + "grad_norm": 2.5454905033111572, + "learning_rate": 1e-06, + "loss": 0.3935, + "mean_token_accuracy": 0.8634486198425293, + "num_tokens": 235899072.0, + "step": 6070 + }, + { + "epoch": 0.7722936013229869, + "grad_norm": 2.55430006980896, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8955094814300537, + "num_tokens": 235931781.0, + "step": 6071 + }, + { + "epoch": 0.7724208116015774, + "grad_norm": 2.5960347652435303, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8822693824768066, + "num_tokens": 235969015.0, + "step": 6072 + }, + { + "epoch": 0.7725480218801679, + "grad_norm": 2.4209232330322266, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8832834362983704, + "num_tokens": 236013435.0, + "step": 6073 + }, + { + "epoch": 0.7726752321587584, + "grad_norm": 2.552431583404541, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8835045695304871, + "num_tokens": 236050282.0, + "step": 6074 + }, + { + "epoch": 0.772802442437349, + "grad_norm": 2.452385187149048, + "learning_rate": 1e-06, + "loss": 0.3713, + "mean_token_accuracy": 0.8695383071899414, + "num_tokens": 236094039.0, + "step": 6075 + }, + { + "epoch": 0.7729296527159395, + "grad_norm": 2.6242005825042725, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8831469416618347, + "num_tokens": 236131796.0, + "step": 6076 + }, + { + "epoch": 0.7730568629945299, + "grad_norm": 2.6207058429718018, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8809012770652771, + "num_tokens": 236167094.0, + "step": 6077 + }, + { + "epoch": 0.7731840732731204, + "grad_norm": 2.454387903213501, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8836128115653992, + "num_tokens": 236207767.0, + "step": 6078 + }, + { + "epoch": 0.773311283551711, + "grad_norm": 2.4655330181121826, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.897607684135437, + "num_tokens": 236243971.0, + "step": 6079 + }, + { + "epoch": 0.7734384938303015, + "grad_norm": 2.872065782546997, + "learning_rate": 1e-06, + "loss": 0.4242, + "mean_token_accuracy": 0.8489867448806763, + "num_tokens": 236282718.0, + "step": 6080 + }, + { + "epoch": 0.773565704108892, + "grad_norm": 2.665604591369629, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.865675687789917, + "num_tokens": 236324337.0, + "step": 6081 + }, + { + "epoch": 0.7736929143874826, + "grad_norm": 2.890286445617676, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8861609697341919, + "num_tokens": 236356069.0, + "step": 6082 + }, + { + "epoch": 0.773820124666073, + "grad_norm": 2.5996711254119873, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8841053247451782, + "num_tokens": 236391631.0, + "step": 6083 + }, + { + "epoch": 0.7739473349446635, + "grad_norm": 2.5407867431640625, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8863009214401245, + "num_tokens": 236428721.0, + "step": 6084 + }, + { + "epoch": 0.774074545223254, + "grad_norm": 2.6273152828216553, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8803377747535706, + "num_tokens": 236464017.0, + "step": 6085 + }, + { + "epoch": 0.7742017555018446, + "grad_norm": 2.3563179969787598, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.894121527671814, + "num_tokens": 236505102.0, + "step": 6086 + }, + { + "epoch": 0.7743289657804351, + "grad_norm": 2.682661294937134, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8811644315719604, + "num_tokens": 236541468.0, + "step": 6087 + }, + { + "epoch": 0.7744561760590256, + "grad_norm": 2.79976487159729, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8761450052261353, + "num_tokens": 236578648.0, + "step": 6088 + }, + { + "epoch": 0.774583386337616, + "grad_norm": 2.5895824432373047, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8771123886108398, + "num_tokens": 236616639.0, + "step": 6089 + }, + { + "epoch": 0.7747105966162066, + "grad_norm": 2.806684970855713, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8807863593101501, + "num_tokens": 236649502.0, + "step": 6090 + }, + { + "epoch": 0.7748378068947971, + "grad_norm": 2.711644172668457, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8852665424346924, + "num_tokens": 236681785.0, + "step": 6091 + }, + { + "epoch": 0.7749650171733876, + "grad_norm": 2.4246864318847656, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8836261034011841, + "num_tokens": 236723706.0, + "step": 6092 + }, + { + "epoch": 0.7750922274519781, + "grad_norm": 2.4416611194610596, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8798249959945679, + "num_tokens": 236767113.0, + "step": 6093 + }, + { + "epoch": 0.7752194377305687, + "grad_norm": 2.518289566040039, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8696001768112183, + "num_tokens": 236812321.0, + "step": 6094 + }, + { + "epoch": 0.7753466480091591, + "grad_norm": 2.6336193084716797, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8782699704170227, + "num_tokens": 236846716.0, + "step": 6095 + }, + { + "epoch": 0.7754738582877496, + "grad_norm": 2.6633026599884033, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8735848665237427, + "num_tokens": 236883174.0, + "step": 6096 + }, + { + "epoch": 0.7756010685663401, + "grad_norm": 2.613898992538452, + "learning_rate": 1e-06, + "loss": 0.3843, + "mean_token_accuracy": 0.8656575083732605, + "num_tokens": 236927367.0, + "step": 6097 + }, + { + "epoch": 0.7757282788449307, + "grad_norm": 2.7489173412323, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8756625652313232, + "num_tokens": 236963260.0, + "step": 6098 + }, + { + "epoch": 0.7758554891235212, + "grad_norm": 2.3767495155334473, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.890219509601593, + "num_tokens": 237005450.0, + "step": 6099 + }, + { + "epoch": 0.7759826994021117, + "grad_norm": 2.559318780899048, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8810436129570007, + "num_tokens": 237043437.0, + "step": 6100 + }, + { + "epoch": 0.7761099096807021, + "grad_norm": 2.57291316986084, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8691131472587585, + "num_tokens": 237083994.0, + "step": 6101 + }, + { + "epoch": 0.7762371199592927, + "grad_norm": 2.4551382064819336, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8773292899131775, + "num_tokens": 237128633.0, + "step": 6102 + }, + { + "epoch": 0.7763643302378832, + "grad_norm": 2.5111773014068604, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8737847208976746, + "num_tokens": 237167827.0, + "step": 6103 + }, + { + "epoch": 0.7764915405164737, + "grad_norm": 2.336406946182251, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8795893788337708, + "num_tokens": 237210933.0, + "step": 6104 + }, + { + "epoch": 0.7766187507950643, + "grad_norm": 2.6761491298675537, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.8690176010131836, + "num_tokens": 237246984.0, + "step": 6105 + }, + { + "epoch": 0.7767459610736548, + "grad_norm": 2.660114288330078, + "learning_rate": 1e-06, + "loss": 0.3976, + "mean_token_accuracy": 0.872180163860321, + "num_tokens": 237289563.0, + "step": 6106 + }, + { + "epoch": 0.7768731713522452, + "grad_norm": 2.4186575412750244, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8842009902000427, + "num_tokens": 237332891.0, + "step": 6107 + }, + { + "epoch": 0.7770003816308357, + "grad_norm": 2.768822193145752, + "learning_rate": 1e-06, + "loss": 0.4225, + "mean_token_accuracy": 0.8496935963630676, + "num_tokens": 237373195.0, + "step": 6108 + }, + { + "epoch": 0.7771275919094263, + "grad_norm": 2.6094110012054443, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8792442679405212, + "num_tokens": 237409420.0, + "step": 6109 + }, + { + "epoch": 0.7772548021880168, + "grad_norm": 2.5581812858581543, + "learning_rate": 1e-06, + "loss": 0.3738, + "mean_token_accuracy": 0.8684284090995789, + "num_tokens": 237449529.0, + "step": 6110 + }, + { + "epoch": 0.7773820124666073, + "grad_norm": 2.7984602451324463, + "learning_rate": 1e-06, + "loss": 0.3757, + "mean_token_accuracy": 0.8735889196395874, + "num_tokens": 237486520.0, + "step": 6111 + }, + { + "epoch": 0.7775092227451978, + "grad_norm": 2.866973400115967, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.867514967918396, + "num_tokens": 237524966.0, + "step": 6112 + }, + { + "epoch": 0.7776364330237884, + "grad_norm": 2.7245728969573975, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8854848146438599, + "num_tokens": 237559586.0, + "step": 6113 + }, + { + "epoch": 0.7777636433023788, + "grad_norm": 2.3250579833984375, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8926865458488464, + "num_tokens": 237600369.0, + "step": 6114 + }, + { + "epoch": 0.7778908535809693, + "grad_norm": 2.706475019454956, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8859890699386597, + "num_tokens": 237636680.0, + "step": 6115 + }, + { + "epoch": 0.7780180638595598, + "grad_norm": 2.722883939743042, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8787656426429749, + "num_tokens": 237672704.0, + "step": 6116 + }, + { + "epoch": 0.7781452741381504, + "grad_norm": 2.6358540058135986, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8735040426254272, + "num_tokens": 237712594.0, + "step": 6117 + }, + { + "epoch": 0.7782724844167409, + "grad_norm": 2.5509138107299805, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8918887972831726, + "num_tokens": 237749852.0, + "step": 6118 + }, + { + "epoch": 0.7783996946953314, + "grad_norm": 2.7075865268707275, + "learning_rate": 1e-06, + "loss": 0.3825, + "mean_token_accuracy": 0.8726422786712646, + "num_tokens": 237786809.0, + "step": 6119 + }, + { + "epoch": 0.7785269049739219, + "grad_norm": 2.7044589519500732, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.873511016368866, + "num_tokens": 237825321.0, + "step": 6120 + }, + { + "epoch": 0.7786541152525124, + "grad_norm": 2.6812331676483154, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8724178075790405, + "num_tokens": 237858475.0, + "step": 6121 + }, + { + "epoch": 0.7787813255311029, + "grad_norm": 2.7412965297698975, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8823177218437195, + "num_tokens": 237893653.0, + "step": 6122 + }, + { + "epoch": 0.7789085358096934, + "grad_norm": 2.416351318359375, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8889220356941223, + "num_tokens": 237932377.0, + "step": 6123 + }, + { + "epoch": 0.779035746088284, + "grad_norm": 2.6193532943725586, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8710153698921204, + "num_tokens": 237967941.0, + "step": 6124 + }, + { + "epoch": 0.7791629563668745, + "grad_norm": 2.732588768005371, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.866566002368927, + "num_tokens": 238005574.0, + "step": 6125 + }, + { + "epoch": 0.7792901666454649, + "grad_norm": 2.541024923324585, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8721716403961182, + "num_tokens": 238043990.0, + "step": 6126 + }, + { + "epoch": 0.7794173769240554, + "grad_norm": 2.5906572341918945, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.876966118812561, + "num_tokens": 238082779.0, + "step": 6127 + }, + { + "epoch": 0.779544587202646, + "grad_norm": 2.5314090251922607, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8924682140350342, + "num_tokens": 238117607.0, + "step": 6128 + }, + { + "epoch": 0.7796717974812365, + "grad_norm": 5.492736339569092, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8766472935676575, + "num_tokens": 238155521.0, + "step": 6129 + }, + { + "epoch": 0.779799007759827, + "grad_norm": 2.4976871013641357, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8785947561264038, + "num_tokens": 238197635.0, + "step": 6130 + }, + { + "epoch": 0.7799262180384176, + "grad_norm": 2.548020362854004, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8828068375587463, + "num_tokens": 238235295.0, + "step": 6131 + }, + { + "epoch": 0.780053428317008, + "grad_norm": 2.700456142425537, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8800462484359741, + "num_tokens": 238273632.0, + "step": 6132 + }, + { + "epoch": 0.7801806385955985, + "grad_norm": 2.8304378986358643, + "learning_rate": 1e-06, + "loss": 0.3869, + "mean_token_accuracy": 0.865410566329956, + "num_tokens": 238311614.0, + "step": 6133 + }, + { + "epoch": 0.780307848874189, + "grad_norm": 3.1058478355407715, + "learning_rate": 1e-06, + "loss": 0.4212, + "mean_token_accuracy": 0.855884313583374, + "num_tokens": 238344654.0, + "step": 6134 + }, + { + "epoch": 0.7804350591527796, + "grad_norm": 2.4281153678894043, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8756239414215088, + "num_tokens": 238390108.0, + "step": 6135 + }, + { + "epoch": 0.7805622694313701, + "grad_norm": 2.840559720993042, + "learning_rate": 1e-06, + "loss": 0.3642, + "mean_token_accuracy": 0.8726779222488403, + "num_tokens": 238424036.0, + "step": 6136 + }, + { + "epoch": 0.7806894797099606, + "grad_norm": 2.523707866668701, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8804133534431458, + "num_tokens": 238461615.0, + "step": 6137 + }, + { + "epoch": 0.780816689988551, + "grad_norm": 2.6389832496643066, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8681294322013855, + "num_tokens": 238499665.0, + "step": 6138 + }, + { + "epoch": 0.7809439002671416, + "grad_norm": 2.507114887237549, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8721635341644287, + "num_tokens": 238542034.0, + "step": 6139 + }, + { + "epoch": 0.7810711105457321, + "grad_norm": 2.4266910552978516, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8865181803703308, + "num_tokens": 238583121.0, + "step": 6140 + }, + { + "epoch": 0.7811983208243226, + "grad_norm": 2.5177621841430664, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8875175714492798, + "num_tokens": 238622220.0, + "step": 6141 + }, + { + "epoch": 0.7813255311029131, + "grad_norm": 2.499962091445923, + "learning_rate": 1e-06, + "loss": 0.3857, + "mean_token_accuracy": 0.8640819787979126, + "num_tokens": 238664408.0, + "step": 6142 + }, + { + "epoch": 0.7814527413815037, + "grad_norm": 2.456212282180786, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.886501133441925, + "num_tokens": 238703064.0, + "step": 6143 + }, + { + "epoch": 0.7815799516600941, + "grad_norm": 2.55906081199646, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8855735063552856, + "num_tokens": 238741937.0, + "step": 6144 + }, + { + "epoch": 0.7817071619386846, + "grad_norm": 2.675337791442871, + "learning_rate": 1e-06, + "loss": 0.3663, + "mean_token_accuracy": 0.8729161024093628, + "num_tokens": 238781904.0, + "step": 6145 + }, + { + "epoch": 0.7818343722172751, + "grad_norm": 2.5186007022857666, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8816196918487549, + "num_tokens": 238822182.0, + "step": 6146 + }, + { + "epoch": 0.7819615824958657, + "grad_norm": 2.975991725921631, + "learning_rate": 1e-06, + "loss": 0.4355, + "mean_token_accuracy": 0.8516008257865906, + "num_tokens": 238860195.0, + "step": 6147 + }, + { + "epoch": 0.7820887927744562, + "grad_norm": 2.5498533248901367, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8785043954849243, + "num_tokens": 238897865.0, + "step": 6148 + }, + { + "epoch": 0.7822160030530467, + "grad_norm": 2.4906558990478516, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8686284422874451, + "num_tokens": 238942572.0, + "step": 6149 + }, + { + "epoch": 0.7823432133316371, + "grad_norm": 2.604494333267212, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8732494115829468, + "num_tokens": 238980113.0, + "step": 6150 + }, + { + "epoch": 0.7824704236102277, + "grad_norm": 2.656437635421753, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8710035681724548, + "num_tokens": 239018611.0, + "step": 6151 + }, + { + "epoch": 0.7825976338888182, + "grad_norm": 2.4451706409454346, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8886144161224365, + "num_tokens": 239054988.0, + "step": 6152 + }, + { + "epoch": 0.7827248441674087, + "grad_norm": 2.481083869934082, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8751080632209778, + "num_tokens": 239098235.0, + "step": 6153 + }, + { + "epoch": 0.7828520544459993, + "grad_norm": 2.414052724838257, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.890813946723938, + "num_tokens": 239138405.0, + "step": 6154 + }, + { + "epoch": 0.7829792647245898, + "grad_norm": 2.5237107276916504, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.872008204460144, + "num_tokens": 239179596.0, + "step": 6155 + }, + { + "epoch": 0.7831064750031802, + "grad_norm": 2.4882640838623047, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8743359446525574, + "num_tokens": 239221726.0, + "step": 6156 + }, + { + "epoch": 0.7832336852817707, + "grad_norm": 2.340048313140869, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8917462229728699, + "num_tokens": 239261646.0, + "step": 6157 + }, + { + "epoch": 0.7833608955603613, + "grad_norm": 2.741419553756714, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8838340640068054, + "num_tokens": 239294214.0, + "step": 6158 + }, + { + "epoch": 0.7834881058389518, + "grad_norm": 2.618825912475586, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.8736295104026794, + "num_tokens": 239334042.0, + "step": 6159 + }, + { + "epoch": 0.7836153161175423, + "grad_norm": 2.640486001968384, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8832533359527588, + "num_tokens": 239372617.0, + "step": 6160 + }, + { + "epoch": 0.7837425263961328, + "grad_norm": 2.5199837684631348, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8868924379348755, + "num_tokens": 239411194.0, + "step": 6161 + }, + { + "epoch": 0.7838697366747234, + "grad_norm": 2.7212843894958496, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8804338574409485, + "num_tokens": 239445466.0, + "step": 6162 + }, + { + "epoch": 0.7839969469533138, + "grad_norm": 2.653315544128418, + "learning_rate": 1e-06, + "loss": 0.4061, + "mean_token_accuracy": 0.8674525022506714, + "num_tokens": 239484373.0, + "step": 6163 + }, + { + "epoch": 0.7841241572319043, + "grad_norm": 2.6183016300201416, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8799782991409302, + "num_tokens": 239522661.0, + "step": 6164 + }, + { + "epoch": 0.7842513675104948, + "grad_norm": 3.0041942596435547, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8743041753768921, + "num_tokens": 239556874.0, + "step": 6165 + }, + { + "epoch": 0.7843785777890854, + "grad_norm": 2.5824058055877686, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8736338019371033, + "num_tokens": 239595913.0, + "step": 6166 + }, + { + "epoch": 0.7845057880676759, + "grad_norm": 2.6113202571868896, + "learning_rate": 1e-06, + "loss": 0.372, + "mean_token_accuracy": 0.8679229617118835, + "num_tokens": 239637168.0, + "step": 6167 + }, + { + "epoch": 0.7846329983462664, + "grad_norm": 2.441210985183716, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8887122869491577, + "num_tokens": 239678555.0, + "step": 6168 + }, + { + "epoch": 0.7847602086248568, + "grad_norm": 2.560824155807495, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8920406103134155, + "num_tokens": 239717703.0, + "step": 6169 + }, + { + "epoch": 0.7848874189034474, + "grad_norm": 2.531440019607544, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8763547539710999, + "num_tokens": 239757915.0, + "step": 6170 + }, + { + "epoch": 0.7850146291820379, + "grad_norm": 2.4926931858062744, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8727408647537231, + "num_tokens": 239800411.0, + "step": 6171 + }, + { + "epoch": 0.7851418394606284, + "grad_norm": 2.7752819061279297, + "learning_rate": 1e-06, + "loss": 0.4202, + "mean_token_accuracy": 0.8524726629257202, + "num_tokens": 239842417.0, + "step": 6172 + }, + { + "epoch": 0.785269049739219, + "grad_norm": 2.675894260406494, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8996670842170715, + "num_tokens": 239874205.0, + "step": 6173 + }, + { + "epoch": 0.7853962600178095, + "grad_norm": 2.5018978118896484, + "learning_rate": 1e-06, + "loss": 0.3964, + "mean_token_accuracy": 0.8627966642379761, + "num_tokens": 239920179.0, + "step": 6174 + }, + { + "epoch": 0.7855234702963999, + "grad_norm": 2.450071334838867, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8790453672409058, + "num_tokens": 239963132.0, + "step": 6175 + }, + { + "epoch": 0.7856506805749904, + "grad_norm": 2.3659915924072266, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8730707168579102, + "num_tokens": 240007796.0, + "step": 6176 + }, + { + "epoch": 0.785777890853581, + "grad_norm": 2.923276424407959, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8790738582611084, + "num_tokens": 240040496.0, + "step": 6177 + }, + { + "epoch": 0.7859051011321715, + "grad_norm": 2.690599203109741, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8807110786437988, + "num_tokens": 240074928.0, + "step": 6178 + }, + { + "epoch": 0.786032311410762, + "grad_norm": 2.6843409538269043, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.875872015953064, + "num_tokens": 240113483.0, + "step": 6179 + }, + { + "epoch": 0.7861595216893525, + "grad_norm": 2.414947271347046, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8860812187194824, + "num_tokens": 240154669.0, + "step": 6180 + }, + { + "epoch": 0.786286731967943, + "grad_norm": 2.7232580184936523, + "learning_rate": 1e-06, + "loss": 0.4008, + "mean_token_accuracy": 0.8653039932250977, + "num_tokens": 240194498.0, + "step": 6181 + }, + { + "epoch": 0.7864139422465335, + "grad_norm": 2.6656494140625, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8821972012519836, + "num_tokens": 240227658.0, + "step": 6182 + }, + { + "epoch": 0.786541152525124, + "grad_norm": 2.978383779525757, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8700781464576721, + "num_tokens": 240263506.0, + "step": 6183 + }, + { + "epoch": 0.7866683628037145, + "grad_norm": 2.4510369300842285, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.867214560508728, + "num_tokens": 240308642.0, + "step": 6184 + }, + { + "epoch": 0.7867955730823051, + "grad_norm": 3.011950969696045, + "learning_rate": 1e-06, + "loss": 0.3666, + "mean_token_accuracy": 0.869321346282959, + "num_tokens": 240340796.0, + "step": 6185 + }, + { + "epoch": 0.7869227833608956, + "grad_norm": 2.8001646995544434, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8811823725700378, + "num_tokens": 240373617.0, + "step": 6186 + }, + { + "epoch": 0.787049993639486, + "grad_norm": 2.4747915267944336, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.873305082321167, + "num_tokens": 240413769.0, + "step": 6187 + }, + { + "epoch": 0.7871772039180766, + "grad_norm": 2.4965319633483887, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8913264274597168, + "num_tokens": 240453164.0, + "step": 6188 + }, + { + "epoch": 0.7873044141966671, + "grad_norm": 2.6337413787841797, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8770562410354614, + "num_tokens": 240490725.0, + "step": 6189 + }, + { + "epoch": 0.7874316244752576, + "grad_norm": 2.5985517501831055, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8726725578308105, + "num_tokens": 240528414.0, + "step": 6190 + }, + { + "epoch": 0.7875588347538481, + "grad_norm": 2.8482179641723633, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8881748914718628, + "num_tokens": 240564765.0, + "step": 6191 + }, + { + "epoch": 0.7876860450324387, + "grad_norm": 2.8741955757141113, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.878541886806488, + "num_tokens": 240595530.0, + "step": 6192 + }, + { + "epoch": 0.7878132553110291, + "grad_norm": 2.7942545413970947, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8825553059577942, + "num_tokens": 240632619.0, + "step": 6193 + }, + { + "epoch": 0.7879404655896196, + "grad_norm": 2.5742645263671875, + "learning_rate": 1e-06, + "loss": 0.3714, + "mean_token_accuracy": 0.8695352077484131, + "num_tokens": 240671025.0, + "step": 6194 + }, + { + "epoch": 0.7880676758682101, + "grad_norm": 2.555699110031128, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8763227462768555, + "num_tokens": 240710365.0, + "step": 6195 + }, + { + "epoch": 0.7881948861468007, + "grad_norm": 2.5217809677124023, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.883091926574707, + "num_tokens": 240748357.0, + "step": 6196 + }, + { + "epoch": 0.7883220964253912, + "grad_norm": 2.6235809326171875, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8702191114425659, + "num_tokens": 240789894.0, + "step": 6197 + }, + { + "epoch": 0.7884493067039817, + "grad_norm": 2.5831377506256104, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8819496631622314, + "num_tokens": 240827971.0, + "step": 6198 + }, + { + "epoch": 0.7885765169825721, + "grad_norm": 2.7212278842926025, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8707772493362427, + "num_tokens": 240870994.0, + "step": 6199 + }, + { + "epoch": 0.7887037272611627, + "grad_norm": 2.3293650150299072, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8868516683578491, + "num_tokens": 240912200.0, + "step": 6200 + }, + { + "epoch": 0.7888309375397532, + "grad_norm": 2.6326262950897217, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8780194520950317, + "num_tokens": 240951716.0, + "step": 6201 + }, + { + "epoch": 0.7889581478183437, + "grad_norm": 2.3837344646453857, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8871161937713623, + "num_tokens": 240993508.0, + "step": 6202 + }, + { + "epoch": 0.7890853580969343, + "grad_norm": 2.491441488265991, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8689454793930054, + "num_tokens": 241033500.0, + "step": 6203 + }, + { + "epoch": 0.7892125683755248, + "grad_norm": 2.301056146621704, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8938727974891663, + "num_tokens": 241074227.0, + "step": 6204 + }, + { + "epoch": 0.7893397786541152, + "grad_norm": 2.517578125, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.882286787033081, + "num_tokens": 241114599.0, + "step": 6205 + }, + { + "epoch": 0.7894669889327057, + "grad_norm": 2.690486192703247, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8858333826065063, + "num_tokens": 241150147.0, + "step": 6206 + }, + { + "epoch": 0.7895941992112963, + "grad_norm": 2.3896992206573486, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8771119117736816, + "num_tokens": 241194004.0, + "step": 6207 + }, + { + "epoch": 0.7897214094898868, + "grad_norm": 2.4974515438079834, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8910268545150757, + "num_tokens": 241228937.0, + "step": 6208 + }, + { + "epoch": 0.7898486197684773, + "grad_norm": 2.8766043186187744, + "learning_rate": 1e-06, + "loss": 0.397, + "mean_token_accuracy": 0.8632767200469971, + "num_tokens": 241264157.0, + "step": 6209 + }, + { + "epoch": 0.7899758300470678, + "grad_norm": 2.927617073059082, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8859518766403198, + "num_tokens": 241294730.0, + "step": 6210 + }, + { + "epoch": 0.7901030403256584, + "grad_norm": 2.3457632064819336, + "learning_rate": 1e-06, + "loss": 0.3635, + "mean_token_accuracy": 0.8754305243492126, + "num_tokens": 241342434.0, + "step": 6211 + }, + { + "epoch": 0.7902302506042488, + "grad_norm": 2.3811910152435303, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8913263082504272, + "num_tokens": 241380703.0, + "step": 6212 + }, + { + "epoch": 0.7903574608828393, + "grad_norm": 2.79880428314209, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8816525936126709, + "num_tokens": 241415864.0, + "step": 6213 + }, + { + "epoch": 0.7904846711614298, + "grad_norm": 2.86728835105896, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8868050575256348, + "num_tokens": 241454134.0, + "step": 6214 + }, + { + "epoch": 0.7906118814400204, + "grad_norm": 3.0058562755584717, + "learning_rate": 1e-06, + "loss": 0.3747, + "mean_token_accuracy": 0.8726670742034912, + "num_tokens": 241485140.0, + "step": 6215 + }, + { + "epoch": 0.7907390917186109, + "grad_norm": 2.325681686401367, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8874104619026184, + "num_tokens": 241528617.0, + "step": 6216 + }, + { + "epoch": 0.7908663019972014, + "grad_norm": 2.5048177242279053, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.888971209526062, + "num_tokens": 241565808.0, + "step": 6217 + }, + { + "epoch": 0.7909935122757918, + "grad_norm": 2.344843864440918, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8810337781906128, + "num_tokens": 241609595.0, + "step": 6218 + }, + { + "epoch": 0.7911207225543824, + "grad_norm": 2.5726404190063477, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8679673671722412, + "num_tokens": 241647972.0, + "step": 6219 + }, + { + "epoch": 0.7912479328329729, + "grad_norm": 2.64961838722229, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8728436231613159, + "num_tokens": 241685351.0, + "step": 6220 + }, + { + "epoch": 0.7913751431115634, + "grad_norm": 2.6140871047973633, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8793779611587524, + "num_tokens": 241726841.0, + "step": 6221 + }, + { + "epoch": 0.791502353390154, + "grad_norm": 2.412235975265503, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8825427293777466, + "num_tokens": 241767924.0, + "step": 6222 + }, + { + "epoch": 0.7916295636687445, + "grad_norm": 2.21008563041687, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8886759281158447, + "num_tokens": 241812858.0, + "step": 6223 + }, + { + "epoch": 0.7917567739473349, + "grad_norm": 2.5430190563201904, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8742659091949463, + "num_tokens": 241852041.0, + "step": 6224 + }, + { + "epoch": 0.7918839842259254, + "grad_norm": 2.5875399112701416, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8878493905067444, + "num_tokens": 241887559.0, + "step": 6225 + }, + { + "epoch": 0.792011194504516, + "grad_norm": 2.652121067047119, + "learning_rate": 1e-06, + "loss": 0.3764, + "mean_token_accuracy": 0.8677232265472412, + "num_tokens": 241927126.0, + "step": 6226 + }, + { + "epoch": 0.7921384047831065, + "grad_norm": 2.6911532878875732, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8774427771568298, + "num_tokens": 241964995.0, + "step": 6227 + }, + { + "epoch": 0.792265615061697, + "grad_norm": 2.585967540740967, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8852213621139526, + "num_tokens": 242002390.0, + "step": 6228 + }, + { + "epoch": 0.7923928253402875, + "grad_norm": 2.2967801094055176, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8946400880813599, + "num_tokens": 242043940.0, + "step": 6229 + }, + { + "epoch": 0.792520035618878, + "grad_norm": 2.6122689247131348, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8872811794281006, + "num_tokens": 242081107.0, + "step": 6230 + }, + { + "epoch": 0.7926472458974685, + "grad_norm": 2.628997564315796, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8835675716400146, + "num_tokens": 242117647.0, + "step": 6231 + }, + { + "epoch": 0.792774456176059, + "grad_norm": 2.3807592391967773, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8705686330795288, + "num_tokens": 242165948.0, + "step": 6232 + }, + { + "epoch": 0.7929016664546495, + "grad_norm": 2.5337636470794678, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.874980628490448, + "num_tokens": 242207341.0, + "step": 6233 + }, + { + "epoch": 0.7930288767332401, + "grad_norm": 3.011791706085205, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8859571218490601, + "num_tokens": 242234494.0, + "step": 6234 + }, + { + "epoch": 0.7931560870118306, + "grad_norm": 2.462374687194824, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8854560256004333, + "num_tokens": 242275774.0, + "step": 6235 + }, + { + "epoch": 0.793283297290421, + "grad_norm": 2.560220956802368, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8739867806434631, + "num_tokens": 242320870.0, + "step": 6236 + }, + { + "epoch": 0.7934105075690115, + "grad_norm": 2.6252124309539795, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8776285648345947, + "num_tokens": 242358665.0, + "step": 6237 + }, + { + "epoch": 0.7935377178476021, + "grad_norm": 2.6408050060272217, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8825868368148804, + "num_tokens": 242396016.0, + "step": 6238 + }, + { + "epoch": 0.7936649281261926, + "grad_norm": 2.3599069118499756, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8890312910079956, + "num_tokens": 242436468.0, + "step": 6239 + }, + { + "epoch": 0.7937921384047831, + "grad_norm": 2.6589348316192627, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8916611671447754, + "num_tokens": 242469885.0, + "step": 6240 + }, + { + "epoch": 0.7939193486833737, + "grad_norm": 2.643242597579956, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8865112066268921, + "num_tokens": 242504277.0, + "step": 6241 + }, + { + "epoch": 0.7940465589619641, + "grad_norm": 2.7876412868499756, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8735669255256653, + "num_tokens": 242542418.0, + "step": 6242 + }, + { + "epoch": 0.7941737692405546, + "grad_norm": 2.4043118953704834, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8758087158203125, + "num_tokens": 242587834.0, + "step": 6243 + }, + { + "epoch": 0.7943009795191451, + "grad_norm": 2.640716791152954, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8846442699432373, + "num_tokens": 242626152.0, + "step": 6244 + }, + { + "epoch": 0.7944281897977357, + "grad_norm": 2.666558027267456, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8820014595985413, + "num_tokens": 242663331.0, + "step": 6245 + }, + { + "epoch": 0.7945554000763262, + "grad_norm": 2.6192290782928467, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8867484331130981, + "num_tokens": 242698912.0, + "step": 6246 + }, + { + "epoch": 0.7946826103549167, + "grad_norm": 2.8555185794830322, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8769891262054443, + "num_tokens": 242733601.0, + "step": 6247 + }, + { + "epoch": 0.7948098206335071, + "grad_norm": 2.4052915573120117, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8826799392700195, + "num_tokens": 242776418.0, + "step": 6248 + }, + { + "epoch": 0.7949370309120977, + "grad_norm": 2.386049747467041, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8891839385032654, + "num_tokens": 242818871.0, + "step": 6249 + }, + { + "epoch": 0.7950642411906882, + "grad_norm": 2.4735279083251953, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8805117011070251, + "num_tokens": 242861707.0, + "step": 6250 + }, + { + "epoch": 0.7951914514692787, + "grad_norm": 2.7690484523773193, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8849667906761169, + "num_tokens": 242892047.0, + "step": 6251 + }, + { + "epoch": 0.7953186617478692, + "grad_norm": 2.4506123065948486, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.8782863616943359, + "num_tokens": 242940162.0, + "step": 6252 + }, + { + "epoch": 0.7954458720264598, + "grad_norm": 2.416339635848999, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8870174288749695, + "num_tokens": 242981913.0, + "step": 6253 + }, + { + "epoch": 0.7955730823050502, + "grad_norm": 2.6584229469299316, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8763234615325928, + "num_tokens": 243017830.0, + "step": 6254 + }, + { + "epoch": 0.7957002925836407, + "grad_norm": 2.5283396244049072, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.879927396774292, + "num_tokens": 243060384.0, + "step": 6255 + }, + { + "epoch": 0.7958275028622313, + "grad_norm": 2.4817583560943604, + "learning_rate": 1e-06, + "loss": 0.3694, + "mean_token_accuracy": 0.871992290019989, + "num_tokens": 243108749.0, + "step": 6256 + }, + { + "epoch": 0.7959547131408218, + "grad_norm": 2.737187385559082, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8746651411056519, + "num_tokens": 243148919.0, + "step": 6257 + }, + { + "epoch": 0.7960819234194123, + "grad_norm": 2.946768045425415, + "learning_rate": 1e-06, + "loss": 0.396, + "mean_token_accuracy": 0.8659989237785339, + "num_tokens": 243183102.0, + "step": 6258 + }, + { + "epoch": 0.7962091336980028, + "grad_norm": 2.603757858276367, + "learning_rate": 1e-06, + "loss": 0.4066, + "mean_token_accuracy": 0.8574531078338623, + "num_tokens": 243228284.0, + "step": 6259 + }, + { + "epoch": 0.7963363439765934, + "grad_norm": 2.8136441707611084, + "learning_rate": 1e-06, + "loss": 0.389, + "mean_token_accuracy": 0.8706730604171753, + "num_tokens": 243263227.0, + "step": 6260 + }, + { + "epoch": 0.7964635542551838, + "grad_norm": 2.517819881439209, + "learning_rate": 1e-06, + "loss": 0.3772, + "mean_token_accuracy": 0.8702875375747681, + "num_tokens": 243307624.0, + "step": 6261 + }, + { + "epoch": 0.7965907645337743, + "grad_norm": 2.701573610305786, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8892937302589417, + "num_tokens": 243340080.0, + "step": 6262 + }, + { + "epoch": 0.7967179748123648, + "grad_norm": 2.421276807785034, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8770148158073425, + "num_tokens": 243384563.0, + "step": 6263 + }, + { + "epoch": 0.7968451850909554, + "grad_norm": 2.518131971359253, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8770895004272461, + "num_tokens": 243422822.0, + "step": 6264 + }, + { + "epoch": 0.7969723953695459, + "grad_norm": 2.672297239303589, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.873714804649353, + "num_tokens": 243463681.0, + "step": 6265 + }, + { + "epoch": 0.7970996056481364, + "grad_norm": 2.3952479362487793, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.88274085521698, + "num_tokens": 243506182.0, + "step": 6266 + }, + { + "epoch": 0.7972268159267268, + "grad_norm": 2.4265904426574707, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8832031488418579, + "num_tokens": 243547498.0, + "step": 6267 + }, + { + "epoch": 0.7973540262053174, + "grad_norm": 2.736901044845581, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8897193074226379, + "num_tokens": 243580272.0, + "step": 6268 + }, + { + "epoch": 0.7974812364839079, + "grad_norm": 2.6533961296081543, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8686957955360413, + "num_tokens": 243620253.0, + "step": 6269 + }, + { + "epoch": 0.7976084467624984, + "grad_norm": 2.7621452808380127, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8641983866691589, + "num_tokens": 243657431.0, + "step": 6270 + }, + { + "epoch": 0.797735657041089, + "grad_norm": 2.464763641357422, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8781575560569763, + "num_tokens": 243701591.0, + "step": 6271 + }, + { + "epoch": 0.7978628673196795, + "grad_norm": 2.483809232711792, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8823525905609131, + "num_tokens": 243736495.0, + "step": 6272 + }, + { + "epoch": 0.7979900775982699, + "grad_norm": 2.6516780853271484, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8805139064788818, + "num_tokens": 243770375.0, + "step": 6273 + }, + { + "epoch": 0.7981172878768604, + "grad_norm": 3.040614366531372, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8767366409301758, + "num_tokens": 243799704.0, + "step": 6274 + }, + { + "epoch": 0.798244498155451, + "grad_norm": 2.3640928268432617, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8877276182174683, + "num_tokens": 243842146.0, + "step": 6275 + }, + { + "epoch": 0.7983717084340415, + "grad_norm": 2.615849018096924, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8797625303268433, + "num_tokens": 243878985.0, + "step": 6276 + }, + { + "epoch": 0.798498918712632, + "grad_norm": 2.635124921798706, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.886715292930603, + "num_tokens": 243912878.0, + "step": 6277 + }, + { + "epoch": 0.7986261289912225, + "grad_norm": 2.485811710357666, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8825881481170654, + "num_tokens": 243954187.0, + "step": 6278 + }, + { + "epoch": 0.798753339269813, + "grad_norm": 2.645960569381714, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.870957612991333, + "num_tokens": 243993476.0, + "step": 6279 + }, + { + "epoch": 0.7988805495484035, + "grad_norm": 2.797377824783325, + "learning_rate": 1e-06, + "loss": 0.3848, + "mean_token_accuracy": 0.8701293468475342, + "num_tokens": 244032838.0, + "step": 6280 + }, + { + "epoch": 0.799007759826994, + "grad_norm": 2.426532030105591, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8932675123214722, + "num_tokens": 244071070.0, + "step": 6281 + }, + { + "epoch": 0.7991349701055845, + "grad_norm": 2.611102819442749, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8777832984924316, + "num_tokens": 244109455.0, + "step": 6282 + }, + { + "epoch": 0.7992621803841751, + "grad_norm": 2.6278457641601562, + "learning_rate": 1e-06, + "loss": 0.4046, + "mean_token_accuracy": 0.8597728610038757, + "num_tokens": 244152473.0, + "step": 6283 + }, + { + "epoch": 0.7993893906627656, + "grad_norm": 2.543874979019165, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8861182332038879, + "num_tokens": 244187992.0, + "step": 6284 + }, + { + "epoch": 0.799516600941356, + "grad_norm": 2.3895206451416016, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8906634449958801, + "num_tokens": 244228843.0, + "step": 6285 + }, + { + "epoch": 0.7996438112199465, + "grad_norm": 2.59842848777771, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8687296509742737, + "num_tokens": 244270558.0, + "step": 6286 + }, + { + "epoch": 0.7997710214985371, + "grad_norm": 2.6538772583007812, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8757800459861755, + "num_tokens": 244310704.0, + "step": 6287 + }, + { + "epoch": 0.7998982317771276, + "grad_norm": 2.7394280433654785, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8630580306053162, + "num_tokens": 244346285.0, + "step": 6288 + }, + { + "epoch": 0.8000254420557181, + "grad_norm": 2.7579147815704346, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8807411193847656, + "num_tokens": 244378139.0, + "step": 6289 + }, + { + "epoch": 0.8001526523343087, + "grad_norm": 2.475444793701172, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8960063457489014, + "num_tokens": 244414920.0, + "step": 6290 + }, + { + "epoch": 0.8002798626128991, + "grad_norm": 2.4731926918029785, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8728727698326111, + "num_tokens": 244456757.0, + "step": 6291 + }, + { + "epoch": 0.8004070728914896, + "grad_norm": 2.547520637512207, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8771941661834717, + "num_tokens": 244496935.0, + "step": 6292 + }, + { + "epoch": 0.8005342831700801, + "grad_norm": 2.3351850509643555, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.885934054851532, + "num_tokens": 244540045.0, + "step": 6293 + }, + { + "epoch": 0.8006614934486707, + "grad_norm": 2.466310501098633, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8763875961303711, + "num_tokens": 244581377.0, + "step": 6294 + }, + { + "epoch": 0.8007887037272612, + "grad_norm": 2.460771083831787, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8850743770599365, + "num_tokens": 244620615.0, + "step": 6295 + }, + { + "epoch": 0.8009159140058517, + "grad_norm": 2.7869808673858643, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8786101341247559, + "num_tokens": 244653133.0, + "step": 6296 + }, + { + "epoch": 0.8010431242844421, + "grad_norm": 2.653346538543701, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8963879346847534, + "num_tokens": 244684124.0, + "step": 6297 + }, + { + "epoch": 0.8011703345630327, + "grad_norm": 2.4706010818481445, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8794936537742615, + "num_tokens": 244727685.0, + "step": 6298 + }, + { + "epoch": 0.8012975448416232, + "grad_norm": 2.724454164505005, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8711503744125366, + "num_tokens": 244763324.0, + "step": 6299 + }, + { + "epoch": 0.8014247551202137, + "grad_norm": 2.503183126449585, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8768188953399658, + "num_tokens": 244804723.0, + "step": 6300 + }, + { + "epoch": 0.8015519653988042, + "grad_norm": 2.4382176399230957, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8805813789367676, + "num_tokens": 244844275.0, + "step": 6301 + }, + { + "epoch": 0.8016791756773948, + "grad_norm": 2.5359749794006348, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8706643581390381, + "num_tokens": 244886827.0, + "step": 6302 + }, + { + "epoch": 0.8018063859559852, + "grad_norm": 2.8394510746002197, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.872414231300354, + "num_tokens": 244921263.0, + "step": 6303 + }, + { + "epoch": 0.8019335962345757, + "grad_norm": 2.5498249530792236, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8801668882369995, + "num_tokens": 244960400.0, + "step": 6304 + }, + { + "epoch": 0.8020608065131662, + "grad_norm": 2.4341816902160645, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8859447836875916, + "num_tokens": 244999790.0, + "step": 6305 + }, + { + "epoch": 0.8021880167917568, + "grad_norm": 3.018991708755493, + "learning_rate": 1e-06, + "loss": 0.3863, + "mean_token_accuracy": 0.8660945892333984, + "num_tokens": 245033795.0, + "step": 6306 + }, + { + "epoch": 0.8023152270703473, + "grad_norm": 2.7455599308013916, + "learning_rate": 1e-06, + "loss": 0.3597, + "mean_token_accuracy": 0.876596212387085, + "num_tokens": 245070738.0, + "step": 6307 + }, + { + "epoch": 0.8024424373489378, + "grad_norm": 2.7852022647857666, + "learning_rate": 1e-06, + "loss": 0.3898, + "mean_token_accuracy": 0.8634964227676392, + "num_tokens": 245114388.0, + "step": 6308 + }, + { + "epoch": 0.8025696476275284, + "grad_norm": 2.6808536052703857, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8739405870437622, + "num_tokens": 245152044.0, + "step": 6309 + }, + { + "epoch": 0.8026968579061188, + "grad_norm": 2.7652719020843506, + "learning_rate": 1e-06, + "loss": 0.3834, + "mean_token_accuracy": 0.866576075553894, + "num_tokens": 245189943.0, + "step": 6310 + }, + { + "epoch": 0.8028240681847093, + "grad_norm": 2.40238881111145, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8918403387069702, + "num_tokens": 245230976.0, + "step": 6311 + }, + { + "epoch": 0.8029512784632998, + "grad_norm": 2.757322311401367, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8653408288955688, + "num_tokens": 245268717.0, + "step": 6312 + }, + { + "epoch": 0.8030784887418904, + "grad_norm": 2.695302963256836, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8806895017623901, + "num_tokens": 245307585.0, + "step": 6313 + }, + { + "epoch": 0.8032056990204809, + "grad_norm": 2.607738971710205, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8749030828475952, + "num_tokens": 245345612.0, + "step": 6314 + }, + { + "epoch": 0.8033329092990714, + "grad_norm": 2.460897207260132, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8710517287254333, + "num_tokens": 245388108.0, + "step": 6315 + }, + { + "epoch": 0.8034601195776618, + "grad_norm": 2.431494951248169, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8870067000389099, + "num_tokens": 245428611.0, + "step": 6316 + }, + { + "epoch": 0.8035873298562524, + "grad_norm": 2.4710447788238525, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8784569501876831, + "num_tokens": 245470479.0, + "step": 6317 + }, + { + "epoch": 0.8037145401348429, + "grad_norm": 2.8449041843414307, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8718777894973755, + "num_tokens": 245506806.0, + "step": 6318 + }, + { + "epoch": 0.8038417504134334, + "grad_norm": 2.6488325595855713, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8753677606582642, + "num_tokens": 245540346.0, + "step": 6319 + }, + { + "epoch": 0.803968960692024, + "grad_norm": 3.0004873275756836, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8719772100448608, + "num_tokens": 245579663.0, + "step": 6320 + }, + { + "epoch": 0.8040961709706145, + "grad_norm": 2.859804153442383, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8882596492767334, + "num_tokens": 245615287.0, + "step": 6321 + }, + { + "epoch": 0.8042233812492049, + "grad_norm": 2.675452947616577, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8924194574356079, + "num_tokens": 245647681.0, + "step": 6322 + }, + { + "epoch": 0.8043505915277954, + "grad_norm": 2.3972883224487305, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.88995361328125, + "num_tokens": 245691281.0, + "step": 6323 + }, + { + "epoch": 0.804477801806386, + "grad_norm": 2.3992910385131836, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8782558441162109, + "num_tokens": 245732172.0, + "step": 6324 + }, + { + "epoch": 0.8046050120849765, + "grad_norm": 2.7509968280792236, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8802320957183838, + "num_tokens": 245765336.0, + "step": 6325 + }, + { + "epoch": 0.804732222363567, + "grad_norm": 2.6397435665130615, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.8736856579780579, + "num_tokens": 245804985.0, + "step": 6326 + }, + { + "epoch": 0.8048594326421575, + "grad_norm": 2.318357229232788, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8969776630401611, + "num_tokens": 245845258.0, + "step": 6327 + }, + { + "epoch": 0.804986642920748, + "grad_norm": 2.7063772678375244, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.887967586517334, + "num_tokens": 245877493.0, + "step": 6328 + }, + { + "epoch": 0.8051138531993385, + "grad_norm": 2.569385528564453, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8769978284835815, + "num_tokens": 245915391.0, + "step": 6329 + }, + { + "epoch": 0.805241063477929, + "grad_norm": 2.641826629638672, + "learning_rate": 1e-06, + "loss": 0.3634, + "mean_token_accuracy": 0.874656081199646, + "num_tokens": 245953808.0, + "step": 6330 + }, + { + "epoch": 0.8053682737565195, + "grad_norm": 2.6162521839141846, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8790634870529175, + "num_tokens": 245992644.0, + "step": 6331 + }, + { + "epoch": 0.8054954840351101, + "grad_norm": 2.9033968448638916, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8750723004341125, + "num_tokens": 246023955.0, + "step": 6332 + }, + { + "epoch": 0.8056226943137006, + "grad_norm": 2.492418050765991, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8850213885307312, + "num_tokens": 246064280.0, + "step": 6333 + }, + { + "epoch": 0.805749904592291, + "grad_norm": 2.4997191429138184, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8755658268928528, + "num_tokens": 246106592.0, + "step": 6334 + }, + { + "epoch": 0.8058771148708815, + "grad_norm": 2.699693441390991, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8862504959106445, + "num_tokens": 246145013.0, + "step": 6335 + }, + { + "epoch": 0.8060043251494721, + "grad_norm": 2.453239679336548, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8782603144645691, + "num_tokens": 246190048.0, + "step": 6336 + }, + { + "epoch": 0.8061315354280626, + "grad_norm": 2.5204453468322754, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8748642206192017, + "num_tokens": 246234853.0, + "step": 6337 + }, + { + "epoch": 0.8062587457066531, + "grad_norm": 2.3251123428344727, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8959195613861084, + "num_tokens": 246274579.0, + "step": 6338 + }, + { + "epoch": 0.8063859559852437, + "grad_norm": 2.5656046867370605, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8783688545227051, + "num_tokens": 246311555.0, + "step": 6339 + }, + { + "epoch": 0.8065131662638341, + "grad_norm": 2.6181445121765137, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8858621120452881, + "num_tokens": 246344914.0, + "step": 6340 + }, + { + "epoch": 0.8066403765424246, + "grad_norm": 2.588284492492676, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8668089509010315, + "num_tokens": 246387715.0, + "step": 6341 + }, + { + "epoch": 0.8067675868210151, + "grad_norm": 2.7286581993103027, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8808594942092896, + "num_tokens": 246424571.0, + "step": 6342 + }, + { + "epoch": 0.8068947970996057, + "grad_norm": 2.6558375358581543, + "learning_rate": 1e-06, + "loss": 0.3888, + "mean_token_accuracy": 0.8634305000305176, + "num_tokens": 246467753.0, + "step": 6343 + }, + { + "epoch": 0.8070220073781962, + "grad_norm": 2.602403163909912, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8819986581802368, + "num_tokens": 246502726.0, + "step": 6344 + }, + { + "epoch": 0.8071492176567867, + "grad_norm": 2.3161845207214355, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9014174938201904, + "num_tokens": 246543646.0, + "step": 6345 + }, + { + "epoch": 0.8072764279353771, + "grad_norm": 2.527350902557373, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8816533088684082, + "num_tokens": 246580981.0, + "step": 6346 + }, + { + "epoch": 0.8074036382139677, + "grad_norm": 2.5221681594848633, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8785640001296997, + "num_tokens": 246622378.0, + "step": 6347 + }, + { + "epoch": 0.8075308484925582, + "grad_norm": 2.6059348583221436, + "learning_rate": 1e-06, + "loss": 0.4133, + "mean_token_accuracy": 0.8594191074371338, + "num_tokens": 246666247.0, + "step": 6348 + }, + { + "epoch": 0.8076580587711487, + "grad_norm": 2.5380184650421143, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8654739856719971, + "num_tokens": 246708041.0, + "step": 6349 + }, + { + "epoch": 0.8077852690497392, + "grad_norm": 2.4110753536224365, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8761444091796875, + "num_tokens": 246751071.0, + "step": 6350 + }, + { + "epoch": 0.8079124793283298, + "grad_norm": 2.6451077461242676, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8766027092933655, + "num_tokens": 246787173.0, + "step": 6351 + }, + { + "epoch": 0.8080396896069202, + "grad_norm": 2.6207284927368164, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8733972311019897, + "num_tokens": 246828208.0, + "step": 6352 + }, + { + "epoch": 0.8081668998855107, + "grad_norm": 2.8085312843322754, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8836274147033691, + "num_tokens": 246859897.0, + "step": 6353 + }, + { + "epoch": 0.8082941101641012, + "grad_norm": 2.665673017501831, + "learning_rate": 1e-06, + "loss": 0.3914, + "mean_token_accuracy": 0.8657678961753845, + "num_tokens": 246900616.0, + "step": 6354 + }, + { + "epoch": 0.8084213204426918, + "grad_norm": 2.590114116668701, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8772842884063721, + "num_tokens": 246938885.0, + "step": 6355 + }, + { + "epoch": 0.8085485307212823, + "grad_norm": 2.5593960285186768, + "learning_rate": 1e-06, + "loss": 0.3806, + "mean_token_accuracy": 0.8675693273544312, + "num_tokens": 246978394.0, + "step": 6356 + }, + { + "epoch": 0.8086757409998728, + "grad_norm": 2.7161519527435303, + "learning_rate": 1e-06, + "loss": 0.3905, + "mean_token_accuracy": 0.8643943071365356, + "num_tokens": 247016554.0, + "step": 6357 + }, + { + "epoch": 0.8088029512784632, + "grad_norm": 2.4562127590179443, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8766927719116211, + "num_tokens": 247059009.0, + "step": 6358 + }, + { + "epoch": 0.8089301615570538, + "grad_norm": 2.6618006229400635, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8676882386207581, + "num_tokens": 247098356.0, + "step": 6359 + }, + { + "epoch": 0.8090573718356443, + "grad_norm": 2.4126148223876953, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8746280074119568, + "num_tokens": 247137966.0, + "step": 6360 + }, + { + "epoch": 0.8091845821142348, + "grad_norm": 2.4754748344421387, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.881890594959259, + "num_tokens": 247178815.0, + "step": 6361 + }, + { + "epoch": 0.8093117923928254, + "grad_norm": 2.5641915798187256, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8745566606521606, + "num_tokens": 247219005.0, + "step": 6362 + }, + { + "epoch": 0.8094390026714159, + "grad_norm": 2.6010005474090576, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8753812909126282, + "num_tokens": 247256509.0, + "step": 6363 + }, + { + "epoch": 0.8095662129500064, + "grad_norm": 2.7128002643585205, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.870688796043396, + "num_tokens": 247292979.0, + "step": 6364 + }, + { + "epoch": 0.8096934232285968, + "grad_norm": 2.494420289993286, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8782305121421814, + "num_tokens": 247332847.0, + "step": 6365 + }, + { + "epoch": 0.8098206335071874, + "grad_norm": 2.7026822566986084, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8733747005462646, + "num_tokens": 247367668.0, + "step": 6366 + }, + { + "epoch": 0.8099478437857779, + "grad_norm": 2.393867015838623, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.896477460861206, + "num_tokens": 247405016.0, + "step": 6367 + }, + { + "epoch": 0.8100750540643684, + "grad_norm": 2.4168362617492676, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.8999190330505371, + "num_tokens": 247442800.0, + "step": 6368 + }, + { + "epoch": 0.8102022643429589, + "grad_norm": 2.7312190532684326, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8803929686546326, + "num_tokens": 247476370.0, + "step": 6369 + }, + { + "epoch": 0.8103294746215495, + "grad_norm": 2.6570892333984375, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8707687854766846, + "num_tokens": 247512522.0, + "step": 6370 + }, + { + "epoch": 0.8104566849001399, + "grad_norm": 2.4668281078338623, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8829546570777893, + "num_tokens": 247553293.0, + "step": 6371 + }, + { + "epoch": 0.8105838951787304, + "grad_norm": 2.8042726516723633, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8772160410881042, + "num_tokens": 247588216.0, + "step": 6372 + }, + { + "epoch": 0.810711105457321, + "grad_norm": 2.528499126434326, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8792264461517334, + "num_tokens": 247627152.0, + "step": 6373 + }, + { + "epoch": 0.8108383157359115, + "grad_norm": 2.598822593688965, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.883394718170166, + "num_tokens": 247664923.0, + "step": 6374 + }, + { + "epoch": 0.810965526014502, + "grad_norm": 2.6789920330047607, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8766123056411743, + "num_tokens": 247702299.0, + "step": 6375 + }, + { + "epoch": 0.8110927362930925, + "grad_norm": 2.685633420944214, + "learning_rate": 1e-06, + "loss": 0.3591, + "mean_token_accuracy": 0.8729970455169678, + "num_tokens": 247740947.0, + "step": 6376 + }, + { + "epoch": 0.811219946571683, + "grad_norm": 2.9265408515930176, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.867921769618988, + "num_tokens": 247773988.0, + "step": 6377 + }, + { + "epoch": 0.8113471568502735, + "grad_norm": 2.6526081562042236, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8751099705696106, + "num_tokens": 247817589.0, + "step": 6378 + }, + { + "epoch": 0.811474367128864, + "grad_norm": 2.6153552532196045, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8915270566940308, + "num_tokens": 247852721.0, + "step": 6379 + }, + { + "epoch": 0.8116015774074545, + "grad_norm": 2.3450164794921875, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8931049704551697, + "num_tokens": 247894497.0, + "step": 6380 + }, + { + "epoch": 0.8117287876860451, + "grad_norm": 2.6819918155670166, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8701672554016113, + "num_tokens": 247930522.0, + "step": 6381 + }, + { + "epoch": 0.8118559979646356, + "grad_norm": 2.3320460319519043, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8926427960395813, + "num_tokens": 247971859.0, + "step": 6382 + }, + { + "epoch": 0.811983208243226, + "grad_norm": 2.390592098236084, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8808389902114868, + "num_tokens": 248016947.0, + "step": 6383 + }, + { + "epoch": 0.8121104185218165, + "grad_norm": 2.6611995697021484, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8693243861198425, + "num_tokens": 248058570.0, + "step": 6384 + }, + { + "epoch": 0.8122376288004071, + "grad_norm": 2.63942289352417, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.881700873374939, + "num_tokens": 248093269.0, + "step": 6385 + }, + { + "epoch": 0.8123648390789976, + "grad_norm": 2.6651370525360107, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8660660982131958, + "num_tokens": 248131464.0, + "step": 6386 + }, + { + "epoch": 0.8124920493575881, + "grad_norm": 2.613600730895996, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8877116441726685, + "num_tokens": 248165522.0, + "step": 6387 + }, + { + "epoch": 0.8126192596361786, + "grad_norm": 2.5961718559265137, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8809387683868408, + "num_tokens": 248205560.0, + "step": 6388 + }, + { + "epoch": 0.8127464699147691, + "grad_norm": 2.6435980796813965, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.8676550388336182, + "num_tokens": 248245743.0, + "step": 6389 + }, + { + "epoch": 0.8128736801933596, + "grad_norm": 2.4386544227600098, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8825608491897583, + "num_tokens": 248288338.0, + "step": 6390 + }, + { + "epoch": 0.8130008904719501, + "grad_norm": 2.6284079551696777, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8708905577659607, + "num_tokens": 248331270.0, + "step": 6391 + }, + { + "epoch": 0.8131281007505406, + "grad_norm": 2.31534743309021, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8785743713378906, + "num_tokens": 248376316.0, + "step": 6392 + }, + { + "epoch": 0.8132553110291312, + "grad_norm": 2.550368547439575, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8770261406898499, + "num_tokens": 248414240.0, + "step": 6393 + }, + { + "epoch": 0.8133825213077217, + "grad_norm": 2.59621262550354, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.885288655757904, + "num_tokens": 248448366.0, + "step": 6394 + }, + { + "epoch": 0.8135097315863121, + "grad_norm": 2.6552894115448, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8756599426269531, + "num_tokens": 248487557.0, + "step": 6395 + }, + { + "epoch": 0.8136369418649027, + "grad_norm": 2.402282476425171, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8785439729690552, + "num_tokens": 248534508.0, + "step": 6396 + }, + { + "epoch": 0.8137641521434932, + "grad_norm": 2.8296048641204834, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8898837566375732, + "num_tokens": 248562781.0, + "step": 6397 + }, + { + "epoch": 0.8138913624220837, + "grad_norm": 2.516704559326172, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8767344355583191, + "num_tokens": 248600582.0, + "step": 6398 + }, + { + "epoch": 0.8140185727006742, + "grad_norm": 2.4611895084381104, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8744844198226929, + "num_tokens": 248643092.0, + "step": 6399 + }, + { + "epoch": 0.8141457829792648, + "grad_norm": 2.6335227489471436, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8801289200782776, + "num_tokens": 248684137.0, + "step": 6400 + }, + { + "epoch": 0.8142729932578552, + "grad_norm": 2.774601936340332, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8943347930908203, + "num_tokens": 248715918.0, + "step": 6401 + }, + { + "epoch": 0.8144002035364457, + "grad_norm": 2.620758533477783, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8676306009292603, + "num_tokens": 248752571.0, + "step": 6402 + }, + { + "epoch": 0.8145274138150362, + "grad_norm": 2.6149556636810303, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8825666308403015, + "num_tokens": 248787548.0, + "step": 6403 + }, + { + "epoch": 0.8146546240936268, + "grad_norm": 2.6854348182678223, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8742040395736694, + "num_tokens": 248825760.0, + "step": 6404 + }, + { + "epoch": 0.8147818343722173, + "grad_norm": 2.875828504562378, + "learning_rate": 1e-06, + "loss": 0.4224, + "mean_token_accuracy": 0.8527417182922363, + "num_tokens": 248864797.0, + "step": 6405 + }, + { + "epoch": 0.8149090446508078, + "grad_norm": 2.399031162261963, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8875607848167419, + "num_tokens": 248909471.0, + "step": 6406 + }, + { + "epoch": 0.8150362549293982, + "grad_norm": 2.559030532836914, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8793476819992065, + "num_tokens": 248949829.0, + "step": 6407 + }, + { + "epoch": 0.8151634652079888, + "grad_norm": 2.5063107013702393, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8834267854690552, + "num_tokens": 248990608.0, + "step": 6408 + }, + { + "epoch": 0.8152906754865793, + "grad_norm": 2.4472310543060303, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8881946206092834, + "num_tokens": 249031858.0, + "step": 6409 + }, + { + "epoch": 0.8154178857651698, + "grad_norm": 2.656168222427368, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8782641291618347, + "num_tokens": 249069704.0, + "step": 6410 + }, + { + "epoch": 0.8155450960437604, + "grad_norm": 2.530848264694214, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8808739185333252, + "num_tokens": 249108445.0, + "step": 6411 + }, + { + "epoch": 0.8156723063223509, + "grad_norm": 2.4163103103637695, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8846216201782227, + "num_tokens": 249153534.0, + "step": 6412 + }, + { + "epoch": 0.8157995166009414, + "grad_norm": 2.6233670711517334, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8767416477203369, + "num_tokens": 249190008.0, + "step": 6413 + }, + { + "epoch": 0.8159267268795318, + "grad_norm": 2.546464204788208, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8825125694274902, + "num_tokens": 249229505.0, + "step": 6414 + }, + { + "epoch": 0.8160539371581224, + "grad_norm": 2.7406458854675293, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8798659443855286, + "num_tokens": 249265603.0, + "step": 6415 + }, + { + "epoch": 0.8161811474367129, + "grad_norm": 2.5820579528808594, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8788392543792725, + "num_tokens": 249302963.0, + "step": 6416 + }, + { + "epoch": 0.8163083577153034, + "grad_norm": 2.637051582336426, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8789315223693848, + "num_tokens": 249341730.0, + "step": 6417 + }, + { + "epoch": 0.8164355679938939, + "grad_norm": 2.7598578929901123, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8721654415130615, + "num_tokens": 249376280.0, + "step": 6418 + }, + { + "epoch": 0.8165627782724845, + "grad_norm": 2.576484203338623, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8827223777770996, + "num_tokens": 249416394.0, + "step": 6419 + }, + { + "epoch": 0.8166899885510749, + "grad_norm": 2.575167655944824, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8701153993606567, + "num_tokens": 249456930.0, + "step": 6420 + }, + { + "epoch": 0.8168171988296654, + "grad_norm": 2.6348724365234375, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8692812919616699, + "num_tokens": 249494229.0, + "step": 6421 + }, + { + "epoch": 0.8169444091082559, + "grad_norm": 2.518303632736206, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8936793804168701, + "num_tokens": 249530291.0, + "step": 6422 + }, + { + "epoch": 0.8170716193868465, + "grad_norm": 2.534621477127075, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.882636547088623, + "num_tokens": 249568247.0, + "step": 6423 + }, + { + "epoch": 0.817198829665437, + "grad_norm": 2.7036094665527344, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8683959245681763, + "num_tokens": 249604784.0, + "step": 6424 + }, + { + "epoch": 0.8173260399440275, + "grad_norm": 2.7174439430236816, + "learning_rate": 1e-06, + "loss": 0.3921, + "mean_token_accuracy": 0.8621177077293396, + "num_tokens": 249645679.0, + "step": 6425 + }, + { + "epoch": 0.8174532502226179, + "grad_norm": 2.481844186782837, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8926230669021606, + "num_tokens": 249684307.0, + "step": 6426 + }, + { + "epoch": 0.8175804605012085, + "grad_norm": 2.5269854068756104, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8851213455200195, + "num_tokens": 249721466.0, + "step": 6427 + }, + { + "epoch": 0.817707670779799, + "grad_norm": 2.558556318283081, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8783597946166992, + "num_tokens": 249760088.0, + "step": 6428 + }, + { + "epoch": 0.8178348810583895, + "grad_norm": 2.428617477416992, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8804271221160889, + "num_tokens": 249802686.0, + "step": 6429 + }, + { + "epoch": 0.8179620913369801, + "grad_norm": 2.5323543548583984, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8805639743804932, + "num_tokens": 249841487.0, + "step": 6430 + }, + { + "epoch": 0.8180893016155706, + "grad_norm": 2.842478036880493, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8880782127380371, + "num_tokens": 249880268.0, + "step": 6431 + }, + { + "epoch": 0.818216511894161, + "grad_norm": 2.4745779037475586, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8883537650108337, + "num_tokens": 249922201.0, + "step": 6432 + }, + { + "epoch": 0.8183437221727515, + "grad_norm": 3.0081496238708496, + "learning_rate": 1e-06, + "loss": 0.3968, + "mean_token_accuracy": 0.8644542694091797, + "num_tokens": 249956150.0, + "step": 6433 + }, + { + "epoch": 0.8184709324513421, + "grad_norm": 2.817796230316162, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8742129802703857, + "num_tokens": 249991525.0, + "step": 6434 + }, + { + "epoch": 0.8185981427299326, + "grad_norm": 2.5131428241729736, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8827719688415527, + "num_tokens": 250032279.0, + "step": 6435 + }, + { + "epoch": 0.8187253530085231, + "grad_norm": 2.6211624145507812, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8727123737335205, + "num_tokens": 250073237.0, + "step": 6436 + }, + { + "epoch": 0.8188525632871136, + "grad_norm": 2.949720621109009, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8953714370727539, + "num_tokens": 250109125.0, + "step": 6437 + }, + { + "epoch": 0.8189797735657041, + "grad_norm": 2.473498582839966, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8857243061065674, + "num_tokens": 250149328.0, + "step": 6438 + }, + { + "epoch": 0.8191069838442946, + "grad_norm": 2.4523115158081055, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8867047429084778, + "num_tokens": 250187052.0, + "step": 6439 + }, + { + "epoch": 0.8192341941228851, + "grad_norm": 2.678793430328369, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8746507167816162, + "num_tokens": 250223359.0, + "step": 6440 + }, + { + "epoch": 0.8193614044014756, + "grad_norm": 2.7089931964874268, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.8694337606430054, + "num_tokens": 250260838.0, + "step": 6441 + }, + { + "epoch": 0.8194886146800662, + "grad_norm": 2.4463021755218506, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8799116611480713, + "num_tokens": 250303484.0, + "step": 6442 + }, + { + "epoch": 0.8196158249586567, + "grad_norm": 2.6173572540283203, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8774918913841248, + "num_tokens": 250338068.0, + "step": 6443 + }, + { + "epoch": 0.8197430352372471, + "grad_norm": 2.519357681274414, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8682061433792114, + "num_tokens": 250379244.0, + "step": 6444 + }, + { + "epoch": 0.8198702455158376, + "grad_norm": 2.624124050140381, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8801934123039246, + "num_tokens": 250415662.0, + "step": 6445 + }, + { + "epoch": 0.8199974557944282, + "grad_norm": 2.4568631649017334, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8879724740982056, + "num_tokens": 250456798.0, + "step": 6446 + }, + { + "epoch": 0.8201246660730187, + "grad_norm": 2.7273051738739014, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8702985048294067, + "num_tokens": 250493497.0, + "step": 6447 + }, + { + "epoch": 0.8202518763516092, + "grad_norm": 2.767958164215088, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8773972988128662, + "num_tokens": 250528782.0, + "step": 6448 + }, + { + "epoch": 0.8203790866301998, + "grad_norm": 2.6395249366760254, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8921812176704407, + "num_tokens": 250564926.0, + "step": 6449 + }, + { + "epoch": 0.8205062969087902, + "grad_norm": 2.3780319690704346, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8796218633651733, + "num_tokens": 250604784.0, + "step": 6450 + }, + { + "epoch": 0.8206335071873807, + "grad_norm": 2.647155284881592, + "learning_rate": 1e-06, + "loss": 0.4133, + "mean_token_accuracy": 0.8636849522590637, + "num_tokens": 250646067.0, + "step": 6451 + }, + { + "epoch": 0.8207607174659712, + "grad_norm": 2.808867931365967, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8813992738723755, + "num_tokens": 250681133.0, + "step": 6452 + }, + { + "epoch": 0.8208879277445618, + "grad_norm": 2.549161672592163, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8801021575927734, + "num_tokens": 250719310.0, + "step": 6453 + }, + { + "epoch": 0.8210151380231523, + "grad_norm": 2.3085989952087402, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8953073620796204, + "num_tokens": 250760810.0, + "step": 6454 + }, + { + "epoch": 0.8211423483017428, + "grad_norm": 2.3308892250061035, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8924450278282166, + "num_tokens": 250800038.0, + "step": 6455 + }, + { + "epoch": 0.8212695585803332, + "grad_norm": 2.474329710006714, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8769862651824951, + "num_tokens": 250837275.0, + "step": 6456 + }, + { + "epoch": 0.8213967688589238, + "grad_norm": 2.5212106704711914, + "learning_rate": 1e-06, + "loss": 0.3514, + "mean_token_accuracy": 0.8772542476654053, + "num_tokens": 250876793.0, + "step": 6457 + }, + { + "epoch": 0.8215239791375143, + "grad_norm": 2.9605462551116943, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8662995100021362, + "num_tokens": 250908802.0, + "step": 6458 + }, + { + "epoch": 0.8216511894161048, + "grad_norm": 2.3936715126037598, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.868816614151001, + "num_tokens": 250954838.0, + "step": 6459 + }, + { + "epoch": 0.8217783996946953, + "grad_norm": 2.661271810531616, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8632261753082275, + "num_tokens": 250996689.0, + "step": 6460 + }, + { + "epoch": 0.8219056099732859, + "grad_norm": 2.6590750217437744, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8671513795852661, + "num_tokens": 251036971.0, + "step": 6461 + }, + { + "epoch": 0.8220328202518764, + "grad_norm": 2.719590663909912, + "learning_rate": 1e-06, + "loss": 0.3716, + "mean_token_accuracy": 0.8724236488342285, + "num_tokens": 251072505.0, + "step": 6462 + }, + { + "epoch": 0.8221600305304668, + "grad_norm": 2.567760467529297, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8797305822372437, + "num_tokens": 251108665.0, + "step": 6463 + }, + { + "epoch": 0.8222872408090574, + "grad_norm": 2.7048256397247314, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8745777606964111, + "num_tokens": 251146579.0, + "step": 6464 + }, + { + "epoch": 0.8224144510876479, + "grad_norm": 2.415397882461548, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8820594549179077, + "num_tokens": 251188961.0, + "step": 6465 + }, + { + "epoch": 0.8225416613662384, + "grad_norm": 2.5154778957366943, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.9000735282897949, + "num_tokens": 251225916.0, + "step": 6466 + }, + { + "epoch": 0.8226688716448289, + "grad_norm": 2.5829474925994873, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8687146306037903, + "num_tokens": 251267704.0, + "step": 6467 + }, + { + "epoch": 0.8227960819234195, + "grad_norm": 2.8783514499664307, + "learning_rate": 1e-06, + "loss": 0.4151, + "mean_token_accuracy": 0.8591228723526001, + "num_tokens": 251306211.0, + "step": 6468 + }, + { + "epoch": 0.8229232922020099, + "grad_norm": 2.608353614807129, + "learning_rate": 1e-06, + "loss": 0.384, + "mean_token_accuracy": 0.8657739162445068, + "num_tokens": 251351058.0, + "step": 6469 + }, + { + "epoch": 0.8230505024806004, + "grad_norm": 2.4056761264801025, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8972584009170532, + "num_tokens": 251388417.0, + "step": 6470 + }, + { + "epoch": 0.8231777127591909, + "grad_norm": 2.561126708984375, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8732697367668152, + "num_tokens": 251427567.0, + "step": 6471 + }, + { + "epoch": 0.8233049230377815, + "grad_norm": 2.6291446685791016, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8700844645500183, + "num_tokens": 251466451.0, + "step": 6472 + }, + { + "epoch": 0.823432133316372, + "grad_norm": 2.48765230178833, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8813209533691406, + "num_tokens": 251508661.0, + "step": 6473 + }, + { + "epoch": 0.8235593435949625, + "grad_norm": 2.567441463470459, + "learning_rate": 1e-06, + "loss": 0.3953, + "mean_token_accuracy": 0.8645235300064087, + "num_tokens": 251550673.0, + "step": 6474 + }, + { + "epoch": 0.8236865538735529, + "grad_norm": 2.5534489154815674, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8825651407241821, + "num_tokens": 251587162.0, + "step": 6475 + }, + { + "epoch": 0.8238137641521435, + "grad_norm": 2.6966447830200195, + "learning_rate": 1e-06, + "loss": 0.3831, + "mean_token_accuracy": 0.8670090436935425, + "num_tokens": 251626765.0, + "step": 6476 + }, + { + "epoch": 0.823940974430734, + "grad_norm": 2.502969980239868, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8728923201560974, + "num_tokens": 251668184.0, + "step": 6477 + }, + { + "epoch": 0.8240681847093245, + "grad_norm": 2.6725237369537354, + "learning_rate": 1e-06, + "loss": 0.4099, + "mean_token_accuracy": 0.856215238571167, + "num_tokens": 251709794.0, + "step": 6478 + }, + { + "epoch": 0.824195394987915, + "grad_norm": 2.54140567779541, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.876421332359314, + "num_tokens": 251750634.0, + "step": 6479 + }, + { + "epoch": 0.8243226052665056, + "grad_norm": 2.625197410583496, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8792423009872437, + "num_tokens": 251789884.0, + "step": 6480 + }, + { + "epoch": 0.824449815545096, + "grad_norm": 2.5977885723114014, + "learning_rate": 1e-06, + "loss": 0.3835, + "mean_token_accuracy": 0.8668777346611023, + "num_tokens": 251829238.0, + "step": 6481 + }, + { + "epoch": 0.8245770258236865, + "grad_norm": 2.754300355911255, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8750650882720947, + "num_tokens": 251866352.0, + "step": 6482 + }, + { + "epoch": 0.824704236102277, + "grad_norm": 2.5341956615448, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8817112445831299, + "num_tokens": 251908528.0, + "step": 6483 + }, + { + "epoch": 0.8248314463808676, + "grad_norm": 2.686002254486084, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8811620473861694, + "num_tokens": 251942869.0, + "step": 6484 + }, + { + "epoch": 0.8249586566594581, + "grad_norm": 2.489889621734619, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8851199150085449, + "num_tokens": 251981623.0, + "step": 6485 + }, + { + "epoch": 0.8250858669380486, + "grad_norm": 2.233832836151123, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8902517557144165, + "num_tokens": 252029010.0, + "step": 6486 + }, + { + "epoch": 0.8252130772166391, + "grad_norm": 2.799888849258423, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8685423135757446, + "num_tokens": 252064973.0, + "step": 6487 + }, + { + "epoch": 0.8253402874952296, + "grad_norm": 2.6439194679260254, + "learning_rate": 1e-06, + "loss": 0.3799, + "mean_token_accuracy": 0.8734956979751587, + "num_tokens": 252103641.0, + "step": 6488 + }, + { + "epoch": 0.8254674977738201, + "grad_norm": 2.7388408184051514, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.878842830657959, + "num_tokens": 252141801.0, + "step": 6489 + }, + { + "epoch": 0.8255947080524106, + "grad_norm": 2.729644298553467, + "learning_rate": 1e-06, + "loss": 0.4019, + "mean_token_accuracy": 0.8678104877471924, + "num_tokens": 252180760.0, + "step": 6490 + }, + { + "epoch": 0.8257219183310012, + "grad_norm": 2.4687790870666504, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8691526055335999, + "num_tokens": 252223726.0, + "step": 6491 + }, + { + "epoch": 0.8258491286095917, + "grad_norm": 2.480691909790039, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8943290710449219, + "num_tokens": 252260207.0, + "step": 6492 + }, + { + "epoch": 0.8259763388881821, + "grad_norm": 2.580780029296875, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8755176067352295, + "num_tokens": 252294996.0, + "step": 6493 + }, + { + "epoch": 0.8261035491667726, + "grad_norm": 2.378098487854004, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8911347389221191, + "num_tokens": 252337573.0, + "step": 6494 + }, + { + "epoch": 0.8262307594453632, + "grad_norm": 2.5181145668029785, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8802861571311951, + "num_tokens": 252375552.0, + "step": 6495 + }, + { + "epoch": 0.8263579697239537, + "grad_norm": 2.607205390930176, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8932615518569946, + "num_tokens": 252409339.0, + "step": 6496 + }, + { + "epoch": 0.8264851800025442, + "grad_norm": 2.4278206825256348, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8752138614654541, + "num_tokens": 252450800.0, + "step": 6497 + }, + { + "epoch": 0.8266123902811348, + "grad_norm": 2.749807357788086, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8765023946762085, + "num_tokens": 252484444.0, + "step": 6498 + }, + { + "epoch": 0.8267396005597252, + "grad_norm": 2.4732284545898438, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8766346573829651, + "num_tokens": 252522557.0, + "step": 6499 + }, + { + "epoch": 0.8268668108383157, + "grad_norm": 2.3695390224456787, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8821949362754822, + "num_tokens": 252564270.0, + "step": 6500 + }, + { + "epoch": 0.8269940211169062, + "grad_norm": 2.4210946559906006, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8730440139770508, + "num_tokens": 252607409.0, + "step": 6501 + }, + { + "epoch": 0.8271212313954968, + "grad_norm": 2.3870174884796143, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8882309794425964, + "num_tokens": 252650723.0, + "step": 6502 + }, + { + "epoch": 0.8272484416740873, + "grad_norm": 2.4488539695739746, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8692367076873779, + "num_tokens": 252695373.0, + "step": 6503 + }, + { + "epoch": 0.8273756519526778, + "grad_norm": 2.878838062286377, + "learning_rate": 1e-06, + "loss": 0.3699, + "mean_token_accuracy": 0.8714778423309326, + "num_tokens": 252728307.0, + "step": 6504 + }, + { + "epoch": 0.8275028622312682, + "grad_norm": 2.652477741241455, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.870780885219574, + "num_tokens": 252771024.0, + "step": 6505 + }, + { + "epoch": 0.8276300725098588, + "grad_norm": 2.427823305130005, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8913605809211731, + "num_tokens": 252807392.0, + "step": 6506 + }, + { + "epoch": 0.8277572827884493, + "grad_norm": 2.5095396041870117, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.874977171421051, + "num_tokens": 252851008.0, + "step": 6507 + }, + { + "epoch": 0.8278844930670398, + "grad_norm": 2.3526923656463623, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8801863193511963, + "num_tokens": 252900954.0, + "step": 6508 + }, + { + "epoch": 0.8280117033456303, + "grad_norm": 2.8265228271484375, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8811944723129272, + "num_tokens": 252938507.0, + "step": 6509 + }, + { + "epoch": 0.8281389136242209, + "grad_norm": 2.474865436553955, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8828699588775635, + "num_tokens": 252980727.0, + "step": 6510 + }, + { + "epoch": 0.8282661239028114, + "grad_norm": 2.4429471492767334, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8791581392288208, + "num_tokens": 253023863.0, + "step": 6511 + }, + { + "epoch": 0.8283933341814018, + "grad_norm": 2.8558189868927, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8805180788040161, + "num_tokens": 253057200.0, + "step": 6512 + }, + { + "epoch": 0.8285205444599923, + "grad_norm": 2.861389398574829, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.8726400136947632, + "num_tokens": 253090913.0, + "step": 6513 + }, + { + "epoch": 0.8286477547385829, + "grad_norm": 2.7279598712921143, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8752439022064209, + "num_tokens": 253125005.0, + "step": 6514 + }, + { + "epoch": 0.8287749650171734, + "grad_norm": 2.6125540733337402, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8803002238273621, + "num_tokens": 253162012.0, + "step": 6515 + }, + { + "epoch": 0.8289021752957639, + "grad_norm": 2.4947710037231445, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8660341501235962, + "num_tokens": 253203746.0, + "step": 6516 + }, + { + "epoch": 0.8290293855743545, + "grad_norm": 2.3563132286071777, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8797764182090759, + "num_tokens": 253244585.0, + "step": 6517 + }, + { + "epoch": 0.8291565958529449, + "grad_norm": 2.3998348712921143, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.878301203250885, + "num_tokens": 253287363.0, + "step": 6518 + }, + { + "epoch": 0.8292838061315354, + "grad_norm": 2.731900215148926, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8817834854125977, + "num_tokens": 253321160.0, + "step": 6519 + }, + { + "epoch": 0.8294110164101259, + "grad_norm": 2.7723655700683594, + "learning_rate": 1e-06, + "loss": 0.3875, + "mean_token_accuracy": 0.8664548397064209, + "num_tokens": 253358555.0, + "step": 6520 + }, + { + "epoch": 0.8295382266887165, + "grad_norm": 2.7162489891052246, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8727450370788574, + "num_tokens": 253396028.0, + "step": 6521 + }, + { + "epoch": 0.829665436967307, + "grad_norm": 2.424922466278076, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8817576766014099, + "num_tokens": 253436070.0, + "step": 6522 + }, + { + "epoch": 0.8297926472458975, + "grad_norm": 2.239765167236328, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8817806243896484, + "num_tokens": 253482492.0, + "step": 6523 + }, + { + "epoch": 0.8299198575244879, + "grad_norm": 2.3293139934539795, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8936423659324646, + "num_tokens": 253524040.0, + "step": 6524 + }, + { + "epoch": 0.8300470678030785, + "grad_norm": 2.4532272815704346, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8760921955108643, + "num_tokens": 253570035.0, + "step": 6525 + }, + { + "epoch": 0.830174278081669, + "grad_norm": 2.8471012115478516, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8702972531318665, + "num_tokens": 253603712.0, + "step": 6526 + }, + { + "epoch": 0.8303014883602595, + "grad_norm": 2.6371757984161377, + "learning_rate": 1e-06, + "loss": 0.3801, + "mean_token_accuracy": 0.8655391931533813, + "num_tokens": 253642935.0, + "step": 6527 + }, + { + "epoch": 0.83042869863885, + "grad_norm": 2.863077402114868, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.895447850227356, + "num_tokens": 253671317.0, + "step": 6528 + }, + { + "epoch": 0.8305559089174406, + "grad_norm": 2.7269997596740723, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8746333718299866, + "num_tokens": 253706316.0, + "step": 6529 + }, + { + "epoch": 0.830683119196031, + "grad_norm": 2.787126302719116, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8824937343597412, + "num_tokens": 253741237.0, + "step": 6530 + }, + { + "epoch": 0.8308103294746215, + "grad_norm": 2.809450149536133, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.867048442363739, + "num_tokens": 253779316.0, + "step": 6531 + }, + { + "epoch": 0.830937539753212, + "grad_norm": 2.6558773517608643, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8814989924430847, + "num_tokens": 253816121.0, + "step": 6532 + }, + { + "epoch": 0.8310647500318026, + "grad_norm": 2.6080267429351807, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8790504336357117, + "num_tokens": 253855510.0, + "step": 6533 + }, + { + "epoch": 0.8311919603103931, + "grad_norm": 2.463576555252075, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.886111319065094, + "num_tokens": 253893544.0, + "step": 6534 + }, + { + "epoch": 0.8313191705889836, + "grad_norm": 2.6188833713531494, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8785408139228821, + "num_tokens": 253931021.0, + "step": 6535 + }, + { + "epoch": 0.831446380867574, + "grad_norm": 2.6277661323547363, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8931894898414612, + "num_tokens": 253966622.0, + "step": 6536 + }, + { + "epoch": 0.8315735911461646, + "grad_norm": 2.501002550125122, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.885582685470581, + "num_tokens": 254008813.0, + "step": 6537 + }, + { + "epoch": 0.8317008014247551, + "grad_norm": 2.4036972522735596, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8828082084655762, + "num_tokens": 254049546.0, + "step": 6538 + }, + { + "epoch": 0.8318280117033456, + "grad_norm": 2.535533905029297, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8828649520874023, + "num_tokens": 254088312.0, + "step": 6539 + }, + { + "epoch": 0.8319552219819362, + "grad_norm": 2.84478497505188, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8787025809288025, + "num_tokens": 254119991.0, + "step": 6540 + }, + { + "epoch": 0.8320824322605267, + "grad_norm": 2.5078916549682617, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8827238082885742, + "num_tokens": 254159733.0, + "step": 6541 + }, + { + "epoch": 0.8322096425391171, + "grad_norm": 2.368123769760132, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8895962238311768, + "num_tokens": 254197403.0, + "step": 6542 + }, + { + "epoch": 0.8323368528177076, + "grad_norm": 2.514012336730957, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8914554119110107, + "num_tokens": 254233366.0, + "step": 6543 + }, + { + "epoch": 0.8324640630962982, + "grad_norm": 2.691145181655884, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.8675227165222168, + "num_tokens": 254272029.0, + "step": 6544 + }, + { + "epoch": 0.8325912733748887, + "grad_norm": 2.3861327171325684, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8908151388168335, + "num_tokens": 254311008.0, + "step": 6545 + }, + { + "epoch": 0.8327184836534792, + "grad_norm": 2.5349597930908203, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8820798397064209, + "num_tokens": 254351434.0, + "step": 6546 + }, + { + "epoch": 0.8328456939320698, + "grad_norm": 2.469414234161377, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8852478265762329, + "num_tokens": 254390709.0, + "step": 6547 + }, + { + "epoch": 0.8329729042106602, + "grad_norm": 2.7763257026672363, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8861855864524841, + "num_tokens": 254423594.0, + "step": 6548 + }, + { + "epoch": 0.8331001144892507, + "grad_norm": 2.698408603668213, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8796752691268921, + "num_tokens": 254460753.0, + "step": 6549 + }, + { + "epoch": 0.8332273247678412, + "grad_norm": 2.6561498641967773, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8904883861541748, + "num_tokens": 254493490.0, + "step": 6550 + }, + { + "epoch": 0.8333545350464318, + "grad_norm": 2.3919849395751953, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8954617977142334, + "num_tokens": 254534229.0, + "step": 6551 + }, + { + "epoch": 0.8334817453250223, + "grad_norm": 2.6020584106445312, + "learning_rate": 1e-06, + "loss": 0.4027, + "mean_token_accuracy": 0.8590108156204224, + "num_tokens": 254578294.0, + "step": 6552 + }, + { + "epoch": 0.8336089556036128, + "grad_norm": 2.5648772716522217, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.8735056519508362, + "num_tokens": 254618256.0, + "step": 6553 + }, + { + "epoch": 0.8337361658822032, + "grad_norm": 2.6008028984069824, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8772818446159363, + "num_tokens": 254657930.0, + "step": 6554 + }, + { + "epoch": 0.8338633761607938, + "grad_norm": 2.6800525188446045, + "learning_rate": 1e-06, + "loss": 0.4018, + "mean_token_accuracy": 0.8596311807632446, + "num_tokens": 254697538.0, + "step": 6555 + }, + { + "epoch": 0.8339905864393843, + "grad_norm": 2.5700523853302, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8795920610427856, + "num_tokens": 254734472.0, + "step": 6556 + }, + { + "epoch": 0.8341177967179748, + "grad_norm": 2.470348358154297, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.876238226890564, + "num_tokens": 254774346.0, + "step": 6557 + }, + { + "epoch": 0.8342450069965653, + "grad_norm": 2.6698124408721924, + "learning_rate": 1e-06, + "loss": 0.388, + "mean_token_accuracy": 0.8666664958000183, + "num_tokens": 254813635.0, + "step": 6558 + }, + { + "epoch": 0.8343722172751559, + "grad_norm": 2.5616261959075928, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8802300691604614, + "num_tokens": 254850735.0, + "step": 6559 + }, + { + "epoch": 0.8344994275537464, + "grad_norm": 2.4173760414123535, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8741836547851562, + "num_tokens": 254892813.0, + "step": 6560 + }, + { + "epoch": 0.8346266378323368, + "grad_norm": 2.5668351650238037, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8725887537002563, + "num_tokens": 254934202.0, + "step": 6561 + }, + { + "epoch": 0.8347538481109273, + "grad_norm": 2.534928560256958, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8829890489578247, + "num_tokens": 254971171.0, + "step": 6562 + }, + { + "epoch": 0.8348810583895179, + "grad_norm": 2.28539776802063, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.890662431716919, + "num_tokens": 255012511.0, + "step": 6563 + }, + { + "epoch": 0.8350082686681084, + "grad_norm": 2.725663900375366, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8840806484222412, + "num_tokens": 255046340.0, + "step": 6564 + }, + { + "epoch": 0.8351354789466989, + "grad_norm": 2.6978275775909424, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.8726109266281128, + "num_tokens": 255081863.0, + "step": 6565 + }, + { + "epoch": 0.8352626892252895, + "grad_norm": 2.769517183303833, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8787843585014343, + "num_tokens": 255116709.0, + "step": 6566 + }, + { + "epoch": 0.8353898995038799, + "grad_norm": 2.6795127391815186, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8747774362564087, + "num_tokens": 255153497.0, + "step": 6567 + }, + { + "epoch": 0.8355171097824704, + "grad_norm": 2.5473172664642334, + "learning_rate": 1e-06, + "loss": 0.3873, + "mean_token_accuracy": 0.8634772300720215, + "num_tokens": 255197998.0, + "step": 6568 + }, + { + "epoch": 0.8356443200610609, + "grad_norm": 2.638199806213379, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8861594796180725, + "num_tokens": 255233462.0, + "step": 6569 + }, + { + "epoch": 0.8357715303396515, + "grad_norm": 2.4178900718688965, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8864063024520874, + "num_tokens": 255274767.0, + "step": 6570 + }, + { + "epoch": 0.835898740618242, + "grad_norm": 2.4092578887939453, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.891767144203186, + "num_tokens": 255313756.0, + "step": 6571 + }, + { + "epoch": 0.8360259508968325, + "grad_norm": 2.468726396560669, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8847554922103882, + "num_tokens": 255355533.0, + "step": 6572 + }, + { + "epoch": 0.8361531611754229, + "grad_norm": 2.5088915824890137, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.8717867732048035, + "num_tokens": 255396376.0, + "step": 6573 + }, + { + "epoch": 0.8362803714540135, + "grad_norm": 2.5426113605499268, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8850142955780029, + "num_tokens": 255435770.0, + "step": 6574 + }, + { + "epoch": 0.836407581732604, + "grad_norm": 2.6283907890319824, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8827589750289917, + "num_tokens": 255470772.0, + "step": 6575 + }, + { + "epoch": 0.8365347920111945, + "grad_norm": 2.5408856868743896, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8729344010353088, + "num_tokens": 255509977.0, + "step": 6576 + }, + { + "epoch": 0.836662002289785, + "grad_norm": 2.5396227836608887, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8745014667510986, + "num_tokens": 255548928.0, + "step": 6577 + }, + { + "epoch": 0.8367892125683756, + "grad_norm": 2.6464669704437256, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8815882802009583, + "num_tokens": 255585835.0, + "step": 6578 + }, + { + "epoch": 0.836916422846966, + "grad_norm": 2.653669834136963, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8749549388885498, + "num_tokens": 255624663.0, + "step": 6579 + }, + { + "epoch": 0.8370436331255565, + "grad_norm": 2.520895004272461, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8814150094985962, + "num_tokens": 255664310.0, + "step": 6580 + }, + { + "epoch": 0.837170843404147, + "grad_norm": 2.585275411605835, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8803035020828247, + "num_tokens": 255700793.0, + "step": 6581 + }, + { + "epoch": 0.8372980536827376, + "grad_norm": 2.559400796890259, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8897805213928223, + "num_tokens": 255737504.0, + "step": 6582 + }, + { + "epoch": 0.8374252639613281, + "grad_norm": 2.4671523571014404, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8820252418518066, + "num_tokens": 255782954.0, + "step": 6583 + }, + { + "epoch": 0.8375524742399186, + "grad_norm": 2.4577817916870117, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.888275682926178, + "num_tokens": 255821903.0, + "step": 6584 + }, + { + "epoch": 0.837679684518509, + "grad_norm": 2.419624090194702, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8861640691757202, + "num_tokens": 255861687.0, + "step": 6585 + }, + { + "epoch": 0.8378068947970996, + "grad_norm": 2.462700605392456, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8793339729309082, + "num_tokens": 255901614.0, + "step": 6586 + }, + { + "epoch": 0.8379341050756901, + "grad_norm": 2.683593511581421, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8725985884666443, + "num_tokens": 255939589.0, + "step": 6587 + }, + { + "epoch": 0.8380613153542806, + "grad_norm": 2.7574613094329834, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8778824806213379, + "num_tokens": 255975175.0, + "step": 6588 + }, + { + "epoch": 0.8381885256328712, + "grad_norm": 2.5631041526794434, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8754651546478271, + "num_tokens": 256017635.0, + "step": 6589 + }, + { + "epoch": 0.8383157359114617, + "grad_norm": 2.717386245727539, + "learning_rate": 1e-06, + "loss": 0.3771, + "mean_token_accuracy": 0.8684264421463013, + "num_tokens": 256057928.0, + "step": 6590 + }, + { + "epoch": 0.8384429461900521, + "grad_norm": 2.634101152420044, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8780370950698853, + "num_tokens": 256094185.0, + "step": 6591 + }, + { + "epoch": 0.8385701564686426, + "grad_norm": 2.373497724533081, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8719463348388672, + "num_tokens": 256137757.0, + "step": 6592 + }, + { + "epoch": 0.8386973667472332, + "grad_norm": 2.4150938987731934, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8928934335708618, + "num_tokens": 256174739.0, + "step": 6593 + }, + { + "epoch": 0.8388245770258237, + "grad_norm": 2.6367595195770264, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8806579113006592, + "num_tokens": 256212027.0, + "step": 6594 + }, + { + "epoch": 0.8389517873044142, + "grad_norm": 2.555220603942871, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8822150230407715, + "num_tokens": 256253261.0, + "step": 6595 + }, + { + "epoch": 0.8390789975830047, + "grad_norm": 2.325047731399536, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8876014947891235, + "num_tokens": 256296540.0, + "step": 6596 + }, + { + "epoch": 0.8392062078615952, + "grad_norm": 2.607713222503662, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.8670272827148438, + "num_tokens": 256340407.0, + "step": 6597 + }, + { + "epoch": 0.8393334181401857, + "grad_norm": 2.5334713459014893, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.872109055519104, + "num_tokens": 256378675.0, + "step": 6598 + }, + { + "epoch": 0.8394606284187762, + "grad_norm": 2.5505897998809814, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8706756234169006, + "num_tokens": 256418192.0, + "step": 6599 + }, + { + "epoch": 0.8395878386973668, + "grad_norm": 2.5174295902252197, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8831056356430054, + "num_tokens": 256457115.0, + "step": 6600 + }, + { + "epoch": 0.8397150489759573, + "grad_norm": 2.7832634449005127, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8749576807022095, + "num_tokens": 256495894.0, + "step": 6601 + }, + { + "epoch": 0.8398422592545478, + "grad_norm": 2.449301242828369, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8860979080200195, + "num_tokens": 256536185.0, + "step": 6602 + }, + { + "epoch": 0.8399694695331382, + "grad_norm": 2.5285654067993164, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8731102347373962, + "num_tokens": 256580945.0, + "step": 6603 + }, + { + "epoch": 0.8400966798117288, + "grad_norm": 2.5354011058807373, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8788071870803833, + "num_tokens": 256620747.0, + "step": 6604 + }, + { + "epoch": 0.8402238900903193, + "grad_norm": 2.411829710006714, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8714241981506348, + "num_tokens": 256667127.0, + "step": 6605 + }, + { + "epoch": 0.8403511003689098, + "grad_norm": 2.6233980655670166, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8799710869789124, + "num_tokens": 256705898.0, + "step": 6606 + }, + { + "epoch": 0.8404783106475003, + "grad_norm": 2.6076018810272217, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8841114044189453, + "num_tokens": 256738693.0, + "step": 6607 + }, + { + "epoch": 0.8406055209260909, + "grad_norm": 2.6581625938415527, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8764158487319946, + "num_tokens": 256778583.0, + "step": 6608 + }, + { + "epoch": 0.8407327312046813, + "grad_norm": 2.5610804557800293, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8792708516120911, + "num_tokens": 256818448.0, + "step": 6609 + }, + { + "epoch": 0.8408599414832718, + "grad_norm": 2.7406246662139893, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8772678375244141, + "num_tokens": 256854753.0, + "step": 6610 + }, + { + "epoch": 0.8409871517618623, + "grad_norm": 2.868473529815674, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.8651963472366333, + "num_tokens": 256890038.0, + "step": 6611 + }, + { + "epoch": 0.8411143620404529, + "grad_norm": 2.8877944946289062, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8814989328384399, + "num_tokens": 256927564.0, + "step": 6612 + }, + { + "epoch": 0.8412415723190434, + "grad_norm": 2.491227149963379, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8799871206283569, + "num_tokens": 256966533.0, + "step": 6613 + }, + { + "epoch": 0.8413687825976339, + "grad_norm": 2.4778833389282227, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8852505683898926, + "num_tokens": 257004737.0, + "step": 6614 + }, + { + "epoch": 0.8414959928762245, + "grad_norm": 2.5523736476898193, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8835654258728027, + "num_tokens": 257042917.0, + "step": 6615 + }, + { + "epoch": 0.8416232031548149, + "grad_norm": 2.6894311904907227, + "learning_rate": 1e-06, + "loss": 0.3533, + "mean_token_accuracy": 0.8748969435691833, + "num_tokens": 257082331.0, + "step": 6616 + }, + { + "epoch": 0.8417504134334054, + "grad_norm": 2.861172914505005, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8811091780662537, + "num_tokens": 257115247.0, + "step": 6617 + }, + { + "epoch": 0.8418776237119959, + "grad_norm": 2.6643965244293213, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8901588916778564, + "num_tokens": 257150886.0, + "step": 6618 + }, + { + "epoch": 0.8420048339905865, + "grad_norm": 2.695460319519043, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8731483221054077, + "num_tokens": 257189030.0, + "step": 6619 + }, + { + "epoch": 0.842132044269177, + "grad_norm": 2.4104673862457275, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8845566511154175, + "num_tokens": 257227146.0, + "step": 6620 + }, + { + "epoch": 0.8422592545477675, + "grad_norm": 2.6318609714508057, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.867815375328064, + "num_tokens": 257265536.0, + "step": 6621 + }, + { + "epoch": 0.8423864648263579, + "grad_norm": 2.331026315689087, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8906313180923462, + "num_tokens": 257306344.0, + "step": 6622 + }, + { + "epoch": 0.8425136751049485, + "grad_norm": 2.6319363117218018, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8825228214263916, + "num_tokens": 257341374.0, + "step": 6623 + }, + { + "epoch": 0.842640885383539, + "grad_norm": 2.7193350791931152, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8696316480636597, + "num_tokens": 257378262.0, + "step": 6624 + }, + { + "epoch": 0.8427680956621295, + "grad_norm": 2.4613189697265625, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8823211789131165, + "num_tokens": 257420081.0, + "step": 6625 + }, + { + "epoch": 0.84289530594072, + "grad_norm": 2.6359188556671143, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8763203024864197, + "num_tokens": 257457296.0, + "step": 6626 + }, + { + "epoch": 0.8430225162193106, + "grad_norm": 2.633647918701172, + "learning_rate": 1e-06, + "loss": 0.4361, + "mean_token_accuracy": 0.8507528305053711, + "num_tokens": 257500855.0, + "step": 6627 + }, + { + "epoch": 0.843149726497901, + "grad_norm": 2.5855414867401123, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8904143571853638, + "num_tokens": 257534137.0, + "step": 6628 + }, + { + "epoch": 0.8432769367764915, + "grad_norm": 2.756176233291626, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8746001720428467, + "num_tokens": 257567837.0, + "step": 6629 + }, + { + "epoch": 0.843404147055082, + "grad_norm": 2.1980738639831543, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8895506858825684, + "num_tokens": 257613967.0, + "step": 6630 + }, + { + "epoch": 0.8435313573336726, + "grad_norm": 2.4566972255706787, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8915817737579346, + "num_tokens": 257652917.0, + "step": 6631 + }, + { + "epoch": 0.8436585676122631, + "grad_norm": 2.601259469985962, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8860111236572266, + "num_tokens": 257686454.0, + "step": 6632 + }, + { + "epoch": 0.8437857778908536, + "grad_norm": 2.350558280944824, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8800396919250488, + "num_tokens": 257730221.0, + "step": 6633 + }, + { + "epoch": 0.843912988169444, + "grad_norm": 2.7017064094543457, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8770670890808105, + "num_tokens": 257764359.0, + "step": 6634 + }, + { + "epoch": 0.8440401984480346, + "grad_norm": 2.6853818893432617, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8759777545928955, + "num_tokens": 257799405.0, + "step": 6635 + }, + { + "epoch": 0.8441674087266251, + "grad_norm": 2.361424446105957, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8772295713424683, + "num_tokens": 257847327.0, + "step": 6636 + }, + { + "epoch": 0.8442946190052156, + "grad_norm": 2.4992048740386963, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8780536651611328, + "num_tokens": 257889369.0, + "step": 6637 + }, + { + "epoch": 0.8444218292838062, + "grad_norm": 2.5401976108551025, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.875996470451355, + "num_tokens": 257932516.0, + "step": 6638 + }, + { + "epoch": 0.8445490395623967, + "grad_norm": 2.6442766189575195, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.8753126263618469, + "num_tokens": 257970974.0, + "step": 6639 + }, + { + "epoch": 0.8446762498409871, + "grad_norm": 2.5108087062835693, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8807003498077393, + "num_tokens": 258010187.0, + "step": 6640 + }, + { + "epoch": 0.8448034601195776, + "grad_norm": 2.4625144004821777, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8853706121444702, + "num_tokens": 258049748.0, + "step": 6641 + }, + { + "epoch": 0.8449306703981682, + "grad_norm": 2.4511351585388184, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8818619251251221, + "num_tokens": 258088079.0, + "step": 6642 + }, + { + "epoch": 0.8450578806767587, + "grad_norm": 2.603753089904785, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8831695318222046, + "num_tokens": 258126394.0, + "step": 6643 + }, + { + "epoch": 0.8451850909553492, + "grad_norm": 2.524526834487915, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8826442360877991, + "num_tokens": 258164735.0, + "step": 6644 + }, + { + "epoch": 0.8453123012339397, + "grad_norm": 2.4673306941986084, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8779141306877136, + "num_tokens": 258207831.0, + "step": 6645 + }, + { + "epoch": 0.8454395115125302, + "grad_norm": 2.3986423015594482, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8830749988555908, + "num_tokens": 258253992.0, + "step": 6646 + }, + { + "epoch": 0.8455667217911207, + "grad_norm": 2.556199312210083, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8741956949234009, + "num_tokens": 258293275.0, + "step": 6647 + }, + { + "epoch": 0.8456939320697112, + "grad_norm": 2.6980252265930176, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8742382526397705, + "num_tokens": 258334537.0, + "step": 6648 + }, + { + "epoch": 0.8458211423483017, + "grad_norm": 2.8266830444335938, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8767377138137817, + "num_tokens": 258366476.0, + "step": 6649 + }, + { + "epoch": 0.8459483526268923, + "grad_norm": 2.450845241546631, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8783210515975952, + "num_tokens": 258410229.0, + "step": 6650 + }, + { + "epoch": 0.8460755629054828, + "grad_norm": 2.6864428520202637, + "learning_rate": 1e-06, + "loss": 0.3943, + "mean_token_accuracy": 0.8665916919708252, + "num_tokens": 258448969.0, + "step": 6651 + }, + { + "epoch": 0.8462027731840732, + "grad_norm": 2.707322120666504, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8845621943473816, + "num_tokens": 258488246.0, + "step": 6652 + }, + { + "epoch": 0.8463299834626637, + "grad_norm": 2.696139097213745, + "learning_rate": 1e-06, + "loss": 0.4178, + "mean_token_accuracy": 0.855414628982544, + "num_tokens": 258530286.0, + "step": 6653 + }, + { + "epoch": 0.8464571937412543, + "grad_norm": 2.6342830657958984, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8637638092041016, + "num_tokens": 258570269.0, + "step": 6654 + }, + { + "epoch": 0.8465844040198448, + "grad_norm": 2.6240124702453613, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.898471474647522, + "num_tokens": 258604143.0, + "step": 6655 + }, + { + "epoch": 0.8467116142984353, + "grad_norm": 2.8098394870758057, + "learning_rate": 1e-06, + "loss": 0.3901, + "mean_token_accuracy": 0.8669180870056152, + "num_tokens": 258637726.0, + "step": 6656 + }, + { + "epoch": 0.8468388245770259, + "grad_norm": 2.4217488765716553, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8854149580001831, + "num_tokens": 258675819.0, + "step": 6657 + }, + { + "epoch": 0.8469660348556163, + "grad_norm": 2.546891450881958, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.863612949848175, + "num_tokens": 258718250.0, + "step": 6658 + }, + { + "epoch": 0.8470932451342068, + "grad_norm": 2.8881425857543945, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8685562610626221, + "num_tokens": 258758946.0, + "step": 6659 + }, + { + "epoch": 0.8472204554127973, + "grad_norm": 2.385413885116577, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8871182203292847, + "num_tokens": 258802085.0, + "step": 6660 + }, + { + "epoch": 0.8473476656913879, + "grad_norm": 2.428398609161377, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8931617736816406, + "num_tokens": 258839973.0, + "step": 6661 + }, + { + "epoch": 0.8474748759699784, + "grad_norm": 2.801919460296631, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.8692241907119751, + "num_tokens": 258874654.0, + "step": 6662 + }, + { + "epoch": 0.8476020862485689, + "grad_norm": 2.6195192337036133, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8730791807174683, + "num_tokens": 258915678.0, + "step": 6663 + }, + { + "epoch": 0.8477292965271594, + "grad_norm": 2.5389411449432373, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8803110718727112, + "num_tokens": 258955163.0, + "step": 6664 + }, + { + "epoch": 0.8478565068057499, + "grad_norm": 2.694612503051758, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8920132517814636, + "num_tokens": 258992086.0, + "step": 6665 + }, + { + "epoch": 0.8479837170843404, + "grad_norm": 2.6596944332122803, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8782281279563904, + "num_tokens": 259027819.0, + "step": 6666 + }, + { + "epoch": 0.8481109273629309, + "grad_norm": 2.726516008377075, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.880851149559021, + "num_tokens": 259063386.0, + "step": 6667 + }, + { + "epoch": 0.8482381376415215, + "grad_norm": 2.4781322479248047, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.886040210723877, + "num_tokens": 259100657.0, + "step": 6668 + }, + { + "epoch": 0.848365347920112, + "grad_norm": 2.778048515319824, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8705560564994812, + "num_tokens": 259134112.0, + "step": 6669 + }, + { + "epoch": 0.8484925581987025, + "grad_norm": 2.7035229206085205, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8818965554237366, + "num_tokens": 259168525.0, + "step": 6670 + }, + { + "epoch": 0.8486197684772929, + "grad_norm": 2.824127674102783, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8758049011230469, + "num_tokens": 259204308.0, + "step": 6671 + }, + { + "epoch": 0.8487469787558835, + "grad_norm": 2.67257022857666, + "learning_rate": 1e-06, + "loss": 0.4007, + "mean_token_accuracy": 0.8622124791145325, + "num_tokens": 259246428.0, + "step": 6672 + }, + { + "epoch": 0.848874189034474, + "grad_norm": 2.5908682346343994, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.879611611366272, + "num_tokens": 259283137.0, + "step": 6673 + }, + { + "epoch": 0.8490013993130645, + "grad_norm": 2.351219654083252, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8824433088302612, + "num_tokens": 259325693.0, + "step": 6674 + }, + { + "epoch": 0.849128609591655, + "grad_norm": 2.41347599029541, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.882423996925354, + "num_tokens": 259364706.0, + "step": 6675 + }, + { + "epoch": 0.8492558198702456, + "grad_norm": 2.466614007949829, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8854443430900574, + "num_tokens": 259405376.0, + "step": 6676 + }, + { + "epoch": 0.849383030148836, + "grad_norm": 2.5398032665252686, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8875619769096375, + "num_tokens": 259443580.0, + "step": 6677 + }, + { + "epoch": 0.8495102404274265, + "grad_norm": 2.76322865486145, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8756029009819031, + "num_tokens": 259481140.0, + "step": 6678 + }, + { + "epoch": 0.849637450706017, + "grad_norm": 2.5067763328552246, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8774049878120422, + "num_tokens": 259523250.0, + "step": 6679 + }, + { + "epoch": 0.8497646609846076, + "grad_norm": 2.4954047203063965, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8764198422431946, + "num_tokens": 259563245.0, + "step": 6680 + }, + { + "epoch": 0.8498918712631981, + "grad_norm": 2.6886560916900635, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8716946244239807, + "num_tokens": 259596764.0, + "step": 6681 + }, + { + "epoch": 0.8500190815417886, + "grad_norm": 2.523736000061035, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8738765716552734, + "num_tokens": 259635753.0, + "step": 6682 + }, + { + "epoch": 0.850146291820379, + "grad_norm": 2.748892068862915, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.8670814633369446, + "num_tokens": 259670539.0, + "step": 6683 + }, + { + "epoch": 0.8502735020989696, + "grad_norm": 2.5395147800445557, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8695611357688904, + "num_tokens": 259713714.0, + "step": 6684 + }, + { + "epoch": 0.8504007123775601, + "grad_norm": 2.6108758449554443, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8816366791725159, + "num_tokens": 259749848.0, + "step": 6685 + }, + { + "epoch": 0.8505279226561506, + "grad_norm": 2.5670039653778076, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8777129650115967, + "num_tokens": 259788275.0, + "step": 6686 + }, + { + "epoch": 0.8506551329347412, + "grad_norm": 2.5304837226867676, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8848208785057068, + "num_tokens": 259828656.0, + "step": 6687 + }, + { + "epoch": 0.8507823432133317, + "grad_norm": 2.7098090648651123, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8787318468093872, + "num_tokens": 259863366.0, + "step": 6688 + }, + { + "epoch": 0.8509095534919221, + "grad_norm": 2.492152690887451, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8820454478263855, + "num_tokens": 259902195.0, + "step": 6689 + }, + { + "epoch": 0.8510367637705126, + "grad_norm": 2.5897576808929443, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8853214979171753, + "num_tokens": 259935662.0, + "step": 6690 + }, + { + "epoch": 0.8511639740491032, + "grad_norm": 2.702881097793579, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8794822096824646, + "num_tokens": 259969320.0, + "step": 6691 + }, + { + "epoch": 0.8512911843276937, + "grad_norm": 2.596680164337158, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8797264099121094, + "num_tokens": 260008256.0, + "step": 6692 + }, + { + "epoch": 0.8514183946062842, + "grad_norm": 2.725679636001587, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.866192102432251, + "num_tokens": 260044066.0, + "step": 6693 + }, + { + "epoch": 0.8515456048848747, + "grad_norm": 2.5829050540924072, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8868973851203918, + "num_tokens": 260080764.0, + "step": 6694 + }, + { + "epoch": 0.8516728151634652, + "grad_norm": 2.717869281768799, + "learning_rate": 1e-06, + "loss": 0.3933, + "mean_token_accuracy": 0.8627702593803406, + "num_tokens": 260121258.0, + "step": 6695 + }, + { + "epoch": 0.8518000254420557, + "grad_norm": 2.516289234161377, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8707304000854492, + "num_tokens": 260164829.0, + "step": 6696 + }, + { + "epoch": 0.8519272357206462, + "grad_norm": 2.505460500717163, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8772215843200684, + "num_tokens": 260204618.0, + "step": 6697 + }, + { + "epoch": 0.8520544459992367, + "grad_norm": 2.7014222145080566, + "learning_rate": 1e-06, + "loss": 0.3889, + "mean_token_accuracy": 0.8619970679283142, + "num_tokens": 260243377.0, + "step": 6698 + }, + { + "epoch": 0.8521816562778273, + "grad_norm": 2.760692834854126, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8758402466773987, + "num_tokens": 260280468.0, + "step": 6699 + }, + { + "epoch": 0.8523088665564178, + "grad_norm": 2.77937912940979, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8734538555145264, + "num_tokens": 260314766.0, + "step": 6700 + }, + { + "epoch": 0.8524360768350082, + "grad_norm": 2.4359700679779053, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8699986934661865, + "num_tokens": 260356586.0, + "step": 6701 + }, + { + "epoch": 0.8525632871135987, + "grad_norm": 2.3848469257354736, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8808947801589966, + "num_tokens": 260396654.0, + "step": 6702 + }, + { + "epoch": 0.8526904973921893, + "grad_norm": 2.8702993392944336, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8751988410949707, + "num_tokens": 260430225.0, + "step": 6703 + }, + { + "epoch": 0.8528177076707798, + "grad_norm": 2.621364116668701, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8653022050857544, + "num_tokens": 260471951.0, + "step": 6704 + }, + { + "epoch": 0.8529449179493703, + "grad_norm": 2.543510675430298, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8887752890586853, + "num_tokens": 260508549.0, + "step": 6705 + }, + { + "epoch": 0.8530721282279609, + "grad_norm": 2.6612586975097656, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8783251643180847, + "num_tokens": 260545011.0, + "step": 6706 + }, + { + "epoch": 0.8531993385065513, + "grad_norm": 2.6114180088043213, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8907973766326904, + "num_tokens": 260578359.0, + "step": 6707 + }, + { + "epoch": 0.8533265487851418, + "grad_norm": 2.5624377727508545, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8870568871498108, + "num_tokens": 260616886.0, + "step": 6708 + }, + { + "epoch": 0.8534537590637323, + "grad_norm": 2.503190755844116, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.872830331325531, + "num_tokens": 260657707.0, + "step": 6709 + }, + { + "epoch": 0.8535809693423229, + "grad_norm": 2.6328580379486084, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8697845935821533, + "num_tokens": 260695932.0, + "step": 6710 + }, + { + "epoch": 0.8537081796209134, + "grad_norm": 2.5891902446746826, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.879802405834198, + "num_tokens": 260734652.0, + "step": 6711 + }, + { + "epoch": 0.8538353898995039, + "grad_norm": 2.4725093841552734, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8803485631942749, + "num_tokens": 260777005.0, + "step": 6712 + }, + { + "epoch": 0.8539626001780944, + "grad_norm": 2.8364617824554443, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8738387823104858, + "num_tokens": 260813578.0, + "step": 6713 + }, + { + "epoch": 0.8540898104566849, + "grad_norm": 2.5638699531555176, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.8702833652496338, + "num_tokens": 260854872.0, + "step": 6714 + }, + { + "epoch": 0.8542170207352754, + "grad_norm": 2.547454833984375, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8859215974807739, + "num_tokens": 260891986.0, + "step": 6715 + }, + { + "epoch": 0.8543442310138659, + "grad_norm": 2.7345502376556396, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8821854591369629, + "num_tokens": 260924112.0, + "step": 6716 + }, + { + "epoch": 0.8544714412924564, + "grad_norm": 2.6203629970550537, + "learning_rate": 1e-06, + "loss": 0.3894, + "mean_token_accuracy": 0.8652387857437134, + "num_tokens": 260963637.0, + "step": 6717 + }, + { + "epoch": 0.854598651571047, + "grad_norm": 2.625202178955078, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.889666736125946, + "num_tokens": 261002820.0, + "step": 6718 + }, + { + "epoch": 0.8547258618496375, + "grad_norm": 2.57928204536438, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8636840581893921, + "num_tokens": 261044413.0, + "step": 6719 + }, + { + "epoch": 0.8548530721282279, + "grad_norm": 2.6584768295288086, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8772541880607605, + "num_tokens": 261081126.0, + "step": 6720 + }, + { + "epoch": 0.8549802824068184, + "grad_norm": 2.3726460933685303, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8814501762390137, + "num_tokens": 261124832.0, + "step": 6721 + }, + { + "epoch": 0.855107492685409, + "grad_norm": 2.5479636192321777, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8938708901405334, + "num_tokens": 261158717.0, + "step": 6722 + }, + { + "epoch": 0.8552347029639995, + "grad_norm": 2.636908531188965, + "learning_rate": 1e-06, + "loss": 0.3828, + "mean_token_accuracy": 0.8665980100631714, + "num_tokens": 261200161.0, + "step": 6723 + }, + { + "epoch": 0.85536191324259, + "grad_norm": 3.1195080280303955, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8850778341293335, + "num_tokens": 261232912.0, + "step": 6724 + }, + { + "epoch": 0.8554891235211806, + "grad_norm": 2.6019694805145264, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8710212111473083, + "num_tokens": 261271573.0, + "step": 6725 + }, + { + "epoch": 0.855616333799771, + "grad_norm": 2.4872190952301025, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8774625658988953, + "num_tokens": 261316712.0, + "step": 6726 + }, + { + "epoch": 0.8557435440783615, + "grad_norm": 2.453784227371216, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8820183873176575, + "num_tokens": 261356300.0, + "step": 6727 + }, + { + "epoch": 0.855870754356952, + "grad_norm": 2.614415168762207, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8822523355484009, + "num_tokens": 261393213.0, + "step": 6728 + }, + { + "epoch": 0.8559979646355426, + "grad_norm": 2.5245306491851807, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8810233473777771, + "num_tokens": 261433685.0, + "step": 6729 + }, + { + "epoch": 0.8561251749141331, + "grad_norm": 2.5173864364624023, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.868289589881897, + "num_tokens": 261475536.0, + "step": 6730 + }, + { + "epoch": 0.8562523851927236, + "grad_norm": 2.5555765628814697, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8855392932891846, + "num_tokens": 261512579.0, + "step": 6731 + }, + { + "epoch": 0.856379595471314, + "grad_norm": 2.5447893142700195, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8712755441665649, + "num_tokens": 261552247.0, + "step": 6732 + }, + { + "epoch": 0.8565068057499046, + "grad_norm": 2.7621424198150635, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8776096701622009, + "num_tokens": 261587013.0, + "step": 6733 + }, + { + "epoch": 0.8566340160284951, + "grad_norm": 2.7154462337493896, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8761602640151978, + "num_tokens": 261623937.0, + "step": 6734 + }, + { + "epoch": 0.8567612263070856, + "grad_norm": 2.6909685134887695, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8695738315582275, + "num_tokens": 261663897.0, + "step": 6735 + }, + { + "epoch": 0.8568884365856761, + "grad_norm": 2.68310284614563, + "learning_rate": 1e-06, + "loss": 0.4176, + "mean_token_accuracy": 0.854864239692688, + "num_tokens": 261701288.0, + "step": 6736 + }, + { + "epoch": 0.8570156468642667, + "grad_norm": 2.800014019012451, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8680163025856018, + "num_tokens": 261736468.0, + "step": 6737 + }, + { + "epoch": 0.8571428571428571, + "grad_norm": 2.6925203800201416, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8805971741676331, + "num_tokens": 261771399.0, + "step": 6738 + }, + { + "epoch": 0.8572700674214476, + "grad_norm": 2.6694281101226807, + "learning_rate": 1e-06, + "loss": 0.3904, + "mean_token_accuracy": 0.8688468933105469, + "num_tokens": 261812378.0, + "step": 6739 + }, + { + "epoch": 0.8573972777000382, + "grad_norm": 2.410733222961426, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8828176856040955, + "num_tokens": 261854106.0, + "step": 6740 + }, + { + "epoch": 0.8575244879786287, + "grad_norm": 2.6453208923339844, + "learning_rate": 1e-06, + "loss": 0.3521, + "mean_token_accuracy": 0.8790321350097656, + "num_tokens": 261893443.0, + "step": 6741 + }, + { + "epoch": 0.8576516982572192, + "grad_norm": 2.7418975830078125, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8719905018806458, + "num_tokens": 261928394.0, + "step": 6742 + }, + { + "epoch": 0.8577789085358097, + "grad_norm": 2.7615106105804443, + "learning_rate": 1e-06, + "loss": 0.3983, + "mean_token_accuracy": 0.8592497110366821, + "num_tokens": 261968619.0, + "step": 6743 + }, + { + "epoch": 0.8579061188144002, + "grad_norm": 2.514291763305664, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8608627319335938, + "num_tokens": 262016339.0, + "step": 6744 + }, + { + "epoch": 0.8580333290929907, + "grad_norm": 2.5949952602386475, + "learning_rate": 1e-06, + "loss": 0.3692, + "mean_token_accuracy": 0.8705841898918152, + "num_tokens": 262059587.0, + "step": 6745 + }, + { + "epoch": 0.8581605393715812, + "grad_norm": 2.4973082542419434, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8766720294952393, + "num_tokens": 262098989.0, + "step": 6746 + }, + { + "epoch": 0.8582877496501717, + "grad_norm": 2.5792620182037354, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8852546215057373, + "num_tokens": 262141358.0, + "step": 6747 + }, + { + "epoch": 0.8584149599287623, + "grad_norm": 2.6219446659088135, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8848362565040588, + "num_tokens": 262176356.0, + "step": 6748 + }, + { + "epoch": 0.8585421702073528, + "grad_norm": 2.9247612953186035, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8757070899009705, + "num_tokens": 262210232.0, + "step": 6749 + }, + { + "epoch": 0.8586693804859432, + "grad_norm": 2.6680212020874023, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.88093101978302, + "num_tokens": 262247511.0, + "step": 6750 + }, + { + "epoch": 0.8587965907645337, + "grad_norm": 2.4882678985595703, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8794461488723755, + "num_tokens": 262289730.0, + "step": 6751 + }, + { + "epoch": 0.8589238010431243, + "grad_norm": 2.3872644901275635, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8957126140594482, + "num_tokens": 262329928.0, + "step": 6752 + }, + { + "epoch": 0.8590510113217148, + "grad_norm": 2.419147491455078, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8806190490722656, + "num_tokens": 262370081.0, + "step": 6753 + }, + { + "epoch": 0.8591782216003053, + "grad_norm": 2.6252353191375732, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8835285305976868, + "num_tokens": 262406025.0, + "step": 6754 + }, + { + "epoch": 0.8593054318788959, + "grad_norm": 2.4385862350463867, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8982268571853638, + "num_tokens": 262442665.0, + "step": 6755 + }, + { + "epoch": 0.8594326421574863, + "grad_norm": 2.624481678009033, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8896360993385315, + "num_tokens": 262476277.0, + "step": 6756 + }, + { + "epoch": 0.8595598524360768, + "grad_norm": 2.5957343578338623, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8680933117866516, + "num_tokens": 262514905.0, + "step": 6757 + }, + { + "epoch": 0.8596870627146673, + "grad_norm": 2.4687857627868652, + "learning_rate": 1e-06, + "loss": 0.3518, + "mean_token_accuracy": 0.8758275508880615, + "num_tokens": 262555483.0, + "step": 6758 + }, + { + "epoch": 0.8598142729932579, + "grad_norm": 2.4756641387939453, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8851020336151123, + "num_tokens": 262594777.0, + "step": 6759 + }, + { + "epoch": 0.8599414832718484, + "grad_norm": 2.41599702835083, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8892266750335693, + "num_tokens": 262632141.0, + "step": 6760 + }, + { + "epoch": 0.8600686935504389, + "grad_norm": 2.6019344329833984, + "learning_rate": 1e-06, + "loss": 0.366, + "mean_token_accuracy": 0.8726673722267151, + "num_tokens": 262675335.0, + "step": 6761 + }, + { + "epoch": 0.8601959038290294, + "grad_norm": 2.580965995788574, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8889943957328796, + "num_tokens": 262709530.0, + "step": 6762 + }, + { + "epoch": 0.8603231141076199, + "grad_norm": 2.9522812366485596, + "learning_rate": 1e-06, + "loss": 0.3829, + "mean_token_accuracy": 0.8660482168197632, + "num_tokens": 262740622.0, + "step": 6763 + }, + { + "epoch": 0.8604503243862104, + "grad_norm": 2.5663421154022217, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8840749263763428, + "num_tokens": 262779248.0, + "step": 6764 + }, + { + "epoch": 0.8605775346648009, + "grad_norm": 2.4758810997009277, + "learning_rate": 1e-06, + "loss": 0.3745, + "mean_token_accuracy": 0.872294545173645, + "num_tokens": 262825168.0, + "step": 6765 + }, + { + "epoch": 0.8607047449433914, + "grad_norm": 2.3032312393188477, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8981238603591919, + "num_tokens": 262866365.0, + "step": 6766 + }, + { + "epoch": 0.860831955221982, + "grad_norm": 2.7227060794830322, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8859323263168335, + "num_tokens": 262901118.0, + "step": 6767 + }, + { + "epoch": 0.8609591655005725, + "grad_norm": 2.509352207183838, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8857605457305908, + "num_tokens": 262941783.0, + "step": 6768 + }, + { + "epoch": 0.8610863757791629, + "grad_norm": 2.5561952590942383, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8986189961433411, + "num_tokens": 262977823.0, + "step": 6769 + }, + { + "epoch": 0.8612135860577534, + "grad_norm": 2.505856513977051, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8769037127494812, + "num_tokens": 263018496.0, + "step": 6770 + }, + { + "epoch": 0.861340796336344, + "grad_norm": 2.4806387424468994, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8874109983444214, + "num_tokens": 263060345.0, + "step": 6771 + }, + { + "epoch": 0.8614680066149345, + "grad_norm": 2.6132829189300537, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8875724077224731, + "num_tokens": 263093914.0, + "step": 6772 + }, + { + "epoch": 0.861595216893525, + "grad_norm": 2.4011476039886475, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8816252946853638, + "num_tokens": 263135284.0, + "step": 6773 + }, + { + "epoch": 0.8617224271721156, + "grad_norm": 2.4347071647644043, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8836714029312134, + "num_tokens": 263174706.0, + "step": 6774 + }, + { + "epoch": 0.861849637450706, + "grad_norm": 2.6955626010894775, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8790731430053711, + "num_tokens": 263213840.0, + "step": 6775 + }, + { + "epoch": 0.8619768477292965, + "grad_norm": 2.4250409603118896, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8786497712135315, + "num_tokens": 263257659.0, + "step": 6776 + }, + { + "epoch": 0.862104058007887, + "grad_norm": 2.4925177097320557, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.890312910079956, + "num_tokens": 263297907.0, + "step": 6777 + }, + { + "epoch": 0.8622312682864776, + "grad_norm": 2.417928695678711, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8800855875015259, + "num_tokens": 263341587.0, + "step": 6778 + }, + { + "epoch": 0.8623584785650681, + "grad_norm": 2.404283046722412, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8871171474456787, + "num_tokens": 263382023.0, + "step": 6779 + }, + { + "epoch": 0.8624856888436586, + "grad_norm": 2.814519166946411, + "learning_rate": 1e-06, + "loss": 0.3883, + "mean_token_accuracy": 0.8668532371520996, + "num_tokens": 263422358.0, + "step": 6780 + }, + { + "epoch": 0.862612899122249, + "grad_norm": 2.500044822692871, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8970177173614502, + "num_tokens": 263457932.0, + "step": 6781 + }, + { + "epoch": 0.8627401094008396, + "grad_norm": 2.5106444358825684, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8841903209686279, + "num_tokens": 263496402.0, + "step": 6782 + }, + { + "epoch": 0.8628673196794301, + "grad_norm": 2.9310812950134277, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8914565443992615, + "num_tokens": 263525392.0, + "step": 6783 + }, + { + "epoch": 0.8629945299580206, + "grad_norm": 2.5743770599365234, + "learning_rate": 1e-06, + "loss": 0.3897, + "mean_token_accuracy": 0.8612993955612183, + "num_tokens": 263567453.0, + "step": 6784 + }, + { + "epoch": 0.8631217402366111, + "grad_norm": 2.4312586784362793, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8740218877792358, + "num_tokens": 263613209.0, + "step": 6785 + }, + { + "epoch": 0.8632489505152017, + "grad_norm": 2.7022645473480225, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8858896493911743, + "num_tokens": 263648758.0, + "step": 6786 + }, + { + "epoch": 0.8633761607937921, + "grad_norm": 2.6106042861938477, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8828511834144592, + "num_tokens": 263692283.0, + "step": 6787 + }, + { + "epoch": 0.8635033710723826, + "grad_norm": 2.6349856853485107, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8756518363952637, + "num_tokens": 263729100.0, + "step": 6788 + }, + { + "epoch": 0.8636305813509731, + "grad_norm": 2.761941909790039, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8825429081916809, + "num_tokens": 263762270.0, + "step": 6789 + }, + { + "epoch": 0.8637577916295637, + "grad_norm": 2.4196078777313232, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8774129748344421, + "num_tokens": 263803370.0, + "step": 6790 + }, + { + "epoch": 0.8638850019081542, + "grad_norm": 2.6548428535461426, + "learning_rate": 1e-06, + "loss": 0.3686, + "mean_token_accuracy": 0.8726232051849365, + "num_tokens": 263839793.0, + "step": 6791 + }, + { + "epoch": 0.8640122121867447, + "grad_norm": 2.595555305480957, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8880473971366882, + "num_tokens": 263879907.0, + "step": 6792 + }, + { + "epoch": 0.8641394224653351, + "grad_norm": 2.408076524734497, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8864467144012451, + "num_tokens": 263921515.0, + "step": 6793 + }, + { + "epoch": 0.8642666327439257, + "grad_norm": 2.5617456436157227, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8798049092292786, + "num_tokens": 263958166.0, + "step": 6794 + }, + { + "epoch": 0.8643938430225162, + "grad_norm": 2.5586888790130615, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8781386017799377, + "num_tokens": 263999941.0, + "step": 6795 + }, + { + "epoch": 0.8645210533011067, + "grad_norm": 2.497964382171631, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8862396478652954, + "num_tokens": 264036724.0, + "step": 6796 + }, + { + "epoch": 0.8646482635796973, + "grad_norm": 2.826791763305664, + "learning_rate": 1e-06, + "loss": 0.3823, + "mean_token_accuracy": 0.8640391826629639, + "num_tokens": 264071714.0, + "step": 6797 + }, + { + "epoch": 0.8647754738582878, + "grad_norm": 2.5717976093292236, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.8696476221084595, + "num_tokens": 264112216.0, + "step": 6798 + }, + { + "epoch": 0.8649026841368782, + "grad_norm": 2.691720962524414, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8753723502159119, + "num_tokens": 264147640.0, + "step": 6799 + }, + { + "epoch": 0.8650298944154687, + "grad_norm": 2.6166584491729736, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8857474327087402, + "num_tokens": 264185828.0, + "step": 6800 + }, + { + "epoch": 0.8651571046940593, + "grad_norm": 2.7255537509918213, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8762326240539551, + "num_tokens": 264221453.0, + "step": 6801 + }, + { + "epoch": 0.8652843149726498, + "grad_norm": 2.376498222351074, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8791259527206421, + "num_tokens": 264268311.0, + "step": 6802 + }, + { + "epoch": 0.8654115252512403, + "grad_norm": 2.7800190448760986, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8834679126739502, + "num_tokens": 264301417.0, + "step": 6803 + }, + { + "epoch": 0.8655387355298308, + "grad_norm": 2.4837374687194824, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8877589106559753, + "num_tokens": 264336555.0, + "step": 6804 + }, + { + "epoch": 0.8656659458084213, + "grad_norm": 2.428997278213501, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8919163942337036, + "num_tokens": 264372481.0, + "step": 6805 + }, + { + "epoch": 0.8657931560870118, + "grad_norm": 2.7341489791870117, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8789840340614319, + "num_tokens": 264407128.0, + "step": 6806 + }, + { + "epoch": 0.8659203663656023, + "grad_norm": 2.8542094230651855, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8685125112533569, + "num_tokens": 264443004.0, + "step": 6807 + }, + { + "epoch": 0.8660475766441929, + "grad_norm": 2.3752079010009766, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8896753191947937, + "num_tokens": 264486623.0, + "step": 6808 + }, + { + "epoch": 0.8661747869227834, + "grad_norm": 2.645597457885742, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8713428378105164, + "num_tokens": 264523413.0, + "step": 6809 + }, + { + "epoch": 0.8663019972013739, + "grad_norm": 2.7969515323638916, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8777784109115601, + "num_tokens": 264555829.0, + "step": 6810 + }, + { + "epoch": 0.8664292074799644, + "grad_norm": 2.688394546508789, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8900988101959229, + "num_tokens": 264587878.0, + "step": 6811 + }, + { + "epoch": 0.8665564177585549, + "grad_norm": 2.5183956623077393, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8746544718742371, + "num_tokens": 264632635.0, + "step": 6812 + }, + { + "epoch": 0.8666836280371454, + "grad_norm": 2.4225759506225586, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8790528774261475, + "num_tokens": 264675891.0, + "step": 6813 + }, + { + "epoch": 0.8668108383157359, + "grad_norm": 2.5442519187927246, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8695654273033142, + "num_tokens": 264718124.0, + "step": 6814 + }, + { + "epoch": 0.8669380485943264, + "grad_norm": 2.2499992847442627, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8946425318717957, + "num_tokens": 264761715.0, + "step": 6815 + }, + { + "epoch": 0.867065258872917, + "grad_norm": 2.775881290435791, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.88018399477005, + "num_tokens": 264793166.0, + "step": 6816 + }, + { + "epoch": 0.8671924691515075, + "grad_norm": 3.067615509033203, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8771752119064331, + "num_tokens": 264825783.0, + "step": 6817 + }, + { + "epoch": 0.8673196794300979, + "grad_norm": 2.43406343460083, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8862327337265015, + "num_tokens": 264866481.0, + "step": 6818 + }, + { + "epoch": 0.8674468897086884, + "grad_norm": 2.764110565185547, + "learning_rate": 1e-06, + "loss": 0.3877, + "mean_token_accuracy": 0.863551676273346, + "num_tokens": 264900992.0, + "step": 6819 + }, + { + "epoch": 0.867574099987279, + "grad_norm": 2.7666079998016357, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.873823881149292, + "num_tokens": 264934529.0, + "step": 6820 + }, + { + "epoch": 0.8677013102658695, + "grad_norm": 2.494601011276245, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8899821043014526, + "num_tokens": 264973785.0, + "step": 6821 + }, + { + "epoch": 0.86782852054446, + "grad_norm": 2.688811779022217, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.868991494178772, + "num_tokens": 265010906.0, + "step": 6822 + }, + { + "epoch": 0.8679557308230506, + "grad_norm": 2.6591198444366455, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8789899349212646, + "num_tokens": 265046718.0, + "step": 6823 + }, + { + "epoch": 0.868082941101641, + "grad_norm": 2.2778425216674805, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8950744271278381, + "num_tokens": 265088523.0, + "step": 6824 + }, + { + "epoch": 0.8682101513802315, + "grad_norm": 2.4976067543029785, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8838760852813721, + "num_tokens": 265130316.0, + "step": 6825 + }, + { + "epoch": 0.868337361658822, + "grad_norm": 2.6356635093688965, + "learning_rate": 1e-06, + "loss": 0.3817, + "mean_token_accuracy": 0.8642106652259827, + "num_tokens": 265170823.0, + "step": 6826 + }, + { + "epoch": 0.8684645719374126, + "grad_norm": 2.792501926422119, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8694196939468384, + "num_tokens": 265204429.0, + "step": 6827 + }, + { + "epoch": 0.8685917822160031, + "grad_norm": 2.528027296066284, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8756375312805176, + "num_tokens": 265245135.0, + "step": 6828 + }, + { + "epoch": 0.8687189924945936, + "grad_norm": 2.3638856410980225, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8888425230979919, + "num_tokens": 265286086.0, + "step": 6829 + }, + { + "epoch": 0.868846202773184, + "grad_norm": 2.412036418914795, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8731794953346252, + "num_tokens": 265328387.0, + "step": 6830 + }, + { + "epoch": 0.8689734130517746, + "grad_norm": 3.1288747787475586, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8713776469230652, + "num_tokens": 265356441.0, + "step": 6831 + }, + { + "epoch": 0.8691006233303651, + "grad_norm": 2.191904306411743, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8916369080543518, + "num_tokens": 265400748.0, + "step": 6832 + }, + { + "epoch": 0.8692278336089556, + "grad_norm": 2.646920919418335, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8880068063735962, + "num_tokens": 265433254.0, + "step": 6833 + }, + { + "epoch": 0.8693550438875461, + "grad_norm": 2.5690343379974365, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8746776580810547, + "num_tokens": 265474528.0, + "step": 6834 + }, + { + "epoch": 0.8694822541661367, + "grad_norm": 2.3634212017059326, + "learning_rate": 1e-06, + "loss": 0.3846, + "mean_token_accuracy": 0.8651777505874634, + "num_tokens": 265522938.0, + "step": 6835 + }, + { + "epoch": 0.8696094644447271, + "grad_norm": 2.5860204696655273, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.878966212272644, + "num_tokens": 265559970.0, + "step": 6836 + }, + { + "epoch": 0.8697366747233176, + "grad_norm": 2.431199550628662, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8830949068069458, + "num_tokens": 265603414.0, + "step": 6837 + }, + { + "epoch": 0.8698638850019081, + "grad_norm": 2.7051897048950195, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8725854158401489, + "num_tokens": 265638844.0, + "step": 6838 + }, + { + "epoch": 0.8699910952804987, + "grad_norm": 2.6407790184020996, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.870208740234375, + "num_tokens": 265674177.0, + "step": 6839 + }, + { + "epoch": 0.8701183055590892, + "grad_norm": 2.3850152492523193, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8889715075492859, + "num_tokens": 265714563.0, + "step": 6840 + }, + { + "epoch": 0.8702455158376797, + "grad_norm": 2.760148286819458, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.8737631440162659, + "num_tokens": 265754815.0, + "step": 6841 + }, + { + "epoch": 0.8703727261162701, + "grad_norm": 2.467625141143799, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8828247785568237, + "num_tokens": 265793985.0, + "step": 6842 + }, + { + "epoch": 0.8704999363948607, + "grad_norm": 2.4520163536071777, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8741940855979919, + "num_tokens": 265833477.0, + "step": 6843 + }, + { + "epoch": 0.8706271466734512, + "grad_norm": 2.354762315750122, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8812528848648071, + "num_tokens": 265876821.0, + "step": 6844 + }, + { + "epoch": 0.8707543569520417, + "grad_norm": 2.439765691757202, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8835102915763855, + "num_tokens": 265914803.0, + "step": 6845 + }, + { + "epoch": 0.8708815672306323, + "grad_norm": 2.6328749656677246, + "learning_rate": 1e-06, + "loss": 0.4054, + "mean_token_accuracy": 0.85783851146698, + "num_tokens": 265956616.0, + "step": 6846 + }, + { + "epoch": 0.8710087775092228, + "grad_norm": 2.9041454792022705, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8872058391571045, + "num_tokens": 265984938.0, + "step": 6847 + }, + { + "epoch": 0.8711359877878132, + "grad_norm": 2.4942080974578857, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.891046404838562, + "num_tokens": 266018710.0, + "step": 6848 + }, + { + "epoch": 0.8712631980664037, + "grad_norm": 2.5866880416870117, + "learning_rate": 1e-06, + "loss": 0.3613, + "mean_token_accuracy": 0.8741140961647034, + "num_tokens": 266056849.0, + "step": 6849 + }, + { + "epoch": 0.8713904083449943, + "grad_norm": 2.5366618633270264, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8751234412193298, + "num_tokens": 266097850.0, + "step": 6850 + }, + { + "epoch": 0.8715176186235848, + "grad_norm": 2.475623369216919, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.8791535496711731, + "num_tokens": 266138533.0, + "step": 6851 + }, + { + "epoch": 0.8716448289021753, + "grad_norm": 2.351248264312744, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8838719129562378, + "num_tokens": 266180570.0, + "step": 6852 + }, + { + "epoch": 0.8717720391807658, + "grad_norm": 3.1306955814361572, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8862469792366028, + "num_tokens": 266221023.0, + "step": 6853 + }, + { + "epoch": 0.8718992494593563, + "grad_norm": 2.467193365097046, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8814965486526489, + "num_tokens": 266257069.0, + "step": 6854 + }, + { + "epoch": 0.8720264597379468, + "grad_norm": 2.457336187362671, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8946588635444641, + "num_tokens": 266292609.0, + "step": 6855 + }, + { + "epoch": 0.8721536700165373, + "grad_norm": 2.733025312423706, + "learning_rate": 1e-06, + "loss": 0.3936, + "mean_token_accuracy": 0.8643665313720703, + "num_tokens": 266332862.0, + "step": 6856 + }, + { + "epoch": 0.8722808802951278, + "grad_norm": 2.651615858078003, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8670915365219116, + "num_tokens": 266370725.0, + "step": 6857 + }, + { + "epoch": 0.8724080905737184, + "grad_norm": 2.763763666152954, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8814015984535217, + "num_tokens": 266410418.0, + "step": 6858 + }, + { + "epoch": 0.8725353008523089, + "grad_norm": 2.496778964996338, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8771852254867554, + "num_tokens": 266452589.0, + "step": 6859 + }, + { + "epoch": 0.8726625111308994, + "grad_norm": 2.5209386348724365, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8802317976951599, + "num_tokens": 266493616.0, + "step": 6860 + }, + { + "epoch": 0.8727897214094898, + "grad_norm": 2.4923150539398193, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8822256326675415, + "num_tokens": 266533444.0, + "step": 6861 + }, + { + "epoch": 0.8729169316880804, + "grad_norm": 2.4407501220703125, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8969444632530212, + "num_tokens": 266570127.0, + "step": 6862 + }, + { + "epoch": 0.8730441419666709, + "grad_norm": 2.5330755710601807, + "learning_rate": 1e-06, + "loss": 0.3636, + "mean_token_accuracy": 0.8734540939331055, + "num_tokens": 266614592.0, + "step": 6863 + }, + { + "epoch": 0.8731713522452614, + "grad_norm": 2.613851308822632, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8803409337997437, + "num_tokens": 266651284.0, + "step": 6864 + }, + { + "epoch": 0.873298562523852, + "grad_norm": 2.897735118865967, + "learning_rate": 1e-06, + "loss": 0.3793, + "mean_token_accuracy": 0.8637194633483887, + "num_tokens": 266686061.0, + "step": 6865 + }, + { + "epoch": 0.8734257728024425, + "grad_norm": 2.4741148948669434, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8752296566963196, + "num_tokens": 266726922.0, + "step": 6866 + }, + { + "epoch": 0.8735529830810329, + "grad_norm": 2.5975661277770996, + "learning_rate": 1e-06, + "loss": 0.3735, + "mean_token_accuracy": 0.8715323209762573, + "num_tokens": 266764507.0, + "step": 6867 + }, + { + "epoch": 0.8736801933596234, + "grad_norm": 2.756133794784546, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8732485771179199, + "num_tokens": 266801675.0, + "step": 6868 + }, + { + "epoch": 0.873807403638214, + "grad_norm": 2.642148733139038, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8677256107330322, + "num_tokens": 266841548.0, + "step": 6869 + }, + { + "epoch": 0.8739346139168045, + "grad_norm": 2.447270393371582, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8837472796440125, + "num_tokens": 266886011.0, + "step": 6870 + }, + { + "epoch": 0.874061824195395, + "grad_norm": 2.288213014602661, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8824892044067383, + "num_tokens": 266930199.0, + "step": 6871 + }, + { + "epoch": 0.8741890344739855, + "grad_norm": 2.560788631439209, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8838074207305908, + "num_tokens": 266965646.0, + "step": 6872 + }, + { + "epoch": 0.874316244752576, + "grad_norm": 2.761772871017456, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8718209266662598, + "num_tokens": 267000969.0, + "step": 6873 + }, + { + "epoch": 0.8744434550311665, + "grad_norm": 2.460897207260132, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8874351978302002, + "num_tokens": 267038505.0, + "step": 6874 + }, + { + "epoch": 0.874570665309757, + "grad_norm": 2.7515716552734375, + "learning_rate": 1e-06, + "loss": 0.3866, + "mean_token_accuracy": 0.8680932521820068, + "num_tokens": 267076001.0, + "step": 6875 + }, + { + "epoch": 0.8746978755883476, + "grad_norm": 2.472442388534546, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8932172656059265, + "num_tokens": 267114049.0, + "step": 6876 + }, + { + "epoch": 0.8748250858669381, + "grad_norm": 2.4753479957580566, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8759769201278687, + "num_tokens": 267156386.0, + "step": 6877 + }, + { + "epoch": 0.8749522961455286, + "grad_norm": 2.531959056854248, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.873161792755127, + "num_tokens": 267198952.0, + "step": 6878 + }, + { + "epoch": 0.875079506424119, + "grad_norm": 2.5061614513397217, + "learning_rate": 1e-06, + "loss": 0.3606, + "mean_token_accuracy": 0.8723082542419434, + "num_tokens": 267240577.0, + "step": 6879 + }, + { + "epoch": 0.8752067167027096, + "grad_norm": 2.6396188735961914, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8761356472969055, + "num_tokens": 267274837.0, + "step": 6880 + }, + { + "epoch": 0.8753339269813001, + "grad_norm": 2.5326614379882812, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8817353248596191, + "num_tokens": 267317546.0, + "step": 6881 + }, + { + "epoch": 0.8754611372598906, + "grad_norm": 2.665489435195923, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8884649276733398, + "num_tokens": 267352866.0, + "step": 6882 + }, + { + "epoch": 0.8755883475384811, + "grad_norm": 2.4065494537353516, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8746047019958496, + "num_tokens": 267396383.0, + "step": 6883 + }, + { + "epoch": 0.8757155578170717, + "grad_norm": 2.5951032638549805, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.885820746421814, + "num_tokens": 267433793.0, + "step": 6884 + }, + { + "epoch": 0.8758427680956621, + "grad_norm": 2.682314395904541, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8763963580131531, + "num_tokens": 267467939.0, + "step": 6885 + }, + { + "epoch": 0.8759699783742526, + "grad_norm": 2.5098519325256348, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8735582828521729, + "num_tokens": 267511184.0, + "step": 6886 + }, + { + "epoch": 0.8760971886528431, + "grad_norm": 2.566774368286133, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8845206499099731, + "num_tokens": 267548982.0, + "step": 6887 + }, + { + "epoch": 0.8762243989314337, + "grad_norm": 2.388627767562866, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8889896869659424, + "num_tokens": 267589982.0, + "step": 6888 + }, + { + "epoch": 0.8763516092100242, + "grad_norm": 2.4248287677764893, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8921890258789062, + "num_tokens": 267628017.0, + "step": 6889 + }, + { + "epoch": 0.8764788194886147, + "grad_norm": 2.5058624744415283, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8730944395065308, + "num_tokens": 267667567.0, + "step": 6890 + }, + { + "epoch": 0.8766060297672051, + "grad_norm": 2.6488986015319824, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8786098957061768, + "num_tokens": 267703199.0, + "step": 6891 + }, + { + "epoch": 0.8767332400457957, + "grad_norm": 2.7097387313842773, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8832728862762451, + "num_tokens": 267737573.0, + "step": 6892 + }, + { + "epoch": 0.8768604503243862, + "grad_norm": 2.5702500343322754, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8843150734901428, + "num_tokens": 267775171.0, + "step": 6893 + }, + { + "epoch": 0.8769876606029767, + "grad_norm": 2.4347245693206787, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.888330340385437, + "num_tokens": 267814727.0, + "step": 6894 + }, + { + "epoch": 0.8771148708815673, + "grad_norm": 2.473147392272949, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.890200138092041, + "num_tokens": 267852714.0, + "step": 6895 + }, + { + "epoch": 0.8772420811601578, + "grad_norm": 2.50744366645813, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.879758358001709, + "num_tokens": 267894017.0, + "step": 6896 + }, + { + "epoch": 0.8773692914387482, + "grad_norm": 2.526740550994873, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8830603361129761, + "num_tokens": 267932671.0, + "step": 6897 + }, + { + "epoch": 0.8774965017173387, + "grad_norm": 2.627279281616211, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8753232955932617, + "num_tokens": 267972297.0, + "step": 6898 + }, + { + "epoch": 0.8776237119959293, + "grad_norm": 2.532315969467163, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8867454528808594, + "num_tokens": 268010563.0, + "step": 6899 + }, + { + "epoch": 0.8777509222745198, + "grad_norm": 2.391850471496582, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8916468024253845, + "num_tokens": 268052105.0, + "step": 6900 + }, + { + "epoch": 0.8778781325531103, + "grad_norm": 2.6797943115234375, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8769893646240234, + "num_tokens": 268088890.0, + "step": 6901 + }, + { + "epoch": 0.8780053428317008, + "grad_norm": 2.7563884258270264, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8745330572128296, + "num_tokens": 268124270.0, + "step": 6902 + }, + { + "epoch": 0.8781325531102913, + "grad_norm": 2.7770557403564453, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8726576566696167, + "num_tokens": 268159559.0, + "step": 6903 + }, + { + "epoch": 0.8782597633888818, + "grad_norm": 2.4928624629974365, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8911285400390625, + "num_tokens": 268199956.0, + "step": 6904 + }, + { + "epoch": 0.8783869736674723, + "grad_norm": 2.8088221549987793, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.8732567429542542, + "num_tokens": 268236750.0, + "step": 6905 + }, + { + "epoch": 0.8785141839460628, + "grad_norm": 2.6092429161071777, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8774487972259521, + "num_tokens": 268275580.0, + "step": 6906 + }, + { + "epoch": 0.8786413942246534, + "grad_norm": 2.606997489929199, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8850746750831604, + "num_tokens": 268310458.0, + "step": 6907 + }, + { + "epoch": 0.8787686045032439, + "grad_norm": 2.4151833057403564, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8937628269195557, + "num_tokens": 268348599.0, + "step": 6908 + }, + { + "epoch": 0.8788958147818343, + "grad_norm": 2.4204587936401367, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8783439993858337, + "num_tokens": 268393452.0, + "step": 6909 + }, + { + "epoch": 0.8790230250604248, + "grad_norm": 2.596811294555664, + "learning_rate": 1e-06, + "loss": 0.3975, + "mean_token_accuracy": 0.8593733310699463, + "num_tokens": 268434097.0, + "step": 6910 + }, + { + "epoch": 0.8791502353390154, + "grad_norm": 2.777334690093994, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8667739629745483, + "num_tokens": 268468527.0, + "step": 6911 + }, + { + "epoch": 0.8792774456176059, + "grad_norm": 2.937034845352173, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8845019936561584, + "num_tokens": 268500653.0, + "step": 6912 + }, + { + "epoch": 0.8794046558961964, + "grad_norm": 2.4398300647735596, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8745396733283997, + "num_tokens": 268543764.0, + "step": 6913 + }, + { + "epoch": 0.879531866174787, + "grad_norm": 2.492077112197876, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8844503164291382, + "num_tokens": 268582839.0, + "step": 6914 + }, + { + "epoch": 0.8796590764533775, + "grad_norm": 2.7329986095428467, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.8712910413742065, + "num_tokens": 268618604.0, + "step": 6915 + }, + { + "epoch": 0.8797862867319679, + "grad_norm": 2.360769510269165, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8940991759300232, + "num_tokens": 268660427.0, + "step": 6916 + }, + { + "epoch": 0.8799134970105584, + "grad_norm": 2.5781702995300293, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.878098726272583, + "num_tokens": 268699727.0, + "step": 6917 + }, + { + "epoch": 0.880040707289149, + "grad_norm": 2.495720386505127, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8861136436462402, + "num_tokens": 268737252.0, + "step": 6918 + }, + { + "epoch": 0.8801679175677395, + "grad_norm": 2.5001883506774902, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8839479088783264, + "num_tokens": 268775264.0, + "step": 6919 + }, + { + "epoch": 0.88029512784633, + "grad_norm": 2.355963706970215, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8844277858734131, + "num_tokens": 268819120.0, + "step": 6920 + }, + { + "epoch": 0.8804223381249205, + "grad_norm": 2.567126750946045, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.884589672088623, + "num_tokens": 268857335.0, + "step": 6921 + }, + { + "epoch": 0.880549548403511, + "grad_norm": 2.6542301177978516, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8669896721839905, + "num_tokens": 268893616.0, + "step": 6922 + }, + { + "epoch": 0.8806767586821015, + "grad_norm": 2.5685558319091797, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8712458610534668, + "num_tokens": 268932682.0, + "step": 6923 + }, + { + "epoch": 0.880803968960692, + "grad_norm": 2.553215980529785, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.884115993976593, + "num_tokens": 268973745.0, + "step": 6924 + }, + { + "epoch": 0.8809311792392825, + "grad_norm": 2.628887891769409, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8929746150970459, + "num_tokens": 269010722.0, + "step": 6925 + }, + { + "epoch": 0.8810583895178731, + "grad_norm": 2.6650712490081787, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8821853399276733, + "num_tokens": 269044127.0, + "step": 6926 + }, + { + "epoch": 0.8811855997964636, + "grad_norm": 2.337406873703003, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8811588883399963, + "num_tokens": 269088952.0, + "step": 6927 + }, + { + "epoch": 0.881312810075054, + "grad_norm": 2.7803761959075928, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8858115673065186, + "num_tokens": 269121284.0, + "step": 6928 + }, + { + "epoch": 0.8814400203536445, + "grad_norm": 2.6265616416931152, + "learning_rate": 1e-06, + "loss": 0.3924, + "mean_token_accuracy": 0.8634933829307556, + "num_tokens": 269163352.0, + "step": 6929 + }, + { + "epoch": 0.8815672306322351, + "grad_norm": 2.824359178543091, + "learning_rate": 1e-06, + "loss": 0.3755, + "mean_token_accuracy": 0.868995189666748, + "num_tokens": 269199122.0, + "step": 6930 + }, + { + "epoch": 0.8816944409108256, + "grad_norm": 2.524874687194824, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8866714239120483, + "num_tokens": 269236846.0, + "step": 6931 + }, + { + "epoch": 0.8818216511894161, + "grad_norm": 2.709770679473877, + "learning_rate": 1e-06, + "loss": 0.3709, + "mean_token_accuracy": 0.8667932748794556, + "num_tokens": 269274406.0, + "step": 6932 + }, + { + "epoch": 0.8819488614680067, + "grad_norm": 2.6697094440460205, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8767191171646118, + "num_tokens": 269312502.0, + "step": 6933 + }, + { + "epoch": 0.8820760717465971, + "grad_norm": 2.8434908390045166, + "learning_rate": 1e-06, + "loss": 0.3701, + "mean_token_accuracy": 0.8723753094673157, + "num_tokens": 269349182.0, + "step": 6934 + }, + { + "epoch": 0.8822032820251876, + "grad_norm": 2.5467820167541504, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8820533752441406, + "num_tokens": 269388558.0, + "step": 6935 + }, + { + "epoch": 0.8823304923037781, + "grad_norm": 2.6491751670837402, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8735796213150024, + "num_tokens": 269429471.0, + "step": 6936 + }, + { + "epoch": 0.8824577025823687, + "grad_norm": 2.383545398712158, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.8746439814567566, + "num_tokens": 269474600.0, + "step": 6937 + }, + { + "epoch": 0.8825849128609592, + "grad_norm": 2.828031539916992, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8672351837158203, + "num_tokens": 269508779.0, + "step": 6938 + }, + { + "epoch": 0.8827121231395497, + "grad_norm": 2.448206901550293, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8878133296966553, + "num_tokens": 269548426.0, + "step": 6939 + }, + { + "epoch": 0.8828393334181401, + "grad_norm": 2.8872556686401367, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.878075361251831, + "num_tokens": 269581552.0, + "step": 6940 + }, + { + "epoch": 0.8829665436967307, + "grad_norm": 2.6226933002471924, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8884257674217224, + "num_tokens": 269617898.0, + "step": 6941 + }, + { + "epoch": 0.8830937539753212, + "grad_norm": 2.3649654388427734, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8756993412971497, + "num_tokens": 269663533.0, + "step": 6942 + }, + { + "epoch": 0.8832209642539117, + "grad_norm": 2.611130714416504, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.866468071937561, + "num_tokens": 269705269.0, + "step": 6943 + }, + { + "epoch": 0.8833481745325023, + "grad_norm": 2.507749557495117, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8856937289237976, + "num_tokens": 269742919.0, + "step": 6944 + }, + { + "epoch": 0.8834753848110928, + "grad_norm": 2.6016345024108887, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8812611699104309, + "num_tokens": 269781247.0, + "step": 6945 + }, + { + "epoch": 0.8836025950896832, + "grad_norm": 2.6533162593841553, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8746927976608276, + "num_tokens": 269822854.0, + "step": 6946 + }, + { + "epoch": 0.8837298053682737, + "grad_norm": 2.44394850730896, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8909459114074707, + "num_tokens": 269862688.0, + "step": 6947 + }, + { + "epoch": 0.8838570156468643, + "grad_norm": 2.811347723007202, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.861633837223053, + "num_tokens": 269899895.0, + "step": 6948 + }, + { + "epoch": 0.8839842259254548, + "grad_norm": 2.578799247741699, + "learning_rate": 1e-06, + "loss": 0.3641, + "mean_token_accuracy": 0.8683350086212158, + "num_tokens": 269937671.0, + "step": 6949 + }, + { + "epoch": 0.8841114362040453, + "grad_norm": 2.525531768798828, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8801283240318298, + "num_tokens": 269974171.0, + "step": 6950 + }, + { + "epoch": 0.8842386464826358, + "grad_norm": 2.435915946960449, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8763419389724731, + "num_tokens": 270016805.0, + "step": 6951 + }, + { + "epoch": 0.8843658567612263, + "grad_norm": 2.5565311908721924, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8907277584075928, + "num_tokens": 270056112.0, + "step": 6952 + }, + { + "epoch": 0.8844930670398168, + "grad_norm": 2.5540125370025635, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8762768507003784, + "num_tokens": 270095371.0, + "step": 6953 + }, + { + "epoch": 0.8846202773184073, + "grad_norm": 2.768864393234253, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.875056266784668, + "num_tokens": 270131739.0, + "step": 6954 + }, + { + "epoch": 0.8847474875969978, + "grad_norm": 2.458407163619995, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8809207081794739, + "num_tokens": 270172221.0, + "step": 6955 + }, + { + "epoch": 0.8848746978755884, + "grad_norm": 2.4981746673583984, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8841551542282104, + "num_tokens": 270208634.0, + "step": 6956 + }, + { + "epoch": 0.8850019081541789, + "grad_norm": 2.4371261596679688, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8867020606994629, + "num_tokens": 270250217.0, + "step": 6957 + }, + { + "epoch": 0.8851291184327693, + "grad_norm": 2.520913600921631, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.878819465637207, + "num_tokens": 270289907.0, + "step": 6958 + }, + { + "epoch": 0.8852563287113598, + "grad_norm": 2.5898327827453613, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8786503672599792, + "num_tokens": 270328944.0, + "step": 6959 + }, + { + "epoch": 0.8853835389899504, + "grad_norm": 2.6415629386901855, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8776199221611023, + "num_tokens": 270367615.0, + "step": 6960 + }, + { + "epoch": 0.8855107492685409, + "grad_norm": 2.6768717765808105, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8839084506034851, + "num_tokens": 270405002.0, + "step": 6961 + }, + { + "epoch": 0.8856379595471314, + "grad_norm": 2.6530990600585938, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8699317574501038, + "num_tokens": 270441687.0, + "step": 6962 + }, + { + "epoch": 0.885765169825722, + "grad_norm": 2.586029529571533, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8949894905090332, + "num_tokens": 270474250.0, + "step": 6963 + }, + { + "epoch": 0.8858923801043125, + "grad_norm": 2.7019286155700684, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8787349462509155, + "num_tokens": 270511611.0, + "step": 6964 + }, + { + "epoch": 0.8860195903829029, + "grad_norm": 2.7212557792663574, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.867098331451416, + "num_tokens": 270550466.0, + "step": 6965 + }, + { + "epoch": 0.8861468006614934, + "grad_norm": 2.7667548656463623, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8762441277503967, + "num_tokens": 270583867.0, + "step": 6966 + }, + { + "epoch": 0.886274010940084, + "grad_norm": 2.7561631202697754, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8751910924911499, + "num_tokens": 270618205.0, + "step": 6967 + }, + { + "epoch": 0.8864012212186745, + "grad_norm": 2.6837031841278076, + "learning_rate": 1e-06, + "loss": 0.3765, + "mean_token_accuracy": 0.8746377229690552, + "num_tokens": 270659214.0, + "step": 6968 + }, + { + "epoch": 0.886528431497265, + "grad_norm": 2.583855152130127, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8844829201698303, + "num_tokens": 270693746.0, + "step": 6969 + }, + { + "epoch": 0.8866556417758555, + "grad_norm": 2.7008049488067627, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.881880521774292, + "num_tokens": 270730367.0, + "step": 6970 + }, + { + "epoch": 0.886782852054446, + "grad_norm": 2.269519090652466, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8852963447570801, + "num_tokens": 270775469.0, + "step": 6971 + }, + { + "epoch": 0.8869100623330365, + "grad_norm": 2.7048375606536865, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8685877919197083, + "num_tokens": 270813850.0, + "step": 6972 + }, + { + "epoch": 0.887037272611627, + "grad_norm": 2.369926929473877, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8913649916648865, + "num_tokens": 270853671.0, + "step": 6973 + }, + { + "epoch": 0.8871644828902175, + "grad_norm": 2.6109859943389893, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8666452765464783, + "num_tokens": 270892894.0, + "step": 6974 + }, + { + "epoch": 0.8872916931688081, + "grad_norm": 2.535823106765747, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8775284290313721, + "num_tokens": 270932684.0, + "step": 6975 + }, + { + "epoch": 0.8874189034473986, + "grad_norm": 2.4777486324310303, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8772149085998535, + "num_tokens": 270970916.0, + "step": 6976 + }, + { + "epoch": 0.887546113725989, + "grad_norm": 2.600489854812622, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8888930082321167, + "num_tokens": 271010452.0, + "step": 6977 + }, + { + "epoch": 0.8876733240045795, + "grad_norm": 2.857058048248291, + "learning_rate": 1e-06, + "loss": 0.3767, + "mean_token_accuracy": 0.8664237260818481, + "num_tokens": 271044767.0, + "step": 6978 + }, + { + "epoch": 0.8878005342831701, + "grad_norm": 2.614344835281372, + "learning_rate": 1e-06, + "loss": 0.3827, + "mean_token_accuracy": 0.8687846660614014, + "num_tokens": 271084227.0, + "step": 6979 + }, + { + "epoch": 0.8879277445617606, + "grad_norm": 2.599551200866699, + "learning_rate": 1e-06, + "loss": 0.3887, + "mean_token_accuracy": 0.8666394948959351, + "num_tokens": 271123299.0, + "step": 6980 + }, + { + "epoch": 0.8880549548403511, + "grad_norm": 3.0182995796203613, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.876692533493042, + "num_tokens": 271152823.0, + "step": 6981 + }, + { + "epoch": 0.8881821651189417, + "grad_norm": 2.4998273849487305, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.8648868799209595, + "num_tokens": 271202329.0, + "step": 6982 + }, + { + "epoch": 0.8883093753975321, + "grad_norm": 2.7004809379577637, + "learning_rate": 1e-06, + "loss": 0.3791, + "mean_token_accuracy": 0.870185375213623, + "num_tokens": 271244051.0, + "step": 6983 + }, + { + "epoch": 0.8884365856761226, + "grad_norm": 2.4733493328094482, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8787950277328491, + "num_tokens": 271285318.0, + "step": 6984 + }, + { + "epoch": 0.8885637959547131, + "grad_norm": 2.4709744453430176, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8809437155723572, + "num_tokens": 271325369.0, + "step": 6985 + }, + { + "epoch": 0.8886910062333037, + "grad_norm": 2.623162269592285, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8772951364517212, + "num_tokens": 271361404.0, + "step": 6986 + }, + { + "epoch": 0.8888182165118942, + "grad_norm": 2.4185984134674072, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.87201327085495, + "num_tokens": 271406420.0, + "step": 6987 + }, + { + "epoch": 0.8889454267904847, + "grad_norm": 2.5857396125793457, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8718282580375671, + "num_tokens": 271450131.0, + "step": 6988 + }, + { + "epoch": 0.8890726370690751, + "grad_norm": 2.5800912380218506, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.882536768913269, + "num_tokens": 271488465.0, + "step": 6989 + }, + { + "epoch": 0.8891998473476657, + "grad_norm": 2.6915087699890137, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8704105615615845, + "num_tokens": 271527627.0, + "step": 6990 + }, + { + "epoch": 0.8893270576262562, + "grad_norm": 2.41117262840271, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8824069499969482, + "num_tokens": 271570498.0, + "step": 6991 + }, + { + "epoch": 0.8894542679048467, + "grad_norm": 2.6552321910858154, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.874783992767334, + "num_tokens": 271606939.0, + "step": 6992 + }, + { + "epoch": 0.8895814781834372, + "grad_norm": 2.831817626953125, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8704019784927368, + "num_tokens": 271642600.0, + "step": 6993 + }, + { + "epoch": 0.8897086884620278, + "grad_norm": 2.758227825164795, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8767447471618652, + "num_tokens": 271681171.0, + "step": 6994 + }, + { + "epoch": 0.8898358987406182, + "grad_norm": 2.708482027053833, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.865861177444458, + "num_tokens": 271716470.0, + "step": 6995 + }, + { + "epoch": 0.8899631090192087, + "grad_norm": 2.4837136268615723, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.890274167060852, + "num_tokens": 271753339.0, + "step": 6996 + }, + { + "epoch": 0.8900903192977992, + "grad_norm": 2.5169310569763184, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8728047013282776, + "num_tokens": 271795947.0, + "step": 6997 + }, + { + "epoch": 0.8902175295763898, + "grad_norm": 2.523284673690796, + "learning_rate": 1e-06, + "loss": 0.3806, + "mean_token_accuracy": 0.8660295605659485, + "num_tokens": 271841329.0, + "step": 6998 + }, + { + "epoch": 0.8903447398549803, + "grad_norm": 2.393004894256592, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8833948373794556, + "num_tokens": 271882160.0, + "step": 6999 + }, + { + "epoch": 0.8904719501335708, + "grad_norm": 2.801622152328491, + "learning_rate": 1e-06, + "loss": 0.428, + "mean_token_accuracy": 0.8483622670173645, + "num_tokens": 271924216.0, + "step": 7000 + }, + { + "epoch": 0.8905991604121613, + "grad_norm": 2.7753429412841797, + "learning_rate": 1e-06, + "loss": 0.3844, + "mean_token_accuracy": 0.8671410083770752, + "num_tokens": 271959630.0, + "step": 7001 + }, + { + "epoch": 0.8907263706907518, + "grad_norm": 2.618177652359009, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8664120435714722, + "num_tokens": 272003582.0, + "step": 7002 + }, + { + "epoch": 0.8908535809693423, + "grad_norm": 2.7286505699157715, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8836065530776978, + "num_tokens": 272035958.0, + "step": 7003 + }, + { + "epoch": 0.8909807912479328, + "grad_norm": 2.8592047691345215, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8834432363510132, + "num_tokens": 272068632.0, + "step": 7004 + }, + { + "epoch": 0.8911080015265234, + "grad_norm": 2.431199550628662, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8771481513977051, + "num_tokens": 272110045.0, + "step": 7005 + }, + { + "epoch": 0.8912352118051139, + "grad_norm": 2.6231353282928467, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8805831670761108, + "num_tokens": 272145226.0, + "step": 7006 + }, + { + "epoch": 0.8913624220837043, + "grad_norm": 2.4140725135803223, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8928835988044739, + "num_tokens": 272185504.0, + "step": 7007 + }, + { + "epoch": 0.8914896323622948, + "grad_norm": 2.6359968185424805, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.873888373374939, + "num_tokens": 272221930.0, + "step": 7008 + }, + { + "epoch": 0.8916168426408854, + "grad_norm": 2.6064326763153076, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8838109970092773, + "num_tokens": 272257928.0, + "step": 7009 + }, + { + "epoch": 0.8917440529194759, + "grad_norm": 2.7929751873016357, + "learning_rate": 1e-06, + "loss": 0.3681, + "mean_token_accuracy": 0.868942141532898, + "num_tokens": 272292574.0, + "step": 7010 + }, + { + "epoch": 0.8918712631980664, + "grad_norm": 2.3941967487335205, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8906877040863037, + "num_tokens": 272330069.0, + "step": 7011 + }, + { + "epoch": 0.891998473476657, + "grad_norm": 2.5584986209869385, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8869791626930237, + "num_tokens": 272370273.0, + "step": 7012 + }, + { + "epoch": 0.8921256837552475, + "grad_norm": 2.6533312797546387, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8704309463500977, + "num_tokens": 272405641.0, + "step": 7013 + }, + { + "epoch": 0.8922528940338379, + "grad_norm": 2.5900063514709473, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8803249597549438, + "num_tokens": 272444754.0, + "step": 7014 + }, + { + "epoch": 0.8923801043124284, + "grad_norm": 2.8041746616363525, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8727891445159912, + "num_tokens": 272479852.0, + "step": 7015 + }, + { + "epoch": 0.892507314591019, + "grad_norm": 2.66188645362854, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8717472553253174, + "num_tokens": 272520785.0, + "step": 7016 + }, + { + "epoch": 0.8926345248696095, + "grad_norm": 2.600505828857422, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.879229724407196, + "num_tokens": 272556681.0, + "step": 7017 + }, + { + "epoch": 0.8927617351482, + "grad_norm": 2.6215288639068604, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8889816403388977, + "num_tokens": 272592862.0, + "step": 7018 + }, + { + "epoch": 0.8928889454267905, + "grad_norm": 2.54885196685791, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8740332126617432, + "num_tokens": 272631836.0, + "step": 7019 + }, + { + "epoch": 0.893016155705381, + "grad_norm": 2.6122026443481445, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.878628671169281, + "num_tokens": 272672039.0, + "step": 7020 + }, + { + "epoch": 0.8931433659839715, + "grad_norm": 2.773977279663086, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.8658804297447205, + "num_tokens": 272705408.0, + "step": 7021 + }, + { + "epoch": 0.893270576262562, + "grad_norm": 2.628046751022339, + "learning_rate": 1e-06, + "loss": 0.3779, + "mean_token_accuracy": 0.8700942993164062, + "num_tokens": 272747844.0, + "step": 7022 + }, + { + "epoch": 0.8933977865411525, + "grad_norm": 2.5100786685943604, + "learning_rate": 1e-06, + "loss": 0.3763, + "mean_token_accuracy": 0.8696203231811523, + "num_tokens": 272791567.0, + "step": 7023 + }, + { + "epoch": 0.8935249968197431, + "grad_norm": 2.4322361946105957, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8781411647796631, + "num_tokens": 272832978.0, + "step": 7024 + }, + { + "epoch": 0.8936522070983336, + "grad_norm": 2.6947267055511475, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8866676688194275, + "num_tokens": 272868804.0, + "step": 7025 + }, + { + "epoch": 0.893779417376924, + "grad_norm": 2.4277944564819336, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8882340788841248, + "num_tokens": 272906621.0, + "step": 7026 + }, + { + "epoch": 0.8939066276555145, + "grad_norm": 2.452418088912964, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8840093612670898, + "num_tokens": 272944331.0, + "step": 7027 + }, + { + "epoch": 0.8940338379341051, + "grad_norm": 2.636660099029541, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8819018006324768, + "num_tokens": 272978546.0, + "step": 7028 + }, + { + "epoch": 0.8941610482126956, + "grad_norm": 2.479022264480591, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8824353218078613, + "num_tokens": 273020193.0, + "step": 7029 + }, + { + "epoch": 0.8942882584912861, + "grad_norm": 2.6271042823791504, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8831541538238525, + "num_tokens": 273056131.0, + "step": 7030 + }, + { + "epoch": 0.8944154687698767, + "grad_norm": 2.7174322605133057, + "learning_rate": 1e-06, + "loss": 0.3957, + "mean_token_accuracy": 0.8604806065559387, + "num_tokens": 273097914.0, + "step": 7031 + }, + { + "epoch": 0.8945426790484671, + "grad_norm": 2.717996835708618, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8690604567527771, + "num_tokens": 273137693.0, + "step": 7032 + }, + { + "epoch": 0.8946698893270576, + "grad_norm": 2.4610755443573, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.876059889793396, + "num_tokens": 273178462.0, + "step": 7033 + }, + { + "epoch": 0.8947970996056481, + "grad_norm": 2.7799344062805176, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8706026077270508, + "num_tokens": 273213908.0, + "step": 7034 + }, + { + "epoch": 0.8949243098842387, + "grad_norm": 2.753199338912964, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8806948661804199, + "num_tokens": 273248918.0, + "step": 7035 + }, + { + "epoch": 0.8950515201628292, + "grad_norm": 2.720064163208008, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8910430669784546, + "num_tokens": 273280111.0, + "step": 7036 + }, + { + "epoch": 0.8951787304414197, + "grad_norm": 2.513491153717041, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8771469593048096, + "num_tokens": 273317149.0, + "step": 7037 + }, + { + "epoch": 0.8953059407200101, + "grad_norm": 2.9652678966522217, + "learning_rate": 1e-06, + "loss": 0.4103, + "mean_token_accuracy": 0.8570084571838379, + "num_tokens": 273356388.0, + "step": 7038 + }, + { + "epoch": 0.8954331509986007, + "grad_norm": 2.40071964263916, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8769998550415039, + "num_tokens": 273399269.0, + "step": 7039 + }, + { + "epoch": 0.8955603612771912, + "grad_norm": 2.8127996921539307, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8749710321426392, + "num_tokens": 273433227.0, + "step": 7040 + }, + { + "epoch": 0.8956875715557817, + "grad_norm": 2.6281449794769287, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8803298473358154, + "num_tokens": 273476596.0, + "step": 7041 + }, + { + "epoch": 0.8958147818343722, + "grad_norm": 2.4383304119110107, + "learning_rate": 1e-06, + "loss": 0.3671, + "mean_token_accuracy": 0.8714030385017395, + "num_tokens": 273520105.0, + "step": 7042 + }, + { + "epoch": 0.8959419921129628, + "grad_norm": 2.301518678665161, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8898061513900757, + "num_tokens": 273562484.0, + "step": 7043 + }, + { + "epoch": 0.8960692023915532, + "grad_norm": 2.4154422283172607, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8877614736557007, + "num_tokens": 273601642.0, + "step": 7044 + }, + { + "epoch": 0.8961964126701437, + "grad_norm": 2.4620585441589355, + "learning_rate": 1e-06, + "loss": 0.393, + "mean_token_accuracy": 0.8634458780288696, + "num_tokens": 273645648.0, + "step": 7045 + }, + { + "epoch": 0.8963236229487342, + "grad_norm": 2.422808885574341, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8872047066688538, + "num_tokens": 273683564.0, + "step": 7046 + }, + { + "epoch": 0.8964508332273248, + "grad_norm": 2.3121206760406494, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8988842368125916, + "num_tokens": 273721158.0, + "step": 7047 + }, + { + "epoch": 0.8965780435059153, + "grad_norm": 2.645904302597046, + "learning_rate": 1e-06, + "loss": 0.396, + "mean_token_accuracy": 0.8612492084503174, + "num_tokens": 273759677.0, + "step": 7048 + }, + { + "epoch": 0.8967052537845058, + "grad_norm": 2.3214638233184814, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8843894004821777, + "num_tokens": 273805807.0, + "step": 7049 + }, + { + "epoch": 0.8968324640630962, + "grad_norm": 2.7063958644866943, + "learning_rate": 1e-06, + "loss": 0.3683, + "mean_token_accuracy": 0.8709326982498169, + "num_tokens": 273840480.0, + "step": 7050 + }, + { + "epoch": 0.8969596743416868, + "grad_norm": 2.623382806777954, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8817813396453857, + "num_tokens": 273878482.0, + "step": 7051 + }, + { + "epoch": 0.8970868846202773, + "grad_norm": 2.7927491664886475, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8831945061683655, + "num_tokens": 273916598.0, + "step": 7052 + }, + { + "epoch": 0.8972140948988678, + "grad_norm": 2.5811991691589355, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.877738356590271, + "num_tokens": 273954094.0, + "step": 7053 + }, + { + "epoch": 0.8973413051774584, + "grad_norm": 2.307645559310913, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8884933590888977, + "num_tokens": 273997071.0, + "step": 7054 + }, + { + "epoch": 0.8974685154560489, + "grad_norm": 2.667297124862671, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8665028214454651, + "num_tokens": 274034706.0, + "step": 7055 + }, + { + "epoch": 0.8975957257346393, + "grad_norm": 2.4077250957489014, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8847224116325378, + "num_tokens": 274078125.0, + "step": 7056 + }, + { + "epoch": 0.8977229360132298, + "grad_norm": 2.663388967514038, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.889246940612793, + "num_tokens": 274112270.0, + "step": 7057 + }, + { + "epoch": 0.8978501462918204, + "grad_norm": 2.611894130706787, + "learning_rate": 1e-06, + "loss": 0.3587, + "mean_token_accuracy": 0.8733563423156738, + "num_tokens": 274152392.0, + "step": 7058 + }, + { + "epoch": 0.8979773565704109, + "grad_norm": 2.670494556427002, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8884909749031067, + "num_tokens": 274188580.0, + "step": 7059 + }, + { + "epoch": 0.8981045668490014, + "grad_norm": 2.4642927646636963, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.899154782295227, + "num_tokens": 274229394.0, + "step": 7060 + }, + { + "epoch": 0.898231777127592, + "grad_norm": 2.55092716217041, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8720558881759644, + "num_tokens": 274270783.0, + "step": 7061 + }, + { + "epoch": 0.8983589874061825, + "grad_norm": 2.581057548522949, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8859450817108154, + "num_tokens": 274307161.0, + "step": 7062 + }, + { + "epoch": 0.8984861976847729, + "grad_norm": 2.4167141914367676, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8708542585372925, + "num_tokens": 274349885.0, + "step": 7063 + }, + { + "epoch": 0.8986134079633634, + "grad_norm": 2.511709690093994, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8866897225379944, + "num_tokens": 274387311.0, + "step": 7064 + }, + { + "epoch": 0.898740618241954, + "grad_norm": 2.4394733905792236, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8827679753303528, + "num_tokens": 274423719.0, + "step": 7065 + }, + { + "epoch": 0.8988678285205445, + "grad_norm": 2.6346747875213623, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8779150247573853, + "num_tokens": 274463642.0, + "step": 7066 + }, + { + "epoch": 0.898995038799135, + "grad_norm": 2.7633047103881836, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8879963159561157, + "num_tokens": 274495089.0, + "step": 7067 + }, + { + "epoch": 0.8991222490777255, + "grad_norm": 2.509193181991577, + "learning_rate": 1e-06, + "loss": 0.3544, + "mean_token_accuracy": 0.8755055069923401, + "num_tokens": 274533789.0, + "step": 7068 + }, + { + "epoch": 0.899249459356316, + "grad_norm": 2.5821447372436523, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8761698007583618, + "num_tokens": 274571187.0, + "step": 7069 + }, + { + "epoch": 0.8993766696349065, + "grad_norm": 2.5523743629455566, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8817736506462097, + "num_tokens": 274608331.0, + "step": 7070 + }, + { + "epoch": 0.899503879913497, + "grad_norm": 2.382378101348877, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8758766651153564, + "num_tokens": 274652722.0, + "step": 7071 + }, + { + "epoch": 0.8996310901920875, + "grad_norm": 2.521165132522583, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8804750442504883, + "num_tokens": 274692824.0, + "step": 7072 + }, + { + "epoch": 0.8997583004706781, + "grad_norm": 2.627898693084717, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8850506544113159, + "num_tokens": 274729260.0, + "step": 7073 + }, + { + "epoch": 0.8998855107492686, + "grad_norm": 2.6272644996643066, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8741592764854431, + "num_tokens": 274767802.0, + "step": 7074 + }, + { + "epoch": 0.900012721027859, + "grad_norm": 2.5232114791870117, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9008534550666809, + "num_tokens": 274803704.0, + "step": 7075 + }, + { + "epoch": 0.9001399313064495, + "grad_norm": 2.901864528656006, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.8716102242469788, + "num_tokens": 274835808.0, + "step": 7076 + }, + { + "epoch": 0.9002671415850401, + "grad_norm": 2.650604009628296, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8797156810760498, + "num_tokens": 274875471.0, + "step": 7077 + }, + { + "epoch": 0.9003943518636306, + "grad_norm": 2.52868914604187, + "learning_rate": 1e-06, + "loss": 0.3741, + "mean_token_accuracy": 0.867742657661438, + "num_tokens": 274917693.0, + "step": 7078 + }, + { + "epoch": 0.9005215621422211, + "grad_norm": 2.7137763500213623, + "learning_rate": 1e-06, + "loss": 0.379, + "mean_token_accuracy": 0.8693870306015015, + "num_tokens": 274955418.0, + "step": 7079 + }, + { + "epoch": 0.9006487724208116, + "grad_norm": 2.5257835388183594, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8846757411956787, + "num_tokens": 274992992.0, + "step": 7080 + }, + { + "epoch": 0.9007759826994021, + "grad_norm": 2.441869020462036, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8896216154098511, + "num_tokens": 275029462.0, + "step": 7081 + }, + { + "epoch": 0.9009031929779926, + "grad_norm": 2.5273263454437256, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.885546088218689, + "num_tokens": 275069093.0, + "step": 7082 + }, + { + "epoch": 0.9010304032565831, + "grad_norm": 2.4031944274902344, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8790677785873413, + "num_tokens": 275111193.0, + "step": 7083 + }, + { + "epoch": 0.9011576135351737, + "grad_norm": 2.7471134662628174, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.880930483341217, + "num_tokens": 275144516.0, + "step": 7084 + }, + { + "epoch": 0.9012848238137642, + "grad_norm": 2.741718292236328, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8864497542381287, + "num_tokens": 275182271.0, + "step": 7085 + }, + { + "epoch": 0.9014120340923547, + "grad_norm": 2.686558246612549, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8848568797111511, + "num_tokens": 275221431.0, + "step": 7086 + }, + { + "epoch": 0.9015392443709451, + "grad_norm": 2.514897108078003, + "learning_rate": 1e-06, + "loss": 0.3726, + "mean_token_accuracy": 0.8690900802612305, + "num_tokens": 275261590.0, + "step": 7087 + }, + { + "epoch": 0.9016664546495357, + "grad_norm": 2.535598039627075, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8756691813468933, + "num_tokens": 275301207.0, + "step": 7088 + }, + { + "epoch": 0.9017936649281262, + "grad_norm": 2.5982043743133545, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8857548236846924, + "num_tokens": 275336735.0, + "step": 7089 + }, + { + "epoch": 0.9019208752067167, + "grad_norm": 2.6704671382904053, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8856512308120728, + "num_tokens": 275371890.0, + "step": 7090 + }, + { + "epoch": 0.9020480854853072, + "grad_norm": 2.5030808448791504, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8853225111961365, + "num_tokens": 275413947.0, + "step": 7091 + }, + { + "epoch": 0.9021752957638978, + "grad_norm": 2.541248321533203, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8906558752059937, + "num_tokens": 275454231.0, + "step": 7092 + }, + { + "epoch": 0.9023025060424882, + "grad_norm": 2.4808099269866943, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8785738945007324, + "num_tokens": 275496254.0, + "step": 7093 + }, + { + "epoch": 0.9024297163210787, + "grad_norm": 2.857661247253418, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.8725716471672058, + "num_tokens": 275531636.0, + "step": 7094 + }, + { + "epoch": 0.9025569265996692, + "grad_norm": 2.4716360569000244, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.896526575088501, + "num_tokens": 275567640.0, + "step": 7095 + }, + { + "epoch": 0.9026841368782598, + "grad_norm": 2.37519907951355, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8880032300949097, + "num_tokens": 275610687.0, + "step": 7096 + }, + { + "epoch": 0.9028113471568503, + "grad_norm": 2.3606908321380615, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8946127891540527, + "num_tokens": 275649200.0, + "step": 7097 + }, + { + "epoch": 0.9029385574354408, + "grad_norm": 2.5979270935058594, + "learning_rate": 1e-06, + "loss": 0.3596, + "mean_token_accuracy": 0.875542402267456, + "num_tokens": 275686945.0, + "step": 7098 + }, + { + "epoch": 0.9030657677140312, + "grad_norm": 2.3908536434173584, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8932806253433228, + "num_tokens": 275726267.0, + "step": 7099 + }, + { + "epoch": 0.9031929779926218, + "grad_norm": 2.509589672088623, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8808054327964783, + "num_tokens": 275764940.0, + "step": 7100 + }, + { + "epoch": 0.9033201882712123, + "grad_norm": 2.346069812774658, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8933357000350952, + "num_tokens": 275804552.0, + "step": 7101 + }, + { + "epoch": 0.9034473985498028, + "grad_norm": 2.5982112884521484, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8774878978729248, + "num_tokens": 275840425.0, + "step": 7102 + }, + { + "epoch": 0.9035746088283934, + "grad_norm": 2.5691404342651367, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.879381000995636, + "num_tokens": 275876795.0, + "step": 7103 + }, + { + "epoch": 0.9037018191069839, + "grad_norm": 2.5476393699645996, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8785876631736755, + "num_tokens": 275915858.0, + "step": 7104 + }, + { + "epoch": 0.9038290293855743, + "grad_norm": 2.717996120452881, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8776727914810181, + "num_tokens": 275950725.0, + "step": 7105 + }, + { + "epoch": 0.9039562396641648, + "grad_norm": 2.563699960708618, + "learning_rate": 1e-06, + "loss": 0.3842, + "mean_token_accuracy": 0.8686006665229797, + "num_tokens": 275990377.0, + "step": 7106 + }, + { + "epoch": 0.9040834499427554, + "grad_norm": 2.6304266452789307, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8845332860946655, + "num_tokens": 276027035.0, + "step": 7107 + }, + { + "epoch": 0.9042106602213459, + "grad_norm": 2.3896396160125732, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8706971406936646, + "num_tokens": 276071381.0, + "step": 7108 + }, + { + "epoch": 0.9043378704999364, + "grad_norm": 2.728800058364868, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8762612342834473, + "num_tokens": 276105297.0, + "step": 7109 + }, + { + "epoch": 0.9044650807785269, + "grad_norm": 2.548459529876709, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8771786689758301, + "num_tokens": 276141219.0, + "step": 7110 + }, + { + "epoch": 0.9045922910571175, + "grad_norm": 2.8349149227142334, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.879317581653595, + "num_tokens": 276173083.0, + "step": 7111 + }, + { + "epoch": 0.9047195013357079, + "grad_norm": 2.532618284225464, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8790991306304932, + "num_tokens": 276211838.0, + "step": 7112 + }, + { + "epoch": 0.9048467116142984, + "grad_norm": 2.5517160892486572, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8774697780609131, + "num_tokens": 276252641.0, + "step": 7113 + }, + { + "epoch": 0.9049739218928889, + "grad_norm": 2.2087559700012207, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.895158052444458, + "num_tokens": 276297097.0, + "step": 7114 + }, + { + "epoch": 0.9051011321714795, + "grad_norm": 2.7852845191955566, + "learning_rate": 1e-06, + "loss": 0.3853, + "mean_token_accuracy": 0.86491858959198, + "num_tokens": 276332583.0, + "step": 7115 + }, + { + "epoch": 0.90522834245007, + "grad_norm": 2.673910617828369, + "learning_rate": 1e-06, + "loss": 0.4072, + "mean_token_accuracy": 0.8579661846160889, + "num_tokens": 276375420.0, + "step": 7116 + }, + { + "epoch": 0.9053555527286605, + "grad_norm": 2.4842069149017334, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8752427101135254, + "num_tokens": 276417938.0, + "step": 7117 + }, + { + "epoch": 0.905482763007251, + "grad_norm": 2.4391088485717773, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8840823173522949, + "num_tokens": 276459543.0, + "step": 7118 + }, + { + "epoch": 0.9056099732858415, + "grad_norm": 2.4499592781066895, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8808851838111877, + "num_tokens": 276501369.0, + "step": 7119 + }, + { + "epoch": 0.905737183564432, + "grad_norm": 2.5455048084259033, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8847114443778992, + "num_tokens": 276542479.0, + "step": 7120 + }, + { + "epoch": 0.9058643938430225, + "grad_norm": 2.6555328369140625, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8830530643463135, + "num_tokens": 276578524.0, + "step": 7121 + }, + { + "epoch": 0.9059916041216131, + "grad_norm": 2.4177632331848145, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8856884241104126, + "num_tokens": 276620270.0, + "step": 7122 + }, + { + "epoch": 0.9061188144002036, + "grad_norm": 2.641082286834717, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8697138428688049, + "num_tokens": 276660053.0, + "step": 7123 + }, + { + "epoch": 0.906246024678794, + "grad_norm": 2.604004383087158, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8737336993217468, + "num_tokens": 276700866.0, + "step": 7124 + }, + { + "epoch": 0.9063732349573845, + "grad_norm": 2.5656282901763916, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8843291997909546, + "num_tokens": 276737234.0, + "step": 7125 + }, + { + "epoch": 0.9065004452359751, + "grad_norm": 2.4432754516601562, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8785063028335571, + "num_tokens": 276780879.0, + "step": 7126 + }, + { + "epoch": 0.9066276555145656, + "grad_norm": 2.671095132827759, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8852000832557678, + "num_tokens": 276815375.0, + "step": 7127 + }, + { + "epoch": 0.9067548657931561, + "grad_norm": 2.5880064964294434, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8737304210662842, + "num_tokens": 276854771.0, + "step": 7128 + }, + { + "epoch": 0.9068820760717466, + "grad_norm": 2.5361390113830566, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8803762197494507, + "num_tokens": 276894241.0, + "step": 7129 + }, + { + "epoch": 0.9070092863503371, + "grad_norm": 2.63327693939209, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8776476979255676, + "num_tokens": 276931690.0, + "step": 7130 + }, + { + "epoch": 0.9071364966289276, + "grad_norm": 2.57320499420166, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8757242560386658, + "num_tokens": 276970739.0, + "step": 7131 + }, + { + "epoch": 0.9072637069075181, + "grad_norm": 2.5288643836975098, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8844844102859497, + "num_tokens": 277008381.0, + "step": 7132 + }, + { + "epoch": 0.9073909171861086, + "grad_norm": 2.569748640060425, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8826189637184143, + "num_tokens": 277041475.0, + "step": 7133 + }, + { + "epoch": 0.9075181274646992, + "grad_norm": 2.4573330879211426, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.886501133441925, + "num_tokens": 277080462.0, + "step": 7134 + }, + { + "epoch": 0.9076453377432897, + "grad_norm": 2.623128652572632, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8744781017303467, + "num_tokens": 277118782.0, + "step": 7135 + }, + { + "epoch": 0.9077725480218801, + "grad_norm": 2.7943623065948486, + "learning_rate": 1e-06, + "loss": 0.3758, + "mean_token_accuracy": 0.8677089214324951, + "num_tokens": 277152083.0, + "step": 7136 + }, + { + "epoch": 0.9078997583004706, + "grad_norm": 2.662022590637207, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8883185386657715, + "num_tokens": 277185262.0, + "step": 7137 + }, + { + "epoch": 0.9080269685790612, + "grad_norm": 2.8112685680389404, + "learning_rate": 1e-06, + "loss": 0.4172, + "mean_token_accuracy": 0.8569773435592651, + "num_tokens": 277222623.0, + "step": 7138 + }, + { + "epoch": 0.9081541788576517, + "grad_norm": 2.4373719692230225, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8804029226303101, + "num_tokens": 277263796.0, + "step": 7139 + }, + { + "epoch": 0.9082813891362422, + "grad_norm": 3.0428550243377686, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.878049373626709, + "num_tokens": 277295644.0, + "step": 7140 + }, + { + "epoch": 0.9084085994148328, + "grad_norm": 2.730684518814087, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8791502714157104, + "num_tokens": 277334555.0, + "step": 7141 + }, + { + "epoch": 0.9085358096934232, + "grad_norm": 2.36785626411438, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8838165998458862, + "num_tokens": 277376914.0, + "step": 7142 + }, + { + "epoch": 0.9086630199720137, + "grad_norm": 2.461265802383423, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8706291913986206, + "num_tokens": 277423518.0, + "step": 7143 + }, + { + "epoch": 0.9087902302506042, + "grad_norm": 2.5847039222717285, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8899960517883301, + "num_tokens": 277455862.0, + "step": 7144 + }, + { + "epoch": 0.9089174405291948, + "grad_norm": 2.5112013816833496, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8793814182281494, + "num_tokens": 277493228.0, + "step": 7145 + }, + { + "epoch": 0.9090446508077853, + "grad_norm": 2.616061210632324, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8680847883224487, + "num_tokens": 277530472.0, + "step": 7146 + }, + { + "epoch": 0.9091718610863758, + "grad_norm": 2.3588802814483643, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8833878040313721, + "num_tokens": 277574368.0, + "step": 7147 + }, + { + "epoch": 0.9092990713649662, + "grad_norm": 2.671175956726074, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8746173977851868, + "num_tokens": 277614114.0, + "step": 7148 + }, + { + "epoch": 0.9094262816435568, + "grad_norm": 2.662423849105835, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8711278438568115, + "num_tokens": 277651504.0, + "step": 7149 + }, + { + "epoch": 0.9095534919221473, + "grad_norm": 2.5026326179504395, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8843744397163391, + "num_tokens": 277690675.0, + "step": 7150 + }, + { + "epoch": 0.9096807022007378, + "grad_norm": 2.337906837463379, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8775985836982727, + "num_tokens": 277736999.0, + "step": 7151 + }, + { + "epoch": 0.9098079124793284, + "grad_norm": 2.5744333267211914, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.878469705581665, + "num_tokens": 277772372.0, + "step": 7152 + }, + { + "epoch": 0.9099351227579189, + "grad_norm": 2.6252315044403076, + "learning_rate": 1e-06, + "loss": 0.3981, + "mean_token_accuracy": 0.8621779680252075, + "num_tokens": 277814310.0, + "step": 7153 + }, + { + "epoch": 0.9100623330365093, + "grad_norm": 2.3914289474487305, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8875638246536255, + "num_tokens": 277857952.0, + "step": 7154 + }, + { + "epoch": 0.9101895433150998, + "grad_norm": 2.7389373779296875, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8854919672012329, + "num_tokens": 277891432.0, + "step": 7155 + }, + { + "epoch": 0.9103167535936904, + "grad_norm": 2.6148645877838135, + "learning_rate": 1e-06, + "loss": 0.3656, + "mean_token_accuracy": 0.8727360963821411, + "num_tokens": 277929153.0, + "step": 7156 + }, + { + "epoch": 0.9104439638722809, + "grad_norm": 2.338831901550293, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8955498933792114, + "num_tokens": 277966744.0, + "step": 7157 + }, + { + "epoch": 0.9105711741508714, + "grad_norm": 2.490731716156006, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8753222227096558, + "num_tokens": 278007223.0, + "step": 7158 + }, + { + "epoch": 0.9106983844294619, + "grad_norm": 2.3800742626190186, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8896637558937073, + "num_tokens": 278047640.0, + "step": 7159 + }, + { + "epoch": 0.9108255947080524, + "grad_norm": 2.3136494159698486, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8801481127738953, + "num_tokens": 278094731.0, + "step": 7160 + }, + { + "epoch": 0.9109528049866429, + "grad_norm": 2.732283115386963, + "learning_rate": 1e-06, + "loss": 0.3737, + "mean_token_accuracy": 0.8728581070899963, + "num_tokens": 278133839.0, + "step": 7161 + }, + { + "epoch": 0.9110800152652334, + "grad_norm": 2.7502291202545166, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8724240064620972, + "num_tokens": 278170562.0, + "step": 7162 + }, + { + "epoch": 0.9112072255438239, + "grad_norm": 2.6545023918151855, + "learning_rate": 1e-06, + "loss": 0.3731, + "mean_token_accuracy": 0.8725154399871826, + "num_tokens": 278210501.0, + "step": 7163 + }, + { + "epoch": 0.9113344358224145, + "grad_norm": 2.44946551322937, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8853271007537842, + "num_tokens": 278251040.0, + "step": 7164 + }, + { + "epoch": 0.911461646101005, + "grad_norm": 2.575103759765625, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8794527053833008, + "num_tokens": 278290629.0, + "step": 7165 + }, + { + "epoch": 0.9115888563795955, + "grad_norm": 2.7290329933166504, + "learning_rate": 1e-06, + "loss": 0.3768, + "mean_token_accuracy": 0.8697658181190491, + "num_tokens": 278326421.0, + "step": 7166 + }, + { + "epoch": 0.9117160666581859, + "grad_norm": 2.8412702083587646, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8801029920578003, + "num_tokens": 278361381.0, + "step": 7167 + }, + { + "epoch": 0.9118432769367765, + "grad_norm": 2.6737887859344482, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8845540285110474, + "num_tokens": 278396239.0, + "step": 7168 + }, + { + "epoch": 0.911970487215367, + "grad_norm": 2.8369874954223633, + "learning_rate": 1e-06, + "loss": 0.3581, + "mean_token_accuracy": 0.8740288615226746, + "num_tokens": 278429328.0, + "step": 7169 + }, + { + "epoch": 0.9120976974939575, + "grad_norm": 2.268922805786133, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.895331859588623, + "num_tokens": 278470949.0, + "step": 7170 + }, + { + "epoch": 0.912224907772548, + "grad_norm": 2.435687303543091, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8806576132774353, + "num_tokens": 278509594.0, + "step": 7171 + }, + { + "epoch": 0.9123521180511386, + "grad_norm": 2.2580764293670654, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8948730230331421, + "num_tokens": 278554412.0, + "step": 7172 + }, + { + "epoch": 0.912479328329729, + "grad_norm": 2.7108662128448486, + "learning_rate": 1e-06, + "loss": 0.3966, + "mean_token_accuracy": 0.8627241849899292, + "num_tokens": 278590708.0, + "step": 7173 + }, + { + "epoch": 0.9126065386083195, + "grad_norm": 2.401092052459717, + "learning_rate": 1e-06, + "loss": 0.3792, + "mean_token_accuracy": 0.8694133758544922, + "num_tokens": 278637215.0, + "step": 7174 + }, + { + "epoch": 0.9127337488869101, + "grad_norm": 2.5582149028778076, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8806833028793335, + "num_tokens": 278676337.0, + "step": 7175 + }, + { + "epoch": 0.9128609591655006, + "grad_norm": 2.5056798458099365, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8778792023658752, + "num_tokens": 278715278.0, + "step": 7176 + }, + { + "epoch": 0.9129881694440911, + "grad_norm": 2.5606532096862793, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8859539031982422, + "num_tokens": 278753331.0, + "step": 7177 + }, + { + "epoch": 0.9131153797226816, + "grad_norm": 2.4190876483917236, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8748286962509155, + "num_tokens": 278797588.0, + "step": 7178 + }, + { + "epoch": 0.9132425900012721, + "grad_norm": 2.6707310676574707, + "learning_rate": 1e-06, + "loss": 0.3688, + "mean_token_accuracy": 0.8701353669166565, + "num_tokens": 278837149.0, + "step": 7179 + }, + { + "epoch": 0.9133698002798626, + "grad_norm": 2.5569629669189453, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8854696750640869, + "num_tokens": 278877146.0, + "step": 7180 + }, + { + "epoch": 0.9134970105584531, + "grad_norm": 2.5090525150299072, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8758420944213867, + "num_tokens": 278916262.0, + "step": 7181 + }, + { + "epoch": 0.9136242208370436, + "grad_norm": 2.422365427017212, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8733633160591125, + "num_tokens": 278958039.0, + "step": 7182 + }, + { + "epoch": 0.9137514311156342, + "grad_norm": 2.2928361892700195, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.891381561756134, + "num_tokens": 279000131.0, + "step": 7183 + }, + { + "epoch": 0.9138786413942247, + "grad_norm": 2.5758416652679443, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8646079897880554, + "num_tokens": 279043235.0, + "step": 7184 + }, + { + "epoch": 0.9140058516728151, + "grad_norm": 2.458120584487915, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8949609994888306, + "num_tokens": 279081019.0, + "step": 7185 + }, + { + "epoch": 0.9141330619514056, + "grad_norm": 2.5612504482269287, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8864821195602417, + "num_tokens": 279119943.0, + "step": 7186 + }, + { + "epoch": 0.9142602722299962, + "grad_norm": 2.5461955070495605, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8809565901756287, + "num_tokens": 279158969.0, + "step": 7187 + }, + { + "epoch": 0.9143874825085867, + "grad_norm": 2.5671803951263428, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8796738386154175, + "num_tokens": 279195970.0, + "step": 7188 + }, + { + "epoch": 0.9145146927871772, + "grad_norm": 2.3402254581451416, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8825193643569946, + "num_tokens": 279240399.0, + "step": 7189 + }, + { + "epoch": 0.9146419030657678, + "grad_norm": 2.4630281925201416, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.879274845123291, + "num_tokens": 279279743.0, + "step": 7190 + }, + { + "epoch": 0.9147691133443582, + "grad_norm": 2.48714280128479, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8750708699226379, + "num_tokens": 279322048.0, + "step": 7191 + }, + { + "epoch": 0.9148963236229487, + "grad_norm": 2.462456703186035, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8863694667816162, + "num_tokens": 279363072.0, + "step": 7192 + }, + { + "epoch": 0.9150235339015392, + "grad_norm": 2.463014841079712, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8779855370521545, + "num_tokens": 279402609.0, + "step": 7193 + }, + { + "epoch": 0.9151507441801298, + "grad_norm": 2.551741361618042, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8822779655456543, + "num_tokens": 279440286.0, + "step": 7194 + }, + { + "epoch": 0.9152779544587203, + "grad_norm": 2.5489940643310547, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8780033588409424, + "num_tokens": 279477720.0, + "step": 7195 + }, + { + "epoch": 0.9154051647373108, + "grad_norm": 2.381415843963623, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.87269526720047, + "num_tokens": 279525260.0, + "step": 7196 + }, + { + "epoch": 0.9155323750159012, + "grad_norm": 2.5979714393615723, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8783701658248901, + "num_tokens": 279565394.0, + "step": 7197 + }, + { + "epoch": 0.9156595852944918, + "grad_norm": 2.5990214347839355, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8709830045700073, + "num_tokens": 279610071.0, + "step": 7198 + }, + { + "epoch": 0.9157867955730823, + "grad_norm": 2.330796241760254, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8835272789001465, + "num_tokens": 279655120.0, + "step": 7199 + }, + { + "epoch": 0.9159140058516728, + "grad_norm": 2.483466863632202, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8863370418548584, + "num_tokens": 279693953.0, + "step": 7200 + }, + { + "epoch": 0.9160412161302633, + "grad_norm": 2.483051061630249, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.886038064956665, + "num_tokens": 279732097.0, + "step": 7201 + }, + { + "epoch": 0.9161684264088539, + "grad_norm": 2.4520514011383057, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8805669546127319, + "num_tokens": 279771047.0, + "step": 7202 + }, + { + "epoch": 0.9162956366874443, + "grad_norm": 2.59600567817688, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8783191442489624, + "num_tokens": 279807743.0, + "step": 7203 + }, + { + "epoch": 0.9164228469660348, + "grad_norm": 2.426678419113159, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8826345801353455, + "num_tokens": 279848030.0, + "step": 7204 + }, + { + "epoch": 0.9165500572446253, + "grad_norm": 2.6810266971588135, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8816970586776733, + "num_tokens": 279882111.0, + "step": 7205 + }, + { + "epoch": 0.9166772675232159, + "grad_norm": 2.738558292388916, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8834402561187744, + "num_tokens": 279920653.0, + "step": 7206 + }, + { + "epoch": 0.9168044778018064, + "grad_norm": 2.7528209686279297, + "learning_rate": 1e-06, + "loss": 0.381, + "mean_token_accuracy": 0.8616644144058228, + "num_tokens": 279953066.0, + "step": 7207 + }, + { + "epoch": 0.9169316880803969, + "grad_norm": 2.767214059829712, + "learning_rate": 1e-06, + "loss": 0.363, + "mean_token_accuracy": 0.8755714893341064, + "num_tokens": 279988009.0, + "step": 7208 + }, + { + "epoch": 0.9170588983589874, + "grad_norm": 2.7415969371795654, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.877301037311554, + "num_tokens": 280027514.0, + "step": 7209 + }, + { + "epoch": 0.9171861086375779, + "grad_norm": 2.571068286895752, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8887657523155212, + "num_tokens": 280066336.0, + "step": 7210 + }, + { + "epoch": 0.9173133189161684, + "grad_norm": 2.6377458572387695, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8912879228591919, + "num_tokens": 280100462.0, + "step": 7211 + }, + { + "epoch": 0.9174405291947589, + "grad_norm": 2.879206657409668, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8716724514961243, + "num_tokens": 280132246.0, + "step": 7212 + }, + { + "epoch": 0.9175677394733495, + "grad_norm": 2.616708755493164, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8752877712249756, + "num_tokens": 280169111.0, + "step": 7213 + }, + { + "epoch": 0.91769494975194, + "grad_norm": 2.198274850845337, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8869763016700745, + "num_tokens": 280217102.0, + "step": 7214 + }, + { + "epoch": 0.9178221600305305, + "grad_norm": 2.7348175048828125, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8833135962486267, + "num_tokens": 280252800.0, + "step": 7215 + }, + { + "epoch": 0.9179493703091209, + "grad_norm": 2.6599583625793457, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8879597187042236, + "num_tokens": 280285776.0, + "step": 7216 + }, + { + "epoch": 0.9180765805877115, + "grad_norm": 2.611741781234741, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8822094202041626, + "num_tokens": 280324167.0, + "step": 7217 + }, + { + "epoch": 0.918203790866302, + "grad_norm": 2.5833709239959717, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8788626194000244, + "num_tokens": 280363338.0, + "step": 7218 + }, + { + "epoch": 0.9183310011448925, + "grad_norm": 2.519822359085083, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8746271729469299, + "num_tokens": 280402377.0, + "step": 7219 + }, + { + "epoch": 0.918458211423483, + "grad_norm": 2.4375736713409424, + "learning_rate": 1e-06, + "loss": 0.3607, + "mean_token_accuracy": 0.867334246635437, + "num_tokens": 280444989.0, + "step": 7220 + }, + { + "epoch": 0.9185854217020736, + "grad_norm": 2.669508934020996, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8735796809196472, + "num_tokens": 280487740.0, + "step": 7221 + }, + { + "epoch": 0.918712631980664, + "grad_norm": 2.5091211795806885, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8762542605400085, + "num_tokens": 280530974.0, + "step": 7222 + }, + { + "epoch": 0.9188398422592545, + "grad_norm": 2.6891140937805176, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8820680379867554, + "num_tokens": 280568065.0, + "step": 7223 + }, + { + "epoch": 0.918967052537845, + "grad_norm": 2.566774368286133, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8776074051856995, + "num_tokens": 280610833.0, + "step": 7224 + }, + { + "epoch": 0.9190942628164356, + "grad_norm": 2.4617741107940674, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8906646370887756, + "num_tokens": 280646960.0, + "step": 7225 + }, + { + "epoch": 0.9192214730950261, + "grad_norm": 2.69246506690979, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8812806606292725, + "num_tokens": 280682362.0, + "step": 7226 + }, + { + "epoch": 0.9193486833736166, + "grad_norm": 2.453289747238159, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8913503885269165, + "num_tokens": 280719063.0, + "step": 7227 + }, + { + "epoch": 0.919475893652207, + "grad_norm": 2.631376266479492, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8732354640960693, + "num_tokens": 280756212.0, + "step": 7228 + }, + { + "epoch": 0.9196031039307976, + "grad_norm": 2.5882866382598877, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8821261525154114, + "num_tokens": 280791070.0, + "step": 7229 + }, + { + "epoch": 0.9197303142093881, + "grad_norm": 2.26912784576416, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8888328671455383, + "num_tokens": 280835399.0, + "step": 7230 + }, + { + "epoch": 0.9198575244879786, + "grad_norm": 2.7568199634552, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8868317604064941, + "num_tokens": 280870140.0, + "step": 7231 + }, + { + "epoch": 0.9199847347665692, + "grad_norm": 2.5096163749694824, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8800090551376343, + "num_tokens": 280908584.0, + "step": 7232 + }, + { + "epoch": 0.9201119450451597, + "grad_norm": 2.698458671569824, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8830438256263733, + "num_tokens": 280941624.0, + "step": 7233 + }, + { + "epoch": 0.9202391553237501, + "grad_norm": 2.616330862045288, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8797827959060669, + "num_tokens": 280977017.0, + "step": 7234 + }, + { + "epoch": 0.9203663656023406, + "grad_norm": 2.58398175239563, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8748016357421875, + "num_tokens": 281015257.0, + "step": 7235 + }, + { + "epoch": 0.9204935758809312, + "grad_norm": 2.2450814247131348, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8880125284194946, + "num_tokens": 281061830.0, + "step": 7236 + }, + { + "epoch": 0.9206207861595217, + "grad_norm": 2.7755415439605713, + "learning_rate": 1e-06, + "loss": 0.3653, + "mean_token_accuracy": 0.8735601305961609, + "num_tokens": 281100908.0, + "step": 7237 + }, + { + "epoch": 0.9207479964381122, + "grad_norm": 2.7368884086608887, + "learning_rate": 1e-06, + "loss": 0.4026, + "mean_token_accuracy": 0.8616029024124146, + "num_tokens": 281139786.0, + "step": 7238 + }, + { + "epoch": 0.9208752067167028, + "grad_norm": 2.7753021717071533, + "learning_rate": 1e-06, + "loss": 0.3876, + "mean_token_accuracy": 0.8653618693351746, + "num_tokens": 281179403.0, + "step": 7239 + }, + { + "epoch": 0.9210024169952932, + "grad_norm": 2.746021270751953, + "learning_rate": 1e-06, + "loss": 0.3762, + "mean_token_accuracy": 0.8683992624282837, + "num_tokens": 281217158.0, + "step": 7240 + }, + { + "epoch": 0.9211296272738837, + "grad_norm": 2.585254430770874, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8883859515190125, + "num_tokens": 281256679.0, + "step": 7241 + }, + { + "epoch": 0.9212568375524742, + "grad_norm": 2.468444585800171, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8838787078857422, + "num_tokens": 281294689.0, + "step": 7242 + }, + { + "epoch": 0.9213840478310648, + "grad_norm": 2.4343106746673584, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8832725286483765, + "num_tokens": 281336486.0, + "step": 7243 + }, + { + "epoch": 0.9215112581096553, + "grad_norm": 2.548184871673584, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.883162260055542, + "num_tokens": 281374595.0, + "step": 7244 + }, + { + "epoch": 0.9216384683882458, + "grad_norm": 2.798736572265625, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8793736696243286, + "num_tokens": 281408586.0, + "step": 7245 + }, + { + "epoch": 0.9217656786668362, + "grad_norm": 2.5189173221588135, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8837703466415405, + "num_tokens": 281449437.0, + "step": 7246 + }, + { + "epoch": 0.9218928889454268, + "grad_norm": 2.79323410987854, + "learning_rate": 1e-06, + "loss": 0.377, + "mean_token_accuracy": 0.8711338639259338, + "num_tokens": 281487548.0, + "step": 7247 + }, + { + "epoch": 0.9220200992240173, + "grad_norm": 3.0495223999023438, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8717327117919922, + "num_tokens": 281517340.0, + "step": 7248 + }, + { + "epoch": 0.9221473095026078, + "grad_norm": 2.6084322929382324, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8839517831802368, + "num_tokens": 281553605.0, + "step": 7249 + }, + { + "epoch": 0.9222745197811983, + "grad_norm": 2.618176221847534, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8865848183631897, + "num_tokens": 281586270.0, + "step": 7250 + }, + { + "epoch": 0.9224017300597889, + "grad_norm": 2.6644387245178223, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8912661075592041, + "num_tokens": 281626323.0, + "step": 7251 + }, + { + "epoch": 0.9225289403383793, + "grad_norm": 2.52921462059021, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.877772867679596, + "num_tokens": 281665221.0, + "step": 7252 + }, + { + "epoch": 0.9226561506169698, + "grad_norm": 2.4161036014556885, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8750395774841309, + "num_tokens": 281709493.0, + "step": 7253 + }, + { + "epoch": 0.9227833608955603, + "grad_norm": 2.4787776470184326, + "learning_rate": 1e-06, + "loss": 0.3575, + "mean_token_accuracy": 0.873242974281311, + "num_tokens": 281753084.0, + "step": 7254 + }, + { + "epoch": 0.9229105711741509, + "grad_norm": 2.403346538543701, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8886083364486694, + "num_tokens": 281792543.0, + "step": 7255 + }, + { + "epoch": 0.9230377814527414, + "grad_norm": 2.6757805347442627, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.881537914276123, + "num_tokens": 281826659.0, + "step": 7256 + }, + { + "epoch": 0.9231649917313319, + "grad_norm": 2.876966953277588, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8793124556541443, + "num_tokens": 281860343.0, + "step": 7257 + }, + { + "epoch": 0.9232922020099223, + "grad_norm": 2.7736055850982666, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8818560838699341, + "num_tokens": 281894994.0, + "step": 7258 + }, + { + "epoch": 0.9234194122885129, + "grad_norm": 2.461726188659668, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8865170478820801, + "num_tokens": 281932602.0, + "step": 7259 + }, + { + "epoch": 0.9235466225671034, + "grad_norm": 2.8758511543273926, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8789416551589966, + "num_tokens": 281964441.0, + "step": 7260 + }, + { + "epoch": 0.9236738328456939, + "grad_norm": 2.950500726699829, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.8692914843559265, + "num_tokens": 281997617.0, + "step": 7261 + }, + { + "epoch": 0.9238010431242845, + "grad_norm": 2.539363384246826, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8787084817886353, + "num_tokens": 282035465.0, + "step": 7262 + }, + { + "epoch": 0.923928253402875, + "grad_norm": 2.530076026916504, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8779847025871277, + "num_tokens": 282074850.0, + "step": 7263 + }, + { + "epoch": 0.9240554636814655, + "grad_norm": 2.8762271404266357, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.886868953704834, + "num_tokens": 282107061.0, + "step": 7264 + }, + { + "epoch": 0.9241826739600559, + "grad_norm": 2.758326768875122, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8701670169830322, + "num_tokens": 282148194.0, + "step": 7265 + }, + { + "epoch": 0.9243098842386465, + "grad_norm": 2.638867139816284, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8701232075691223, + "num_tokens": 282190023.0, + "step": 7266 + }, + { + "epoch": 0.924437094517237, + "grad_norm": 2.5063295364379883, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8817843794822693, + "num_tokens": 282231994.0, + "step": 7267 + }, + { + "epoch": 0.9245643047958275, + "grad_norm": 2.699326753616333, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8810771107673645, + "num_tokens": 282269251.0, + "step": 7268 + }, + { + "epoch": 0.924691515074418, + "grad_norm": 2.8571553230285645, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8831933736801147, + "num_tokens": 282300600.0, + "step": 7269 + }, + { + "epoch": 0.9248187253530086, + "grad_norm": 2.8532159328460693, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.8706303238868713, + "num_tokens": 282335205.0, + "step": 7270 + }, + { + "epoch": 0.924945935631599, + "grad_norm": 2.611074924468994, + "learning_rate": 1e-06, + "loss": 0.3761, + "mean_token_accuracy": 0.8696209192276001, + "num_tokens": 282376769.0, + "step": 7271 + }, + { + "epoch": 0.9250731459101895, + "grad_norm": 2.67590594291687, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8721964359283447, + "num_tokens": 282414107.0, + "step": 7272 + }, + { + "epoch": 0.92520035618878, + "grad_norm": 2.774646520614624, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.886695146560669, + "num_tokens": 282451523.0, + "step": 7273 + }, + { + "epoch": 0.9253275664673706, + "grad_norm": 2.6997082233428955, + "learning_rate": 1e-06, + "loss": 0.376, + "mean_token_accuracy": 0.8688787221908569, + "num_tokens": 282491856.0, + "step": 7274 + }, + { + "epoch": 0.9254547767459611, + "grad_norm": 2.5806901454925537, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.879991888999939, + "num_tokens": 282530890.0, + "step": 7275 + }, + { + "epoch": 0.9255819870245516, + "grad_norm": 2.612159013748169, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8831411004066467, + "num_tokens": 282565934.0, + "step": 7276 + }, + { + "epoch": 0.925709197303142, + "grad_norm": 2.4430806636810303, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8870850801467896, + "num_tokens": 282604606.0, + "step": 7277 + }, + { + "epoch": 0.9258364075817326, + "grad_norm": 2.665566921234131, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8891721963882446, + "num_tokens": 282637449.0, + "step": 7278 + }, + { + "epoch": 0.9259636178603231, + "grad_norm": 2.751082420349121, + "learning_rate": 1e-06, + "loss": 0.4156, + "mean_token_accuracy": 0.8571013808250427, + "num_tokens": 282677041.0, + "step": 7279 + }, + { + "epoch": 0.9260908281389136, + "grad_norm": 2.711780071258545, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8784886002540588, + "num_tokens": 282710935.0, + "step": 7280 + }, + { + "epoch": 0.9262180384175042, + "grad_norm": 2.4677019119262695, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8932883739471436, + "num_tokens": 282744709.0, + "step": 7281 + }, + { + "epoch": 0.9263452486960947, + "grad_norm": 2.5809214115142822, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8800424337387085, + "num_tokens": 282782076.0, + "step": 7282 + }, + { + "epoch": 0.9264724589746851, + "grad_norm": 2.396888017654419, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8896172046661377, + "num_tokens": 282823928.0, + "step": 7283 + }, + { + "epoch": 0.9265996692532756, + "grad_norm": 2.6787257194519043, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8839961290359497, + "num_tokens": 282856545.0, + "step": 7284 + }, + { + "epoch": 0.9267268795318662, + "grad_norm": 2.430603504180908, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8885483145713806, + "num_tokens": 282894824.0, + "step": 7285 + }, + { + "epoch": 0.9268540898104567, + "grad_norm": 2.4566152095794678, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8813655376434326, + "num_tokens": 282939340.0, + "step": 7286 + }, + { + "epoch": 0.9269813000890472, + "grad_norm": 2.628199577331543, + "learning_rate": 1e-06, + "loss": 0.399, + "mean_token_accuracy": 0.8623625636100769, + "num_tokens": 282979374.0, + "step": 7287 + }, + { + "epoch": 0.9271085103676378, + "grad_norm": 2.441279411315918, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8752981424331665, + "num_tokens": 283023520.0, + "step": 7288 + }, + { + "epoch": 0.9272357206462282, + "grad_norm": 2.496366024017334, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8789302110671997, + "num_tokens": 283063969.0, + "step": 7289 + }, + { + "epoch": 0.9273629309248187, + "grad_norm": 2.6475915908813477, + "learning_rate": 1e-06, + "loss": 0.403, + "mean_token_accuracy": 0.8627215027809143, + "num_tokens": 283105805.0, + "step": 7290 + }, + { + "epoch": 0.9274901412034092, + "grad_norm": 2.481494426727295, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8859301805496216, + "num_tokens": 283147276.0, + "step": 7291 + }, + { + "epoch": 0.9276173514819998, + "grad_norm": 2.4698288440704346, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8746373653411865, + "num_tokens": 283190201.0, + "step": 7292 + }, + { + "epoch": 0.9277445617605903, + "grad_norm": 2.809368848800659, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8736951947212219, + "num_tokens": 283227217.0, + "step": 7293 + }, + { + "epoch": 0.9278717720391808, + "grad_norm": 2.798948287963867, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8858140110969543, + "num_tokens": 283259676.0, + "step": 7294 + }, + { + "epoch": 0.9279989823177712, + "grad_norm": 2.5491902828216553, + "learning_rate": 1e-06, + "loss": 0.4023, + "mean_token_accuracy": 0.861075222492218, + "num_tokens": 283305244.0, + "step": 7295 + }, + { + "epoch": 0.9281261925963618, + "grad_norm": 2.6871962547302246, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8696805834770203, + "num_tokens": 283346646.0, + "step": 7296 + }, + { + "epoch": 0.9282534028749523, + "grad_norm": 2.761106491088867, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8767528533935547, + "num_tokens": 283379737.0, + "step": 7297 + }, + { + "epoch": 0.9283806131535428, + "grad_norm": 2.5224783420562744, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8895313143730164, + "num_tokens": 283421504.0, + "step": 7298 + }, + { + "epoch": 0.9285078234321333, + "grad_norm": 2.61078143119812, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8928980231285095, + "num_tokens": 283452828.0, + "step": 7299 + }, + { + "epoch": 0.9286350337107239, + "grad_norm": 2.2417211532592773, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8894895315170288, + "num_tokens": 283496502.0, + "step": 7300 + }, + { + "epoch": 0.9287622439893143, + "grad_norm": 2.343618392944336, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8877504467964172, + "num_tokens": 283540852.0, + "step": 7301 + }, + { + "epoch": 0.9288894542679048, + "grad_norm": 2.3471732139587402, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8834208250045776, + "num_tokens": 283582693.0, + "step": 7302 + }, + { + "epoch": 0.9290166645464953, + "grad_norm": 2.775238037109375, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8809620141983032, + "num_tokens": 283616497.0, + "step": 7303 + }, + { + "epoch": 0.9291438748250859, + "grad_norm": 2.644406318664551, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8737002611160278, + "num_tokens": 283658148.0, + "step": 7304 + }, + { + "epoch": 0.9292710851036764, + "grad_norm": 2.785992383956909, + "learning_rate": 1e-06, + "loss": 0.3786, + "mean_token_accuracy": 0.8675973415374756, + "num_tokens": 283694593.0, + "step": 7305 + }, + { + "epoch": 0.9293982953822669, + "grad_norm": 2.531557083129883, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.884471595287323, + "num_tokens": 283737347.0, + "step": 7306 + }, + { + "epoch": 0.9295255056608573, + "grad_norm": 2.387118101119995, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8810406923294067, + "num_tokens": 283779339.0, + "step": 7307 + }, + { + "epoch": 0.9296527159394479, + "grad_norm": 2.8017704486846924, + "learning_rate": 1e-06, + "loss": 0.4041, + "mean_token_accuracy": 0.861330509185791, + "num_tokens": 283816904.0, + "step": 7308 + }, + { + "epoch": 0.9297799262180384, + "grad_norm": 2.571946144104004, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8815569877624512, + "num_tokens": 283855992.0, + "step": 7309 + }, + { + "epoch": 0.9299071364966289, + "grad_norm": 2.592775344848633, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8831632137298584, + "num_tokens": 283894224.0, + "step": 7310 + }, + { + "epoch": 0.9300343467752195, + "grad_norm": 2.4075279235839844, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.8723622560501099, + "num_tokens": 283938189.0, + "step": 7311 + }, + { + "epoch": 0.93016155705381, + "grad_norm": 2.686699390411377, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8711574673652649, + "num_tokens": 283976299.0, + "step": 7312 + }, + { + "epoch": 0.9302887673324005, + "grad_norm": 2.7175440788269043, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.87003093957901, + "num_tokens": 284013390.0, + "step": 7313 + }, + { + "epoch": 0.9304159776109909, + "grad_norm": 2.5059545040130615, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8779025077819824, + "num_tokens": 284053292.0, + "step": 7314 + }, + { + "epoch": 0.9305431878895815, + "grad_norm": 2.6356241703033447, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8682848215103149, + "num_tokens": 284091985.0, + "step": 7315 + }, + { + "epoch": 0.930670398168172, + "grad_norm": 2.417649745941162, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8901468515396118, + "num_tokens": 284130989.0, + "step": 7316 + }, + { + "epoch": 0.9307976084467625, + "grad_norm": 2.7511777877807617, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8854751586914062, + "num_tokens": 284162497.0, + "step": 7317 + }, + { + "epoch": 0.930924818725353, + "grad_norm": 2.623677968978882, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.876007080078125, + "num_tokens": 284200001.0, + "step": 7318 + }, + { + "epoch": 0.9310520290039436, + "grad_norm": 2.7181432247161865, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8712455034255981, + "num_tokens": 284236435.0, + "step": 7319 + }, + { + "epoch": 0.931179239282534, + "grad_norm": 2.663170099258423, + "learning_rate": 1e-06, + "loss": 0.3659, + "mean_token_accuracy": 0.8741105198860168, + "num_tokens": 284273417.0, + "step": 7320 + }, + { + "epoch": 0.9313064495611245, + "grad_norm": 2.534478187561035, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8778517246246338, + "num_tokens": 284311514.0, + "step": 7321 + }, + { + "epoch": 0.931433659839715, + "grad_norm": 2.8408331871032715, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8768963813781738, + "num_tokens": 284343005.0, + "step": 7322 + }, + { + "epoch": 0.9315608701183056, + "grad_norm": 2.431269407272339, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8893961310386658, + "num_tokens": 284380649.0, + "step": 7323 + }, + { + "epoch": 0.9316880803968961, + "grad_norm": 2.47602915763855, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8781483173370361, + "num_tokens": 284420042.0, + "step": 7324 + }, + { + "epoch": 0.9318152906754866, + "grad_norm": 2.468428373336792, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8807401061058044, + "num_tokens": 284461076.0, + "step": 7325 + }, + { + "epoch": 0.931942500954077, + "grad_norm": 2.4479053020477295, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.881264328956604, + "num_tokens": 284501985.0, + "step": 7326 + }, + { + "epoch": 0.9320697112326676, + "grad_norm": 2.521280527114868, + "learning_rate": 1e-06, + "loss": 0.3838, + "mean_token_accuracy": 0.8674814105033875, + "num_tokens": 284544942.0, + "step": 7327 + }, + { + "epoch": 0.9321969215112581, + "grad_norm": 2.370220899581909, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8879430890083313, + "num_tokens": 284586034.0, + "step": 7328 + }, + { + "epoch": 0.9323241317898486, + "grad_norm": 2.52577543258667, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8826184272766113, + "num_tokens": 284622434.0, + "step": 7329 + }, + { + "epoch": 0.9324513420684392, + "grad_norm": 2.623983383178711, + "learning_rate": 1e-06, + "loss": 0.3778, + "mean_token_accuracy": 0.8713973760604858, + "num_tokens": 284660233.0, + "step": 7330 + }, + { + "epoch": 0.9325785523470297, + "grad_norm": 2.41048002243042, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8840652704238892, + "num_tokens": 284701675.0, + "step": 7331 + }, + { + "epoch": 0.9327057626256201, + "grad_norm": 2.824493408203125, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8709768056869507, + "num_tokens": 284736291.0, + "step": 7332 + }, + { + "epoch": 0.9328329729042106, + "grad_norm": 2.5683343410491943, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8894144296646118, + "num_tokens": 284772865.0, + "step": 7333 + }, + { + "epoch": 0.9329601831828012, + "grad_norm": 2.3359968662261963, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8824681043624878, + "num_tokens": 284817124.0, + "step": 7334 + }, + { + "epoch": 0.9330873934613917, + "grad_norm": 2.8063066005706787, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8821497559547424, + "num_tokens": 284849117.0, + "step": 7335 + }, + { + "epoch": 0.9332146037399822, + "grad_norm": 2.387239456176758, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8887648582458496, + "num_tokens": 284888038.0, + "step": 7336 + }, + { + "epoch": 0.9333418140185727, + "grad_norm": 2.4323840141296387, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8797411918640137, + "num_tokens": 284927531.0, + "step": 7337 + }, + { + "epoch": 0.9334690242971632, + "grad_norm": 2.8099777698516846, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8778330087661743, + "num_tokens": 284962192.0, + "step": 7338 + }, + { + "epoch": 0.9335962345757537, + "grad_norm": 2.5939457416534424, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8781184554100037, + "num_tokens": 285002724.0, + "step": 7339 + }, + { + "epoch": 0.9337234448543442, + "grad_norm": 2.7056777477264404, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8904401063919067, + "num_tokens": 285036623.0, + "step": 7340 + }, + { + "epoch": 0.9338506551329347, + "grad_norm": 2.5400550365448, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8730531930923462, + "num_tokens": 285077832.0, + "step": 7341 + }, + { + "epoch": 0.9339778654115253, + "grad_norm": 2.8678672313690186, + "learning_rate": 1e-06, + "loss": 0.3954, + "mean_token_accuracy": 0.8594975471496582, + "num_tokens": 285112363.0, + "step": 7342 + }, + { + "epoch": 0.9341050756901158, + "grad_norm": 2.6465182304382324, + "learning_rate": 1e-06, + "loss": 0.37, + "mean_token_accuracy": 0.8699890375137329, + "num_tokens": 285152205.0, + "step": 7343 + }, + { + "epoch": 0.9342322859687062, + "grad_norm": 2.5435304641723633, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8861061930656433, + "num_tokens": 285189747.0, + "step": 7344 + }, + { + "epoch": 0.9343594962472968, + "grad_norm": 3.296302318572998, + "learning_rate": 1e-06, + "loss": 0.4092, + "mean_token_accuracy": 0.8566505312919617, + "num_tokens": 285218842.0, + "step": 7345 + }, + { + "epoch": 0.9344867065258873, + "grad_norm": 2.7637388706207275, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8776699304580688, + "num_tokens": 285252688.0, + "step": 7346 + }, + { + "epoch": 0.9346139168044778, + "grad_norm": 2.6181256771087646, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8790528178215027, + "num_tokens": 285288286.0, + "step": 7347 + }, + { + "epoch": 0.9347411270830683, + "grad_norm": 2.548252820968628, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8865966200828552, + "num_tokens": 285326331.0, + "step": 7348 + }, + { + "epoch": 0.9348683373616589, + "grad_norm": 2.656813859939575, + "learning_rate": 1e-06, + "loss": 0.3608, + "mean_token_accuracy": 0.8746179342269897, + "num_tokens": 285365448.0, + "step": 7349 + }, + { + "epoch": 0.9349955476402493, + "grad_norm": 2.494494676589966, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.879185676574707, + "num_tokens": 285404495.0, + "step": 7350 + }, + { + "epoch": 0.9351227579188398, + "grad_norm": 2.5701115131378174, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8685951232910156, + "num_tokens": 285442983.0, + "step": 7351 + }, + { + "epoch": 0.9352499681974303, + "grad_norm": 2.6516363620758057, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8855544924736023, + "num_tokens": 285477803.0, + "step": 7352 + }, + { + "epoch": 0.9353771784760209, + "grad_norm": 2.405268430709839, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8790351748466492, + "num_tokens": 285520620.0, + "step": 7353 + }, + { + "epoch": 0.9355043887546114, + "grad_norm": 2.577594757080078, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8930137157440186, + "num_tokens": 285555063.0, + "step": 7354 + }, + { + "epoch": 0.9356315990332019, + "grad_norm": 2.437912940979004, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.888730525970459, + "num_tokens": 285594061.0, + "step": 7355 + }, + { + "epoch": 0.9357588093117923, + "grad_norm": 2.8231208324432373, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8741479516029358, + "num_tokens": 285630646.0, + "step": 7356 + }, + { + "epoch": 0.9358860195903829, + "grad_norm": 2.801417589187622, + "learning_rate": 1e-06, + "loss": 0.3881, + "mean_token_accuracy": 0.8649007081985474, + "num_tokens": 285669053.0, + "step": 7357 + }, + { + "epoch": 0.9360132298689734, + "grad_norm": 2.5001049041748047, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8735796809196472, + "num_tokens": 285708600.0, + "step": 7358 + }, + { + "epoch": 0.9361404401475639, + "grad_norm": 2.582378625869751, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8711168766021729, + "num_tokens": 285747330.0, + "step": 7359 + }, + { + "epoch": 0.9362676504261545, + "grad_norm": 2.5944321155548096, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8824487924575806, + "num_tokens": 285782096.0, + "step": 7360 + }, + { + "epoch": 0.936394860704745, + "grad_norm": 2.41996169090271, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8845450282096863, + "num_tokens": 285824822.0, + "step": 7361 + }, + { + "epoch": 0.9365220709833355, + "grad_norm": 2.4735829830169678, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8935050368309021, + "num_tokens": 285862359.0, + "step": 7362 + }, + { + "epoch": 0.9366492812619259, + "grad_norm": 2.880185604095459, + "learning_rate": 1e-06, + "loss": 0.3769, + "mean_token_accuracy": 0.8658910989761353, + "num_tokens": 285898351.0, + "step": 7363 + }, + { + "epoch": 0.9367764915405165, + "grad_norm": 2.461803674697876, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8892626762390137, + "num_tokens": 285937537.0, + "step": 7364 + }, + { + "epoch": 0.936903701819107, + "grad_norm": 2.4036524295806885, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8751442432403564, + "num_tokens": 285983296.0, + "step": 7365 + }, + { + "epoch": 0.9370309120976975, + "grad_norm": 2.6857290267944336, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8755650520324707, + "num_tokens": 286020698.0, + "step": 7366 + }, + { + "epoch": 0.937158122376288, + "grad_norm": 2.4359290599823, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8883084058761597, + "num_tokens": 286060688.0, + "step": 7367 + }, + { + "epoch": 0.9372853326548786, + "grad_norm": 2.329317569732666, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8904621601104736, + "num_tokens": 286103532.0, + "step": 7368 + }, + { + "epoch": 0.937412542933469, + "grad_norm": 2.8126955032348633, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8792060613632202, + "num_tokens": 286136808.0, + "step": 7369 + }, + { + "epoch": 0.9375397532120595, + "grad_norm": 2.63112211227417, + "learning_rate": 1e-06, + "loss": 0.3808, + "mean_token_accuracy": 0.8659411668777466, + "num_tokens": 286176327.0, + "step": 7370 + }, + { + "epoch": 0.93766696349065, + "grad_norm": 2.5719542503356934, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.881467342376709, + "num_tokens": 286214616.0, + "step": 7371 + }, + { + "epoch": 0.9377941737692406, + "grad_norm": 2.4094579219818115, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8931524753570557, + "num_tokens": 286253694.0, + "step": 7372 + }, + { + "epoch": 0.9379213840478311, + "grad_norm": 2.584264039993286, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8804539442062378, + "num_tokens": 286291476.0, + "step": 7373 + }, + { + "epoch": 0.9380485943264216, + "grad_norm": 2.508531332015991, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8833391070365906, + "num_tokens": 286329295.0, + "step": 7374 + }, + { + "epoch": 0.938175804605012, + "grad_norm": 2.4323668479919434, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8752743601799011, + "num_tokens": 286368493.0, + "step": 7375 + }, + { + "epoch": 0.9383030148836026, + "grad_norm": 2.7665903568267822, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8800877332687378, + "num_tokens": 286402704.0, + "step": 7376 + }, + { + "epoch": 0.9384302251621931, + "grad_norm": 2.6540093421936035, + "learning_rate": 1e-06, + "loss": 0.3931, + "mean_token_accuracy": 0.8628160953521729, + "num_tokens": 286445427.0, + "step": 7377 + }, + { + "epoch": 0.9385574354407836, + "grad_norm": 2.5633387565612793, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.877135694026947, + "num_tokens": 286486559.0, + "step": 7378 + }, + { + "epoch": 0.9386846457193742, + "grad_norm": 2.7813668251037598, + "learning_rate": 1e-06, + "loss": 0.3783, + "mean_token_accuracy": 0.8694267272949219, + "num_tokens": 286523375.0, + "step": 7379 + }, + { + "epoch": 0.9388118559979647, + "grad_norm": 2.7046682834625244, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8760030269622803, + "num_tokens": 286557062.0, + "step": 7380 + }, + { + "epoch": 0.9389390662765551, + "grad_norm": 2.6325464248657227, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8737502098083496, + "num_tokens": 286596696.0, + "step": 7381 + }, + { + "epoch": 0.9390662765551456, + "grad_norm": 2.483861207962036, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8735824823379517, + "num_tokens": 286638375.0, + "step": 7382 + }, + { + "epoch": 0.9391934868337362, + "grad_norm": 2.6339120864868164, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8784657716751099, + "num_tokens": 286674970.0, + "step": 7383 + }, + { + "epoch": 0.9393206971123267, + "grad_norm": 2.540220022201538, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8789992332458496, + "num_tokens": 286714568.0, + "step": 7384 + }, + { + "epoch": 0.9394479073909172, + "grad_norm": 2.4593167304992676, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8925064206123352, + "num_tokens": 286751818.0, + "step": 7385 + }, + { + "epoch": 0.9395751176695077, + "grad_norm": 2.5417895317077637, + "learning_rate": 1e-06, + "loss": 0.3849, + "mean_token_accuracy": 0.8736600279808044, + "num_tokens": 286791504.0, + "step": 7386 + }, + { + "epoch": 0.9397023279480982, + "grad_norm": 2.4962379932403564, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8777268528938293, + "num_tokens": 286830166.0, + "step": 7387 + }, + { + "epoch": 0.9398295382266887, + "grad_norm": 2.6494414806365967, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8811783790588379, + "num_tokens": 286868299.0, + "step": 7388 + }, + { + "epoch": 0.9399567485052792, + "grad_norm": 2.8396658897399902, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8650124669075012, + "num_tokens": 286903149.0, + "step": 7389 + }, + { + "epoch": 0.9400839587838697, + "grad_norm": 2.6383090019226074, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8923617601394653, + "num_tokens": 286936719.0, + "step": 7390 + }, + { + "epoch": 0.9402111690624603, + "grad_norm": 2.6459712982177734, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8745225071907043, + "num_tokens": 286975738.0, + "step": 7391 + }, + { + "epoch": 0.9403383793410508, + "grad_norm": 2.5602755546569824, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.875450849533081, + "num_tokens": 287015850.0, + "step": 7392 + }, + { + "epoch": 0.9404655896196412, + "grad_norm": 2.728878974914551, + "learning_rate": 1e-06, + "loss": 0.3719, + "mean_token_accuracy": 0.8721740245819092, + "num_tokens": 287052761.0, + "step": 7393 + }, + { + "epoch": 0.9405927998982317, + "grad_norm": 2.553454637527466, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8827290534973145, + "num_tokens": 287093606.0, + "step": 7394 + }, + { + "epoch": 0.9407200101768223, + "grad_norm": 2.5979182720184326, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8957363963127136, + "num_tokens": 287127105.0, + "step": 7395 + }, + { + "epoch": 0.9408472204554128, + "grad_norm": 2.450007438659668, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.887523889541626, + "num_tokens": 287165182.0, + "step": 7396 + }, + { + "epoch": 0.9409744307340033, + "grad_norm": 2.312037467956543, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8826828002929688, + "num_tokens": 287211954.0, + "step": 7397 + }, + { + "epoch": 0.9411016410125939, + "grad_norm": 2.704873561859131, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8810369968414307, + "num_tokens": 287245356.0, + "step": 7398 + }, + { + "epoch": 0.9412288512911843, + "grad_norm": 2.5385823249816895, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8858292102813721, + "num_tokens": 287282189.0, + "step": 7399 + }, + { + "epoch": 0.9413560615697748, + "grad_norm": 2.687608480453491, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.872978687286377, + "num_tokens": 287319499.0, + "step": 7400 + }, + { + "epoch": 0.9414832718483653, + "grad_norm": 2.581834554672241, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.880027174949646, + "num_tokens": 287357151.0, + "step": 7401 + }, + { + "epoch": 0.9416104821269559, + "grad_norm": 2.7080628871917725, + "learning_rate": 1e-06, + "loss": 0.4129, + "mean_token_accuracy": 0.8677120208740234, + "num_tokens": 287399359.0, + "step": 7402 + }, + { + "epoch": 0.9417376924055464, + "grad_norm": 2.703413248062134, + "learning_rate": 1e-06, + "loss": 0.382, + "mean_token_accuracy": 0.8628619909286499, + "num_tokens": 287434795.0, + "step": 7403 + }, + { + "epoch": 0.9418649026841369, + "grad_norm": 2.5345869064331055, + "learning_rate": 1e-06, + "loss": 0.3515, + "mean_token_accuracy": 0.8775631785392761, + "num_tokens": 287476350.0, + "step": 7404 + }, + { + "epoch": 0.9419921129627273, + "grad_norm": 2.110957145690918, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8993849158287048, + "num_tokens": 287520546.0, + "step": 7405 + }, + { + "epoch": 0.9421193232413179, + "grad_norm": 2.4612674713134766, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8874849081039429, + "num_tokens": 287560141.0, + "step": 7406 + }, + { + "epoch": 0.9422465335199084, + "grad_norm": 2.476865291595459, + "learning_rate": 1e-06, + "loss": 0.3972, + "mean_token_accuracy": 0.8606877326965332, + "num_tokens": 287604540.0, + "step": 7407 + }, + { + "epoch": 0.9423737437984989, + "grad_norm": 2.5245144367218018, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8862472772598267, + "num_tokens": 287644639.0, + "step": 7408 + }, + { + "epoch": 0.9425009540770894, + "grad_norm": 2.797586441040039, + "learning_rate": 1e-06, + "loss": 0.3964, + "mean_token_accuracy": 0.8630627393722534, + "num_tokens": 287682117.0, + "step": 7409 + }, + { + "epoch": 0.94262816435568, + "grad_norm": 2.5734450817108154, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8818924427032471, + "num_tokens": 287720878.0, + "step": 7410 + }, + { + "epoch": 0.9427553746342705, + "grad_norm": 2.6502532958984375, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8699624538421631, + "num_tokens": 287759863.0, + "step": 7411 + }, + { + "epoch": 0.9428825849128609, + "grad_norm": 2.677945852279663, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8732582330703735, + "num_tokens": 287797409.0, + "step": 7412 + }, + { + "epoch": 0.9430097951914514, + "grad_norm": 2.5678765773773193, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.885108470916748, + "num_tokens": 287835011.0, + "step": 7413 + }, + { + "epoch": 0.943137005470042, + "grad_norm": 2.5077006816864014, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8851948380470276, + "num_tokens": 287875704.0, + "step": 7414 + }, + { + "epoch": 0.9432642157486325, + "grad_norm": 2.6103873252868652, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8719784617424011, + "num_tokens": 287911604.0, + "step": 7415 + }, + { + "epoch": 0.943391426027223, + "grad_norm": 2.7027580738067627, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8913117051124573, + "num_tokens": 287944750.0, + "step": 7416 + }, + { + "epoch": 0.9435186363058136, + "grad_norm": 2.8657658100128174, + "learning_rate": 1e-06, + "loss": 0.3893, + "mean_token_accuracy": 0.8638182878494263, + "num_tokens": 287982412.0, + "step": 7417 + }, + { + "epoch": 0.943645846584404, + "grad_norm": 2.7438297271728516, + "learning_rate": 1e-06, + "loss": 0.3703, + "mean_token_accuracy": 0.8698049187660217, + "num_tokens": 288023967.0, + "step": 7418 + }, + { + "epoch": 0.9437730568629945, + "grad_norm": 2.529726266860962, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8869831562042236, + "num_tokens": 288065232.0, + "step": 7419 + }, + { + "epoch": 0.943900267141585, + "grad_norm": 2.4413628578186035, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8888263702392578, + "num_tokens": 288105301.0, + "step": 7420 + }, + { + "epoch": 0.9440274774201756, + "grad_norm": 2.7388436794281006, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.888730525970459, + "num_tokens": 288142482.0, + "step": 7421 + }, + { + "epoch": 0.9441546876987661, + "grad_norm": 2.2321617603302, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.896216869354248, + "num_tokens": 288187840.0, + "step": 7422 + }, + { + "epoch": 0.9442818979773566, + "grad_norm": 2.4367198944091797, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8908524513244629, + "num_tokens": 288226443.0, + "step": 7423 + }, + { + "epoch": 0.944409108255947, + "grad_norm": 2.784827709197998, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8774862289428711, + "num_tokens": 288258570.0, + "step": 7424 + }, + { + "epoch": 0.9445363185345376, + "grad_norm": 2.85331392288208, + "learning_rate": 1e-06, + "loss": 0.4004, + "mean_token_accuracy": 0.861333966255188, + "num_tokens": 288293780.0, + "step": 7425 + }, + { + "epoch": 0.9446635288131281, + "grad_norm": 2.636101007461548, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8729437589645386, + "num_tokens": 288334142.0, + "step": 7426 + }, + { + "epoch": 0.9447907390917186, + "grad_norm": 2.509368896484375, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8819948434829712, + "num_tokens": 288375036.0, + "step": 7427 + }, + { + "epoch": 0.9449179493703092, + "grad_norm": 2.6371946334838867, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.878650426864624, + "num_tokens": 288414432.0, + "step": 7428 + }, + { + "epoch": 0.9450451596488997, + "grad_norm": 2.5552613735198975, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8837321996688843, + "num_tokens": 288452999.0, + "step": 7429 + }, + { + "epoch": 0.9451723699274901, + "grad_norm": 2.398167848587036, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8741150498390198, + "num_tokens": 288498262.0, + "step": 7430 + }, + { + "epoch": 0.9452995802060806, + "grad_norm": 2.574561357498169, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.882624089717865, + "num_tokens": 288533764.0, + "step": 7431 + }, + { + "epoch": 0.9454267904846712, + "grad_norm": 2.4522860050201416, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8861284255981445, + "num_tokens": 288570177.0, + "step": 7432 + }, + { + "epoch": 0.9455540007632617, + "grad_norm": 2.442356586456299, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8871903419494629, + "num_tokens": 288609282.0, + "step": 7433 + }, + { + "epoch": 0.9456812110418522, + "grad_norm": 2.846872091293335, + "learning_rate": 1e-06, + "loss": 0.3878, + "mean_token_accuracy": 0.8652592897415161, + "num_tokens": 288644480.0, + "step": 7434 + }, + { + "epoch": 0.9458084213204427, + "grad_norm": 2.5635712146759033, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8739437460899353, + "num_tokens": 288683951.0, + "step": 7435 + }, + { + "epoch": 0.9459356315990332, + "grad_norm": 2.5615992546081543, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8866627216339111, + "num_tokens": 288721242.0, + "step": 7436 + }, + { + "epoch": 0.9460628418776237, + "grad_norm": 2.4588828086853027, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8897210955619812, + "num_tokens": 288758560.0, + "step": 7437 + }, + { + "epoch": 0.9461900521562142, + "grad_norm": 2.4675745964050293, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8797207474708557, + "num_tokens": 288797849.0, + "step": 7438 + }, + { + "epoch": 0.9463172624348047, + "grad_norm": 2.581392288208008, + "learning_rate": 1e-06, + "loss": 0.3665, + "mean_token_accuracy": 0.8722788095474243, + "num_tokens": 288843379.0, + "step": 7439 + }, + { + "epoch": 0.9464444727133953, + "grad_norm": 2.9503536224365234, + "learning_rate": 1e-06, + "loss": 0.4159, + "mean_token_accuracy": 0.8576900959014893, + "num_tokens": 288875802.0, + "step": 7440 + }, + { + "epoch": 0.9465716829919858, + "grad_norm": 2.5143980979919434, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8792632818222046, + "num_tokens": 288915702.0, + "step": 7441 + }, + { + "epoch": 0.9466988932705762, + "grad_norm": 2.587675094604492, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8654377460479736, + "num_tokens": 288958531.0, + "step": 7442 + }, + { + "epoch": 0.9468261035491667, + "grad_norm": 2.432924509048462, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8772651553153992, + "num_tokens": 288997850.0, + "step": 7443 + }, + { + "epoch": 0.9469533138277573, + "grad_norm": 2.7465856075286865, + "learning_rate": 1e-06, + "loss": 0.3881, + "mean_token_accuracy": 0.8653071522712708, + "num_tokens": 289035969.0, + "step": 7444 + }, + { + "epoch": 0.9470805241063478, + "grad_norm": 2.551530122756958, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8929526805877686, + "num_tokens": 289071504.0, + "step": 7445 + }, + { + "epoch": 0.9472077343849383, + "grad_norm": 2.7068777084350586, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8911924958229065, + "num_tokens": 289106826.0, + "step": 7446 + }, + { + "epoch": 0.9473349446635289, + "grad_norm": 2.395946741104126, + "learning_rate": 1e-06, + "loss": 0.3718, + "mean_token_accuracy": 0.870587170124054, + "num_tokens": 289154433.0, + "step": 7447 + }, + { + "epoch": 0.9474621549421193, + "grad_norm": 2.519697666168213, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8890668153762817, + "num_tokens": 289193867.0, + "step": 7448 + }, + { + "epoch": 0.9475893652207098, + "grad_norm": 2.453986644744873, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8717759251594543, + "num_tokens": 289235189.0, + "step": 7449 + }, + { + "epoch": 0.9477165754993003, + "grad_norm": 2.5836987495422363, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.879703164100647, + "num_tokens": 289273542.0, + "step": 7450 + }, + { + "epoch": 0.9478437857778909, + "grad_norm": 2.6915297508239746, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8837586641311646, + "num_tokens": 289307700.0, + "step": 7451 + }, + { + "epoch": 0.9479709960564814, + "grad_norm": 2.504380941390991, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8767011165618896, + "num_tokens": 289347091.0, + "step": 7452 + }, + { + "epoch": 0.9480982063350719, + "grad_norm": 2.741035223007202, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8818869590759277, + "num_tokens": 289380556.0, + "step": 7453 + }, + { + "epoch": 0.9482254166136623, + "grad_norm": 2.7755417823791504, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.895260214805603, + "num_tokens": 289412500.0, + "step": 7454 + }, + { + "epoch": 0.9483526268922529, + "grad_norm": 2.384227752685547, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8787864446640015, + "num_tokens": 289455800.0, + "step": 7455 + }, + { + "epoch": 0.9484798371708434, + "grad_norm": 2.371220350265503, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8864578008651733, + "num_tokens": 289497863.0, + "step": 7456 + }, + { + "epoch": 0.9486070474494339, + "grad_norm": 2.726929187774658, + "learning_rate": 1e-06, + "loss": 0.3852, + "mean_token_accuracy": 0.8653274178504944, + "num_tokens": 289535044.0, + "step": 7457 + }, + { + "epoch": 0.9487342577280244, + "grad_norm": 2.764472484588623, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8815710544586182, + "num_tokens": 289567483.0, + "step": 7458 + }, + { + "epoch": 0.948861468006615, + "grad_norm": 2.434558153152466, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8972706198692322, + "num_tokens": 289606860.0, + "step": 7459 + }, + { + "epoch": 0.9489886782852054, + "grad_norm": 2.52774715423584, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8736655712127686, + "num_tokens": 289648654.0, + "step": 7460 + }, + { + "epoch": 0.9491158885637959, + "grad_norm": 2.7201452255249023, + "learning_rate": 1e-06, + "loss": 0.355, + "mean_token_accuracy": 0.8785425424575806, + "num_tokens": 289683709.0, + "step": 7461 + }, + { + "epoch": 0.9492430988423864, + "grad_norm": 2.5435147285461426, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8792026042938232, + "num_tokens": 289723199.0, + "step": 7462 + }, + { + "epoch": 0.949370309120977, + "grad_norm": 2.692798614501953, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.8783205151557922, + "num_tokens": 289756692.0, + "step": 7463 + }, + { + "epoch": 0.9494975193995675, + "grad_norm": 2.6348283290863037, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8730902671813965, + "num_tokens": 289799075.0, + "step": 7464 + }, + { + "epoch": 0.949624729678158, + "grad_norm": 2.7277772426605225, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8730589151382446, + "num_tokens": 289834952.0, + "step": 7465 + }, + { + "epoch": 0.9497519399567486, + "grad_norm": 2.6466264724731445, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8750672936439514, + "num_tokens": 289873009.0, + "step": 7466 + }, + { + "epoch": 0.949879150235339, + "grad_norm": 3.000453472137451, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8791204690933228, + "num_tokens": 289912624.0, + "step": 7467 + }, + { + "epoch": 0.9500063605139295, + "grad_norm": 2.6113338470458984, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8818252086639404, + "num_tokens": 289949438.0, + "step": 7468 + }, + { + "epoch": 0.95013357079252, + "grad_norm": 2.499972105026245, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.878275990486145, + "num_tokens": 289989652.0, + "step": 7469 + }, + { + "epoch": 0.9502607810711106, + "grad_norm": 2.5451884269714355, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8966156244277954, + "num_tokens": 290023978.0, + "step": 7470 + }, + { + "epoch": 0.9503879913497011, + "grad_norm": 2.7233502864837646, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.8693312406539917, + "num_tokens": 290062123.0, + "step": 7471 + }, + { + "epoch": 0.9505152016282916, + "grad_norm": 2.7420332431793213, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8878645896911621, + "num_tokens": 290095221.0, + "step": 7472 + }, + { + "epoch": 0.950642411906882, + "grad_norm": 2.429039716720581, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8844711780548096, + "num_tokens": 290136952.0, + "step": 7473 + }, + { + "epoch": 0.9507696221854726, + "grad_norm": 2.3618602752685547, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8823031187057495, + "num_tokens": 290179875.0, + "step": 7474 + }, + { + "epoch": 0.9508968324640631, + "grad_norm": 2.415198802947998, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8783307671546936, + "num_tokens": 290221200.0, + "step": 7475 + }, + { + "epoch": 0.9510240427426536, + "grad_norm": 2.4080727100372314, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8984706401824951, + "num_tokens": 290259357.0, + "step": 7476 + }, + { + "epoch": 0.9511512530212441, + "grad_norm": 2.870867967605591, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8669240474700928, + "num_tokens": 290293171.0, + "step": 7477 + }, + { + "epoch": 0.9512784632998347, + "grad_norm": 2.7006900310516357, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8755884766578674, + "num_tokens": 290329427.0, + "step": 7478 + }, + { + "epoch": 0.9514056735784251, + "grad_norm": 2.5129854679107666, + "learning_rate": 1e-06, + "loss": 0.3453, + "mean_token_accuracy": 0.8784934878349304, + "num_tokens": 290371178.0, + "step": 7479 + }, + { + "epoch": 0.9515328838570156, + "grad_norm": 2.249251127243042, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8873740434646606, + "num_tokens": 290415918.0, + "step": 7480 + }, + { + "epoch": 0.9516600941356061, + "grad_norm": 2.2964184284210205, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8926302194595337, + "num_tokens": 290456791.0, + "step": 7481 + }, + { + "epoch": 0.9517873044141967, + "grad_norm": 2.7254655361175537, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8831876516342163, + "num_tokens": 290491524.0, + "step": 7482 + }, + { + "epoch": 0.9519145146927872, + "grad_norm": 2.2904112339019775, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8939279317855835, + "num_tokens": 290533934.0, + "step": 7483 + }, + { + "epoch": 0.9520417249713777, + "grad_norm": 2.4559807777404785, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8946715593338013, + "num_tokens": 290573287.0, + "step": 7484 + }, + { + "epoch": 0.9521689352499682, + "grad_norm": 2.6179039478302, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8849554061889648, + "num_tokens": 290609257.0, + "step": 7485 + }, + { + "epoch": 0.9522961455285587, + "grad_norm": 2.934992790222168, + "learning_rate": 1e-06, + "loss": 0.4016, + "mean_token_accuracy": 0.8557177186012268, + "num_tokens": 290642271.0, + "step": 7486 + }, + { + "epoch": 0.9524233558071492, + "grad_norm": 2.4735169410705566, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8924425840377808, + "num_tokens": 290682196.0, + "step": 7487 + }, + { + "epoch": 0.9525505660857397, + "grad_norm": 2.8452510833740234, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8826776146888733, + "num_tokens": 290712748.0, + "step": 7488 + }, + { + "epoch": 0.9526777763643303, + "grad_norm": 2.452545404434204, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8748764991760254, + "num_tokens": 290760233.0, + "step": 7489 + }, + { + "epoch": 0.9528049866429208, + "grad_norm": 2.3674771785736084, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.883043646812439, + "num_tokens": 290803996.0, + "step": 7490 + }, + { + "epoch": 0.9529321969215112, + "grad_norm": 2.4741013050079346, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8841660022735596, + "num_tokens": 290844136.0, + "step": 7491 + }, + { + "epoch": 0.9530594072001017, + "grad_norm": 2.7193984985351562, + "learning_rate": 1e-06, + "loss": 0.3819, + "mean_token_accuracy": 0.8661750555038452, + "num_tokens": 290881990.0, + "step": 7492 + }, + { + "epoch": 0.9531866174786923, + "grad_norm": 2.4379665851593018, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8831101655960083, + "num_tokens": 290923020.0, + "step": 7493 + }, + { + "epoch": 0.9533138277572828, + "grad_norm": 2.4751265048980713, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.883125901222229, + "num_tokens": 290964226.0, + "step": 7494 + }, + { + "epoch": 0.9534410380358733, + "grad_norm": 2.24871826171875, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.886053204536438, + "num_tokens": 291011173.0, + "step": 7495 + }, + { + "epoch": 0.9535682483144639, + "grad_norm": 2.505002498626709, + "learning_rate": 1e-06, + "loss": 0.3578, + "mean_token_accuracy": 0.8746821880340576, + "num_tokens": 291052163.0, + "step": 7496 + }, + { + "epoch": 0.9536954585930543, + "grad_norm": 2.4127025604248047, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8817436695098877, + "num_tokens": 291095068.0, + "step": 7497 + }, + { + "epoch": 0.9538226688716448, + "grad_norm": 2.663972854614258, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8814276456832886, + "num_tokens": 291131015.0, + "step": 7498 + }, + { + "epoch": 0.9539498791502353, + "grad_norm": 2.51231050491333, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8810489177703857, + "num_tokens": 291173238.0, + "step": 7499 + }, + { + "epoch": 0.9540770894288259, + "grad_norm": 2.7351973056793213, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8933185935020447, + "num_tokens": 291203092.0, + "step": 7500 + }, + { + "epoch": 0.9542042997074164, + "grad_norm": 2.678743600845337, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8775031566619873, + "num_tokens": 291240517.0, + "step": 7501 + }, + { + "epoch": 0.9543315099860069, + "grad_norm": 2.4268758296966553, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8932228088378906, + "num_tokens": 291277850.0, + "step": 7502 + }, + { + "epoch": 0.9544587202645973, + "grad_norm": 2.6786797046661377, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8844302892684937, + "num_tokens": 291311491.0, + "step": 7503 + }, + { + "epoch": 0.9545859305431879, + "grad_norm": 2.683798313140869, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8811448812484741, + "num_tokens": 291344880.0, + "step": 7504 + }, + { + "epoch": 0.9547131408217784, + "grad_norm": 2.5246031284332275, + "learning_rate": 1e-06, + "loss": 0.3759, + "mean_token_accuracy": 0.8696959018707275, + "num_tokens": 291388506.0, + "step": 7505 + }, + { + "epoch": 0.9548403511003689, + "grad_norm": 2.5016961097717285, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8880566954612732, + "num_tokens": 291427336.0, + "step": 7506 + }, + { + "epoch": 0.9549675613789594, + "grad_norm": 2.4947690963745117, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8808228969573975, + "num_tokens": 291469578.0, + "step": 7507 + }, + { + "epoch": 0.95509477165755, + "grad_norm": 2.574193000793457, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8866879940032959, + "num_tokens": 291506877.0, + "step": 7508 + }, + { + "epoch": 0.9552219819361404, + "grad_norm": 2.581214666366577, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8769821524620056, + "num_tokens": 291545340.0, + "step": 7509 + }, + { + "epoch": 0.9553491922147309, + "grad_norm": 2.6637043952941895, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8827229738235474, + "num_tokens": 291584220.0, + "step": 7510 + }, + { + "epoch": 0.9554764024933214, + "grad_norm": 2.497774839401245, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8853917121887207, + "num_tokens": 291621755.0, + "step": 7511 + }, + { + "epoch": 0.955603612771912, + "grad_norm": 2.599200487136841, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8999953866004944, + "num_tokens": 291653556.0, + "step": 7512 + }, + { + "epoch": 0.9557308230505025, + "grad_norm": 2.9225053787231445, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.878044843673706, + "num_tokens": 291686876.0, + "step": 7513 + }, + { + "epoch": 0.955858033329093, + "grad_norm": 2.608743190765381, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8792921900749207, + "num_tokens": 291722846.0, + "step": 7514 + }, + { + "epoch": 0.9559852436076836, + "grad_norm": 2.442032814025879, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8819082975387573, + "num_tokens": 291762490.0, + "step": 7515 + }, + { + "epoch": 0.956112453886274, + "grad_norm": 3.027769088745117, + "learning_rate": 1e-06, + "loss": 0.3802, + "mean_token_accuracy": 0.860285758972168, + "num_tokens": 291801748.0, + "step": 7516 + }, + { + "epoch": 0.9562396641648645, + "grad_norm": 2.755455255508423, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8815374374389648, + "num_tokens": 291833887.0, + "step": 7517 + }, + { + "epoch": 0.956366874443455, + "grad_norm": 2.7773096561431885, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8838394284248352, + "num_tokens": 291867516.0, + "step": 7518 + }, + { + "epoch": 0.9564940847220456, + "grad_norm": 2.6465964317321777, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8796849250793457, + "num_tokens": 291902858.0, + "step": 7519 + }, + { + "epoch": 0.9566212950006361, + "grad_norm": 2.4864044189453125, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8964628577232361, + "num_tokens": 291937396.0, + "step": 7520 + }, + { + "epoch": 0.9567485052792266, + "grad_norm": 2.6795246601104736, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8823556900024414, + "num_tokens": 291975390.0, + "step": 7521 + }, + { + "epoch": 0.956875715557817, + "grad_norm": 2.484891176223755, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8779579401016235, + "num_tokens": 292017359.0, + "step": 7522 + }, + { + "epoch": 0.9570029258364076, + "grad_norm": 2.670564889907837, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8834708333015442, + "num_tokens": 292052511.0, + "step": 7523 + }, + { + "epoch": 0.9571301361149981, + "grad_norm": 2.811567783355713, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.8719330430030823, + "num_tokens": 292088464.0, + "step": 7524 + }, + { + "epoch": 0.9572573463935886, + "grad_norm": 2.5806944370269775, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8840676546096802, + "num_tokens": 292124616.0, + "step": 7525 + }, + { + "epoch": 0.9573845566721791, + "grad_norm": 2.7051334381103516, + "learning_rate": 1e-06, + "loss": 0.4006, + "mean_token_accuracy": 0.8617864847183228, + "num_tokens": 292163314.0, + "step": 7526 + }, + { + "epoch": 0.9575117669507697, + "grad_norm": 2.592548370361328, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8660378456115723, + "num_tokens": 292207336.0, + "step": 7527 + }, + { + "epoch": 0.9576389772293601, + "grad_norm": 2.4683609008789062, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8773097991943359, + "num_tokens": 292248562.0, + "step": 7528 + }, + { + "epoch": 0.9577661875079506, + "grad_norm": 2.7155165672302246, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.887458860874176, + "num_tokens": 292282440.0, + "step": 7529 + }, + { + "epoch": 0.9578933977865411, + "grad_norm": 2.837404727935791, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8744955062866211, + "num_tokens": 292313480.0, + "step": 7530 + }, + { + "epoch": 0.9580206080651317, + "grad_norm": 2.2852327823638916, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8841285705566406, + "num_tokens": 292358649.0, + "step": 7531 + }, + { + "epoch": 0.9581478183437222, + "grad_norm": 2.399993896484375, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.90071702003479, + "num_tokens": 292393657.0, + "step": 7532 + }, + { + "epoch": 0.9582750286223127, + "grad_norm": 2.651278495788574, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8822543025016785, + "num_tokens": 292428292.0, + "step": 7533 + }, + { + "epoch": 0.9584022389009031, + "grad_norm": 2.4905223846435547, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8745272159576416, + "num_tokens": 292471246.0, + "step": 7534 + }, + { + "epoch": 0.9585294491794937, + "grad_norm": 2.454951286315918, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8903734683990479, + "num_tokens": 292508660.0, + "step": 7535 + }, + { + "epoch": 0.9586566594580842, + "grad_norm": 2.668515682220459, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.896636962890625, + "num_tokens": 292539515.0, + "step": 7536 + }, + { + "epoch": 0.9587838697366747, + "grad_norm": 2.467210054397583, + "learning_rate": 1e-06, + "loss": 0.3655, + "mean_token_accuracy": 0.8760714530944824, + "num_tokens": 292583026.0, + "step": 7537 + }, + { + "epoch": 0.9589110800152653, + "grad_norm": 2.950971841812134, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8816772699356079, + "num_tokens": 292620644.0, + "step": 7538 + }, + { + "epoch": 0.9590382902938558, + "grad_norm": 2.667067527770996, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8850662708282471, + "num_tokens": 292653863.0, + "step": 7539 + }, + { + "epoch": 0.9591655005724462, + "grad_norm": 2.450929880142212, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8768638968467712, + "num_tokens": 292695686.0, + "step": 7540 + }, + { + "epoch": 0.9592927108510367, + "grad_norm": 2.6655967235565186, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8770567178726196, + "num_tokens": 292731935.0, + "step": 7541 + }, + { + "epoch": 0.9594199211296273, + "grad_norm": 2.4966704845428467, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8816375732421875, + "num_tokens": 292771923.0, + "step": 7542 + }, + { + "epoch": 0.9595471314082178, + "grad_norm": 2.7414073944091797, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8900328278541565, + "num_tokens": 292805760.0, + "step": 7543 + }, + { + "epoch": 0.9596743416868083, + "grad_norm": 2.810455322265625, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8801238536834717, + "num_tokens": 292838479.0, + "step": 7544 + }, + { + "epoch": 0.9598015519653988, + "grad_norm": 2.8685739040374756, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8764152526855469, + "num_tokens": 292872789.0, + "step": 7545 + }, + { + "epoch": 0.9599287622439893, + "grad_norm": 2.536094903945923, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8766868114471436, + "num_tokens": 292915088.0, + "step": 7546 + }, + { + "epoch": 0.9600559725225798, + "grad_norm": 2.559016704559326, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8819851875305176, + "num_tokens": 292954467.0, + "step": 7547 + }, + { + "epoch": 0.9601831828011703, + "grad_norm": 2.5185153484344482, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8846092224121094, + "num_tokens": 292993122.0, + "step": 7548 + }, + { + "epoch": 0.9603103930797608, + "grad_norm": 2.516068696975708, + "learning_rate": 1e-06, + "loss": 0.3997, + "mean_token_accuracy": 0.8634954690933228, + "num_tokens": 293041445.0, + "step": 7549 + }, + { + "epoch": 0.9604376033583514, + "grad_norm": 2.5257582664489746, + "learning_rate": 1e-06, + "loss": 0.3789, + "mean_token_accuracy": 0.8671833872795105, + "num_tokens": 293084916.0, + "step": 7550 + }, + { + "epoch": 0.9605648136369419, + "grad_norm": 2.688568353652954, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8786321878433228, + "num_tokens": 293124577.0, + "step": 7551 + }, + { + "epoch": 0.9606920239155323, + "grad_norm": 2.5074639320373535, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8775519728660583, + "num_tokens": 293169672.0, + "step": 7552 + }, + { + "epoch": 0.9608192341941229, + "grad_norm": 2.754422187805176, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8766502141952515, + "num_tokens": 293204982.0, + "step": 7553 + }, + { + "epoch": 0.9609464444727134, + "grad_norm": 2.457704782485962, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8871895670890808, + "num_tokens": 293242920.0, + "step": 7554 + }, + { + "epoch": 0.9610736547513039, + "grad_norm": 2.4043970108032227, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8791205883026123, + "num_tokens": 293286343.0, + "step": 7555 + }, + { + "epoch": 0.9612008650298944, + "grad_norm": 2.6307599544525146, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.882591962814331, + "num_tokens": 293322571.0, + "step": 7556 + }, + { + "epoch": 0.961328075308485, + "grad_norm": 2.4098258018493652, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8880565166473389, + "num_tokens": 293362606.0, + "step": 7557 + }, + { + "epoch": 0.9614552855870754, + "grad_norm": 2.5833611488342285, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.892008364200592, + "num_tokens": 293395138.0, + "step": 7558 + }, + { + "epoch": 0.9615824958656659, + "grad_norm": 2.5066068172454834, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8831215500831604, + "num_tokens": 293435888.0, + "step": 7559 + }, + { + "epoch": 0.9617097061442564, + "grad_norm": 2.69879412651062, + "learning_rate": 1e-06, + "loss": 0.3744, + "mean_token_accuracy": 0.8647879362106323, + "num_tokens": 293475259.0, + "step": 7560 + }, + { + "epoch": 0.961836916422847, + "grad_norm": 2.83967924118042, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8668084144592285, + "num_tokens": 293510486.0, + "step": 7561 + }, + { + "epoch": 0.9619641267014375, + "grad_norm": 2.6904938220977783, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8781006336212158, + "num_tokens": 293542956.0, + "step": 7562 + }, + { + "epoch": 0.962091336980028, + "grad_norm": 2.4966723918914795, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8738078474998474, + "num_tokens": 293588703.0, + "step": 7563 + }, + { + "epoch": 0.9622185472586186, + "grad_norm": 2.4413881301879883, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8757324814796448, + "num_tokens": 293629879.0, + "step": 7564 + }, + { + "epoch": 0.962345757537209, + "grad_norm": 2.57112193107605, + "learning_rate": 1e-06, + "loss": 0.3629, + "mean_token_accuracy": 0.871482253074646, + "num_tokens": 293671609.0, + "step": 7565 + }, + { + "epoch": 0.9624729678157995, + "grad_norm": 2.5704092979431152, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8794527053833008, + "num_tokens": 293709280.0, + "step": 7566 + }, + { + "epoch": 0.96260017809439, + "grad_norm": 2.3967185020446777, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8850114345550537, + "num_tokens": 293753496.0, + "step": 7567 + }, + { + "epoch": 0.9627273883729806, + "grad_norm": 2.443462610244751, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8965772390365601, + "num_tokens": 293789754.0, + "step": 7568 + }, + { + "epoch": 0.9628545986515711, + "grad_norm": 2.6502842903137207, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8823802471160889, + "num_tokens": 293826556.0, + "step": 7569 + }, + { + "epoch": 0.9629818089301616, + "grad_norm": 2.437169313430786, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8881742358207703, + "num_tokens": 293866213.0, + "step": 7570 + }, + { + "epoch": 0.963109019208752, + "grad_norm": 2.3928983211517334, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8840029239654541, + "num_tokens": 293908490.0, + "step": 7571 + }, + { + "epoch": 0.9632362294873426, + "grad_norm": 2.3656649589538574, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8799445033073425, + "num_tokens": 293951255.0, + "step": 7572 + }, + { + "epoch": 0.9633634397659331, + "grad_norm": 2.6049537658691406, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8817810416221619, + "num_tokens": 293986956.0, + "step": 7573 + }, + { + "epoch": 0.9634906500445236, + "grad_norm": 2.4061200618743896, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8830596804618835, + "num_tokens": 294030083.0, + "step": 7574 + }, + { + "epoch": 0.9636178603231141, + "grad_norm": 2.3588578701019287, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8864424824714661, + "num_tokens": 294072388.0, + "step": 7575 + }, + { + "epoch": 0.9637450706017047, + "grad_norm": 2.4343693256378174, + "learning_rate": 1e-06, + "loss": 0.4012, + "mean_token_accuracy": 0.859572172164917, + "num_tokens": 294118771.0, + "step": 7576 + }, + { + "epoch": 0.9638722808802951, + "grad_norm": 2.5047311782836914, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8814195394515991, + "num_tokens": 294157676.0, + "step": 7577 + }, + { + "epoch": 0.9639994911588856, + "grad_norm": 2.790036678314209, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8856335282325745, + "num_tokens": 294190491.0, + "step": 7578 + }, + { + "epoch": 0.9641267014374761, + "grad_norm": 2.586130380630493, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.882209062576294, + "num_tokens": 294227149.0, + "step": 7579 + }, + { + "epoch": 0.9642539117160667, + "grad_norm": 2.469489336013794, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8835209608078003, + "num_tokens": 294264171.0, + "step": 7580 + }, + { + "epoch": 0.9643811219946572, + "grad_norm": 2.5403378009796143, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8956336975097656, + "num_tokens": 294301045.0, + "step": 7581 + }, + { + "epoch": 0.9645083322732477, + "grad_norm": 2.5255157947540283, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8927539587020874, + "num_tokens": 294333702.0, + "step": 7582 + }, + { + "epoch": 0.9646355425518381, + "grad_norm": 2.503969430923462, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8723928928375244, + "num_tokens": 294374909.0, + "step": 7583 + }, + { + "epoch": 0.9647627528304287, + "grad_norm": 2.3553459644317627, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.890727162361145, + "num_tokens": 294413100.0, + "step": 7584 + }, + { + "epoch": 0.9648899631090192, + "grad_norm": 2.5819599628448486, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.879527747631073, + "num_tokens": 294451966.0, + "step": 7585 + }, + { + "epoch": 0.9650171733876097, + "grad_norm": 2.779470443725586, + "learning_rate": 1e-06, + "loss": 0.4046, + "mean_token_accuracy": 0.8571674227714539, + "num_tokens": 294488834.0, + "step": 7586 + }, + { + "epoch": 0.9651443836662003, + "grad_norm": 2.3952372074127197, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8803596496582031, + "num_tokens": 294531146.0, + "step": 7587 + }, + { + "epoch": 0.9652715939447908, + "grad_norm": 2.354229211807251, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8914520740509033, + "num_tokens": 294571580.0, + "step": 7588 + }, + { + "epoch": 0.9653988042233812, + "grad_norm": 2.53171968460083, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8731328248977661, + "num_tokens": 294614054.0, + "step": 7589 + }, + { + "epoch": 0.9655260145019717, + "grad_norm": 2.6031851768493652, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8723597526550293, + "num_tokens": 294652921.0, + "step": 7590 + }, + { + "epoch": 0.9656532247805623, + "grad_norm": 2.5319597721099854, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8906115293502808, + "num_tokens": 294691760.0, + "step": 7591 + }, + { + "epoch": 0.9657804350591528, + "grad_norm": 2.5132651329040527, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8845093846321106, + "num_tokens": 294730465.0, + "step": 7592 + }, + { + "epoch": 0.9659076453377433, + "grad_norm": 2.471688747406006, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8844351172447205, + "num_tokens": 294769104.0, + "step": 7593 + }, + { + "epoch": 0.9660348556163338, + "grad_norm": 2.80996036529541, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8626773357391357, + "num_tokens": 294804738.0, + "step": 7594 + }, + { + "epoch": 0.9661620658949243, + "grad_norm": 2.6421988010406494, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8695498704910278, + "num_tokens": 294841641.0, + "step": 7595 + }, + { + "epoch": 0.9662892761735148, + "grad_norm": 2.6929211616516113, + "learning_rate": 1e-06, + "loss": 0.3798, + "mean_token_accuracy": 0.8639717102050781, + "num_tokens": 294879297.0, + "step": 7596 + }, + { + "epoch": 0.9664164864521053, + "grad_norm": 2.74709415435791, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8787180185317993, + "num_tokens": 294911255.0, + "step": 7597 + }, + { + "epoch": 0.9665436967306958, + "grad_norm": 2.3156285285949707, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8946232795715332, + "num_tokens": 294950436.0, + "step": 7598 + }, + { + "epoch": 0.9666709070092864, + "grad_norm": 2.388965368270874, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8868453502655029, + "num_tokens": 294990241.0, + "step": 7599 + }, + { + "epoch": 0.9667981172878769, + "grad_norm": 2.662853717803955, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8793803453445435, + "num_tokens": 295027088.0, + "step": 7600 + }, + { + "epoch": 0.9669253275664673, + "grad_norm": 2.7813913822174072, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8808183670043945, + "num_tokens": 295060636.0, + "step": 7601 + }, + { + "epoch": 0.9670525378450578, + "grad_norm": 2.7949013710021973, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.881932258605957, + "num_tokens": 295094517.0, + "step": 7602 + }, + { + "epoch": 0.9671797481236484, + "grad_norm": 2.4744515419006348, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8930220603942871, + "num_tokens": 295129568.0, + "step": 7603 + }, + { + "epoch": 0.9673069584022389, + "grad_norm": 2.519652843475342, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8780485391616821, + "num_tokens": 295172059.0, + "step": 7604 + }, + { + "epoch": 0.9674341686808294, + "grad_norm": 2.30049729347229, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8978569507598877, + "num_tokens": 295213641.0, + "step": 7605 + }, + { + "epoch": 0.96756137895942, + "grad_norm": 2.6751670837402344, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8802999258041382, + "num_tokens": 295249139.0, + "step": 7606 + }, + { + "epoch": 0.9676885892380104, + "grad_norm": 2.809903144836426, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.868059515953064, + "num_tokens": 295286514.0, + "step": 7607 + }, + { + "epoch": 0.9678157995166009, + "grad_norm": 2.5261831283569336, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8891711831092834, + "num_tokens": 295324485.0, + "step": 7608 + }, + { + "epoch": 0.9679430097951914, + "grad_norm": 2.35958194732666, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.890882134437561, + "num_tokens": 295365256.0, + "step": 7609 + }, + { + "epoch": 0.968070220073782, + "grad_norm": 2.509892225265503, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8876729011535645, + "num_tokens": 295403176.0, + "step": 7610 + }, + { + "epoch": 0.9681974303523725, + "grad_norm": 2.7578744888305664, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8894206881523132, + "num_tokens": 295435441.0, + "step": 7611 + }, + { + "epoch": 0.968324640630963, + "grad_norm": 2.6999692916870117, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8877966403961182, + "num_tokens": 295469028.0, + "step": 7612 + }, + { + "epoch": 0.9684518509095535, + "grad_norm": 2.689865827560425, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8819330930709839, + "num_tokens": 295503802.0, + "step": 7613 + }, + { + "epoch": 0.968579061188144, + "grad_norm": 2.8338570594787598, + "learning_rate": 1e-06, + "loss": 0.3532, + "mean_token_accuracy": 0.8750291466712952, + "num_tokens": 295539649.0, + "step": 7614 + }, + { + "epoch": 0.9687062714667345, + "grad_norm": 2.44195294380188, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8904813528060913, + "num_tokens": 295580669.0, + "step": 7615 + }, + { + "epoch": 0.968833481745325, + "grad_norm": 2.5195183753967285, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8762954473495483, + "num_tokens": 295618305.0, + "step": 7616 + }, + { + "epoch": 0.9689606920239155, + "grad_norm": 2.413287401199341, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8835389614105225, + "num_tokens": 295658027.0, + "step": 7617 + }, + { + "epoch": 0.9690879023025061, + "grad_norm": 2.5449697971343994, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8795939683914185, + "num_tokens": 295694720.0, + "step": 7618 + }, + { + "epoch": 0.9692151125810966, + "grad_norm": 2.933652877807617, + "learning_rate": 1e-06, + "loss": 0.3951, + "mean_token_accuracy": 0.864248514175415, + "num_tokens": 295728426.0, + "step": 7619 + }, + { + "epoch": 0.969342322859687, + "grad_norm": 2.6175687313079834, + "learning_rate": 1e-06, + "loss": 0.3411, + "mean_token_accuracy": 0.880578875541687, + "num_tokens": 295765844.0, + "step": 7620 + }, + { + "epoch": 0.9694695331382776, + "grad_norm": 2.5493690967559814, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.883138120174408, + "num_tokens": 295804317.0, + "step": 7621 + }, + { + "epoch": 0.9695967434168681, + "grad_norm": 2.402451515197754, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8823263645172119, + "num_tokens": 295845221.0, + "step": 7622 + }, + { + "epoch": 0.9697239536954586, + "grad_norm": 2.5023317337036133, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8812370300292969, + "num_tokens": 295885059.0, + "step": 7623 + }, + { + "epoch": 0.9698511639740491, + "grad_norm": 2.465177059173584, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8818327784538269, + "num_tokens": 295926084.0, + "step": 7624 + }, + { + "epoch": 0.9699783742526397, + "grad_norm": 2.6580874919891357, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8705108165740967, + "num_tokens": 295965495.0, + "step": 7625 + }, + { + "epoch": 0.9701055845312301, + "grad_norm": 2.7542357444763184, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8756488561630249, + "num_tokens": 296000387.0, + "step": 7626 + }, + { + "epoch": 0.9702327948098206, + "grad_norm": 2.369807243347168, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.879204273223877, + "num_tokens": 296043059.0, + "step": 7627 + }, + { + "epoch": 0.9703600050884111, + "grad_norm": 2.7327511310577393, + "learning_rate": 1e-06, + "loss": 0.4068, + "mean_token_accuracy": 0.8625490069389343, + "num_tokens": 296084965.0, + "step": 7628 + }, + { + "epoch": 0.9704872153670017, + "grad_norm": 2.4164910316467285, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8953919410705566, + "num_tokens": 296119658.0, + "step": 7629 + }, + { + "epoch": 0.9706144256455922, + "grad_norm": 2.3837220668792725, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8725465536117554, + "num_tokens": 296165687.0, + "step": 7630 + }, + { + "epoch": 0.9707416359241827, + "grad_norm": 3.2262260913848877, + "learning_rate": 1e-06, + "loss": 0.3794, + "mean_token_accuracy": 0.8713670969009399, + "num_tokens": 296203025.0, + "step": 7631 + }, + { + "epoch": 0.9708688462027731, + "grad_norm": 2.578951597213745, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8822876214981079, + "num_tokens": 296241322.0, + "step": 7632 + }, + { + "epoch": 0.9709960564813637, + "grad_norm": 2.6057071685791016, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8891339302062988, + "num_tokens": 296278265.0, + "step": 7633 + }, + { + "epoch": 0.9711232667599542, + "grad_norm": 2.447831630706787, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8926975131034851, + "num_tokens": 296319162.0, + "step": 7634 + }, + { + "epoch": 0.9712504770385447, + "grad_norm": 3.0364012718200684, + "learning_rate": 1e-06, + "loss": 0.4569, + "mean_token_accuracy": 0.8535134792327881, + "num_tokens": 296353203.0, + "step": 7635 + }, + { + "epoch": 0.9713776873171353, + "grad_norm": 2.68939471244812, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8842992186546326, + "num_tokens": 296388425.0, + "step": 7636 + }, + { + "epoch": 0.9715048975957258, + "grad_norm": 2.334056854248047, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8886678218841553, + "num_tokens": 296429687.0, + "step": 7637 + }, + { + "epoch": 0.9716321078743162, + "grad_norm": 2.573932409286499, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8756186962127686, + "num_tokens": 296469890.0, + "step": 7638 + }, + { + "epoch": 0.9717593181529067, + "grad_norm": 2.6060216426849365, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.870015025138855, + "num_tokens": 296510308.0, + "step": 7639 + }, + { + "epoch": 0.9718865284314973, + "grad_norm": 2.635322332382202, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8685771226882935, + "num_tokens": 296544315.0, + "step": 7640 + }, + { + "epoch": 0.9720137387100878, + "grad_norm": 2.725576162338257, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8811025619506836, + "num_tokens": 296577957.0, + "step": 7641 + }, + { + "epoch": 0.9721409489886783, + "grad_norm": 2.9146971702575684, + "learning_rate": 1e-06, + "loss": 0.3845, + "mean_token_accuracy": 0.8699567317962646, + "num_tokens": 296609271.0, + "step": 7642 + }, + { + "epoch": 0.9722681592672688, + "grad_norm": 2.5389153957366943, + "learning_rate": 1e-06, + "loss": 0.373, + "mean_token_accuracy": 0.8691096901893616, + "num_tokens": 296652167.0, + "step": 7643 + }, + { + "epoch": 0.9723953695458593, + "grad_norm": 2.556535243988037, + "learning_rate": 1e-06, + "loss": 0.383, + "mean_token_accuracy": 0.8661438226699829, + "num_tokens": 296696233.0, + "step": 7644 + }, + { + "epoch": 0.9725225798244498, + "grad_norm": 2.2896804809570312, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8909586668014526, + "num_tokens": 296736239.0, + "step": 7645 + }, + { + "epoch": 0.9726497901030403, + "grad_norm": 2.494771957397461, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8889846205711365, + "num_tokens": 296773848.0, + "step": 7646 + }, + { + "epoch": 0.9727770003816308, + "grad_norm": 2.1222596168518066, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8961175084114075, + "num_tokens": 296815439.0, + "step": 7647 + }, + { + "epoch": 0.9729042106602214, + "grad_norm": 2.5632052421569824, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8905282616615295, + "num_tokens": 296850466.0, + "step": 7648 + }, + { + "epoch": 0.9730314209388119, + "grad_norm": 2.698371648788452, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.889617383480072, + "num_tokens": 296881852.0, + "step": 7649 + }, + { + "epoch": 0.9731586312174023, + "grad_norm": 2.483860969543457, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8879261612892151, + "num_tokens": 296920810.0, + "step": 7650 + }, + { + "epoch": 0.9732858414959928, + "grad_norm": 2.6669728755950928, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8836556077003479, + "num_tokens": 296956568.0, + "step": 7651 + }, + { + "epoch": 0.9734130517745834, + "grad_norm": 2.8446969985961914, + "learning_rate": 1e-06, + "loss": 0.3862, + "mean_token_accuracy": 0.8725986480712891, + "num_tokens": 296993723.0, + "step": 7652 + }, + { + "epoch": 0.9735402620531739, + "grad_norm": 2.6313188076019287, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.891780436038971, + "num_tokens": 297028416.0, + "step": 7653 + }, + { + "epoch": 0.9736674723317644, + "grad_norm": 2.86826753616333, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8793046474456787, + "num_tokens": 297063525.0, + "step": 7654 + }, + { + "epoch": 0.973794682610355, + "grad_norm": 2.483686685562134, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8768476247787476, + "num_tokens": 297102150.0, + "step": 7655 + }, + { + "epoch": 0.9739218928889454, + "grad_norm": 2.815786600112915, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8821359276771545, + "num_tokens": 297135129.0, + "step": 7656 + }, + { + "epoch": 0.9740491031675359, + "grad_norm": 2.729367256164551, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8836120963096619, + "num_tokens": 297168760.0, + "step": 7657 + }, + { + "epoch": 0.9741763134461264, + "grad_norm": 2.550366163253784, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8737010955810547, + "num_tokens": 297206917.0, + "step": 7658 + }, + { + "epoch": 0.974303523724717, + "grad_norm": 2.4170634746551514, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8828547596931458, + "num_tokens": 297251012.0, + "step": 7659 + }, + { + "epoch": 0.9744307340033075, + "grad_norm": 2.5505127906799316, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8941994309425354, + "num_tokens": 297287198.0, + "step": 7660 + }, + { + "epoch": 0.974557944281898, + "grad_norm": 2.6795895099639893, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8844853639602661, + "num_tokens": 297319827.0, + "step": 7661 + }, + { + "epoch": 0.9746851545604885, + "grad_norm": 2.6915483474731445, + "learning_rate": 1e-06, + "loss": 0.4321, + "mean_token_accuracy": 0.8521102070808411, + "num_tokens": 297361371.0, + "step": 7662 + }, + { + "epoch": 0.974812364839079, + "grad_norm": 2.4868295192718506, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8776183724403381, + "num_tokens": 297404080.0, + "step": 7663 + }, + { + "epoch": 0.9749395751176695, + "grad_norm": 2.564432144165039, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8696438074111938, + "num_tokens": 297447269.0, + "step": 7664 + }, + { + "epoch": 0.97506678539626, + "grad_norm": 2.4776930809020996, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8671000599861145, + "num_tokens": 297490039.0, + "step": 7665 + }, + { + "epoch": 0.9751939956748505, + "grad_norm": 2.6366240978240967, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8694458603858948, + "num_tokens": 297528814.0, + "step": 7666 + }, + { + "epoch": 0.9753212059534411, + "grad_norm": 2.486269950866699, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8823907971382141, + "num_tokens": 297566724.0, + "step": 7667 + }, + { + "epoch": 0.9754484162320316, + "grad_norm": 2.825806140899658, + "learning_rate": 1e-06, + "loss": 0.3742, + "mean_token_accuracy": 0.8725261688232422, + "num_tokens": 297600338.0, + "step": 7668 + }, + { + "epoch": 0.975575626510622, + "grad_norm": 2.5397818088531494, + "learning_rate": 1e-06, + "loss": 0.3725, + "mean_token_accuracy": 0.8709362149238586, + "num_tokens": 297638378.0, + "step": 7669 + }, + { + "epoch": 0.9757028367892125, + "grad_norm": 2.719245672225952, + "learning_rate": 1e-06, + "loss": 0.38, + "mean_token_accuracy": 0.8702698945999146, + "num_tokens": 297673737.0, + "step": 7670 + }, + { + "epoch": 0.9758300470678031, + "grad_norm": 2.4982733726501465, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8832237720489502, + "num_tokens": 297710461.0, + "step": 7671 + }, + { + "epoch": 0.9759572573463936, + "grad_norm": 2.4395177364349365, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8828780651092529, + "num_tokens": 297749099.0, + "step": 7672 + }, + { + "epoch": 0.9760844676249841, + "grad_norm": 2.47189998626709, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.880984902381897, + "num_tokens": 297790741.0, + "step": 7673 + }, + { + "epoch": 0.9762116779035747, + "grad_norm": 2.561849355697632, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8869709372520447, + "num_tokens": 297827420.0, + "step": 7674 + }, + { + "epoch": 0.9763388881821651, + "grad_norm": 2.627964973449707, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8832435607910156, + "num_tokens": 297863990.0, + "step": 7675 + }, + { + "epoch": 0.9764660984607556, + "grad_norm": 2.6454808712005615, + "learning_rate": 1e-06, + "loss": 0.3637, + "mean_token_accuracy": 0.8715439438819885, + "num_tokens": 297899203.0, + "step": 7676 + }, + { + "epoch": 0.9765933087393461, + "grad_norm": 2.423191785812378, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8847886919975281, + "num_tokens": 297940207.0, + "step": 7677 + }, + { + "epoch": 0.9767205190179367, + "grad_norm": 2.330967426300049, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8839894533157349, + "num_tokens": 297978336.0, + "step": 7678 + }, + { + "epoch": 0.9768477292965272, + "grad_norm": 2.6851377487182617, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8819611072540283, + "num_tokens": 298012977.0, + "step": 7679 + }, + { + "epoch": 0.9769749395751177, + "grad_norm": 2.723935842514038, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8708021640777588, + "num_tokens": 298048239.0, + "step": 7680 + }, + { + "epoch": 0.9771021498537081, + "grad_norm": 2.5637946128845215, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8909339308738708, + "num_tokens": 298083606.0, + "step": 7681 + }, + { + "epoch": 0.9772293601322987, + "grad_norm": 2.5953147411346436, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.872800350189209, + "num_tokens": 298125612.0, + "step": 7682 + }, + { + "epoch": 0.9773565704108892, + "grad_norm": 2.4361376762390137, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8824523687362671, + "num_tokens": 298163869.0, + "step": 7683 + }, + { + "epoch": 0.9774837806894797, + "grad_norm": 2.6134915351867676, + "learning_rate": 1e-06, + "loss": 0.3978, + "mean_token_accuracy": 0.8662823438644409, + "num_tokens": 298206977.0, + "step": 7684 + }, + { + "epoch": 0.9776109909680702, + "grad_norm": 2.4102108478546143, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9001211524009705, + "num_tokens": 298244852.0, + "step": 7685 + }, + { + "epoch": 0.9777382012466608, + "grad_norm": 2.400753974914551, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8805220127105713, + "num_tokens": 298287335.0, + "step": 7686 + }, + { + "epoch": 0.9778654115252512, + "grad_norm": 2.5366244316101074, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8884938955307007, + "num_tokens": 298329339.0, + "step": 7687 + }, + { + "epoch": 0.9779926218038417, + "grad_norm": 2.864870309829712, + "learning_rate": 1e-06, + "loss": 0.3682, + "mean_token_accuracy": 0.8725678324699402, + "num_tokens": 298363764.0, + "step": 7688 + }, + { + "epoch": 0.9781198320824323, + "grad_norm": 2.293004035949707, + "learning_rate": 1e-06, + "loss": 0.3565, + "mean_token_accuracy": 0.8717898726463318, + "num_tokens": 298409951.0, + "step": 7689 + }, + { + "epoch": 0.9782470423610228, + "grad_norm": 2.5871663093566895, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.891538143157959, + "num_tokens": 298446951.0, + "step": 7690 + }, + { + "epoch": 0.9783742526396133, + "grad_norm": 2.3161978721618652, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8911832571029663, + "num_tokens": 298489019.0, + "step": 7691 + }, + { + "epoch": 0.9785014629182038, + "grad_norm": 2.2910375595092773, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8854832649230957, + "num_tokens": 298532053.0, + "step": 7692 + }, + { + "epoch": 0.9786286731967943, + "grad_norm": 2.4617905616760254, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8847886919975281, + "num_tokens": 298571767.0, + "step": 7693 + }, + { + "epoch": 0.9787558834753848, + "grad_norm": 2.4518065452575684, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8809208273887634, + "num_tokens": 298615499.0, + "step": 7694 + }, + { + "epoch": 0.9788830937539753, + "grad_norm": 2.3440279960632324, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8820679187774658, + "num_tokens": 298657352.0, + "step": 7695 + }, + { + "epoch": 0.9790103040325658, + "grad_norm": 2.4548563957214355, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8655116558074951, + "num_tokens": 298702658.0, + "step": 7696 + }, + { + "epoch": 0.9791375143111564, + "grad_norm": 2.7187793254852295, + "learning_rate": 1e-06, + "loss": 0.3774, + "mean_token_accuracy": 0.8693639039993286, + "num_tokens": 298741403.0, + "step": 7697 + }, + { + "epoch": 0.9792647245897469, + "grad_norm": 2.7150235176086426, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8779222369194031, + "num_tokens": 298775534.0, + "step": 7698 + }, + { + "epoch": 0.9793919348683373, + "grad_norm": 2.5707762241363525, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8787220120429993, + "num_tokens": 298814753.0, + "step": 7699 + }, + { + "epoch": 0.9795191451469278, + "grad_norm": 2.618772029876709, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8726464509963989, + "num_tokens": 298850996.0, + "step": 7700 + }, + { + "epoch": 0.9796463554255184, + "grad_norm": 2.4594531059265137, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8795425891876221, + "num_tokens": 298893952.0, + "step": 7701 + }, + { + "epoch": 0.9797735657041089, + "grad_norm": 2.8539881706237793, + "learning_rate": 1e-06, + "loss": 0.3891, + "mean_token_accuracy": 0.8678945302963257, + "num_tokens": 298930428.0, + "step": 7702 + }, + { + "epoch": 0.9799007759826994, + "grad_norm": 2.531078577041626, + "learning_rate": 1e-06, + "loss": 0.3661, + "mean_token_accuracy": 0.8716228008270264, + "num_tokens": 298974054.0, + "step": 7703 + }, + { + "epoch": 0.98002798626129, + "grad_norm": 2.594270944595337, + "learning_rate": 1e-06, + "loss": 0.378, + "mean_token_accuracy": 0.8713251948356628, + "num_tokens": 299016006.0, + "step": 7704 + }, + { + "epoch": 0.9801551965398804, + "grad_norm": 2.6001288890838623, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.8814495801925659, + "num_tokens": 299052294.0, + "step": 7705 + }, + { + "epoch": 0.9802824068184709, + "grad_norm": 2.2699191570281982, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8901579976081848, + "num_tokens": 299097767.0, + "step": 7706 + }, + { + "epoch": 0.9804096170970614, + "grad_norm": 2.5245158672332764, + "learning_rate": 1e-06, + "loss": 0.36, + "mean_token_accuracy": 0.8734551072120667, + "num_tokens": 299138354.0, + "step": 7707 + }, + { + "epoch": 0.980536827375652, + "grad_norm": 2.526625871658325, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8857564926147461, + "num_tokens": 299178347.0, + "step": 7708 + }, + { + "epoch": 0.9806640376542425, + "grad_norm": 2.223970413208008, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8966896533966064, + "num_tokens": 299219572.0, + "step": 7709 + }, + { + "epoch": 0.980791247932833, + "grad_norm": 2.3345892429351807, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8877121806144714, + "num_tokens": 299268360.0, + "step": 7710 + }, + { + "epoch": 0.9809184582114235, + "grad_norm": 2.525453567504883, + "learning_rate": 1e-06, + "loss": 0.3435, + "mean_token_accuracy": 0.8818077445030212, + "num_tokens": 299309916.0, + "step": 7711 + }, + { + "epoch": 0.981045668490014, + "grad_norm": 2.6690824031829834, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8766089677810669, + "num_tokens": 299348375.0, + "step": 7712 + }, + { + "epoch": 0.9811728787686045, + "grad_norm": 2.609916925430298, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8852482438087463, + "num_tokens": 299383595.0, + "step": 7713 + }, + { + "epoch": 0.981300089047195, + "grad_norm": 2.356252670288086, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8792840242385864, + "num_tokens": 299424718.0, + "step": 7714 + }, + { + "epoch": 0.9814272993257855, + "grad_norm": 2.3998630046844482, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8909898400306702, + "num_tokens": 299461982.0, + "step": 7715 + }, + { + "epoch": 0.9815545096043761, + "grad_norm": 2.4576449394226074, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8823910355567932, + "num_tokens": 299503697.0, + "step": 7716 + }, + { + "epoch": 0.9816817198829666, + "grad_norm": 2.9142534732818604, + "learning_rate": 1e-06, + "loss": 0.417, + "mean_token_accuracy": 0.856926441192627, + "num_tokens": 299538359.0, + "step": 7717 + }, + { + "epoch": 0.981808930161557, + "grad_norm": 2.928250312805176, + "learning_rate": 1e-06, + "loss": 0.3851, + "mean_token_accuracy": 0.8647788763046265, + "num_tokens": 299573449.0, + "step": 7718 + }, + { + "epoch": 0.9819361404401475, + "grad_norm": 2.7020816802978516, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8783940672874451, + "num_tokens": 299609465.0, + "step": 7719 + }, + { + "epoch": 0.9820633507187381, + "grad_norm": 2.384908676147461, + "learning_rate": 1e-06, + "loss": 0.3474, + "mean_token_accuracy": 0.8807478547096252, + "num_tokens": 299653511.0, + "step": 7720 + }, + { + "epoch": 0.9821905609973286, + "grad_norm": 2.5584981441497803, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8950504064559937, + "num_tokens": 299689255.0, + "step": 7721 + }, + { + "epoch": 0.9823177712759191, + "grad_norm": 2.469522714614868, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8829680681228638, + "num_tokens": 299727629.0, + "step": 7722 + }, + { + "epoch": 0.9824449815545097, + "grad_norm": 2.629894971847534, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8845756649971008, + "num_tokens": 299761245.0, + "step": 7723 + }, + { + "epoch": 0.9825721918331001, + "grad_norm": 2.5224344730377197, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8831624388694763, + "num_tokens": 299799539.0, + "step": 7724 + }, + { + "epoch": 0.9826994021116906, + "grad_norm": 2.6781551837921143, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8722233772277832, + "num_tokens": 299841265.0, + "step": 7725 + }, + { + "epoch": 0.9828266123902811, + "grad_norm": 2.2449095249176025, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8817479610443115, + "num_tokens": 299889617.0, + "step": 7726 + }, + { + "epoch": 0.9829538226688717, + "grad_norm": 2.5109903812408447, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.880570650100708, + "num_tokens": 299926356.0, + "step": 7727 + }, + { + "epoch": 0.9830810329474622, + "grad_norm": 2.2965314388275146, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.878888726234436, + "num_tokens": 299972365.0, + "step": 7728 + }, + { + "epoch": 0.9832082432260527, + "grad_norm": 2.4051215648651123, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8760684728622437, + "num_tokens": 300015359.0, + "step": 7729 + }, + { + "epoch": 0.9833354535046431, + "grad_norm": 2.3706259727478027, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8882606029510498, + "num_tokens": 300057255.0, + "step": 7730 + }, + { + "epoch": 0.9834626637832337, + "grad_norm": 2.375767946243286, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.879809558391571, + "num_tokens": 300099243.0, + "step": 7731 + }, + { + "epoch": 0.9835898740618242, + "grad_norm": 2.3372833728790283, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.893746018409729, + "num_tokens": 300136670.0, + "step": 7732 + }, + { + "epoch": 0.9837170843404147, + "grad_norm": 2.4347660541534424, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8902801275253296, + "num_tokens": 300173520.0, + "step": 7733 + }, + { + "epoch": 0.9838442946190052, + "grad_norm": 2.3558084964752197, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8845654129981995, + "num_tokens": 300214564.0, + "step": 7734 + }, + { + "epoch": 0.9839715048975958, + "grad_norm": 2.50726056098938, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8773189783096313, + "num_tokens": 300251323.0, + "step": 7735 + }, + { + "epoch": 0.9840987151761862, + "grad_norm": 2.733696222305298, + "learning_rate": 1e-06, + "loss": 0.3984, + "mean_token_accuracy": 0.8604826927185059, + "num_tokens": 300291680.0, + "step": 7736 + }, + { + "epoch": 0.9842259254547767, + "grad_norm": 2.332085371017456, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8859454989433289, + "num_tokens": 300336770.0, + "step": 7737 + }, + { + "epoch": 0.9843531357333672, + "grad_norm": 2.4663426876068115, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8836457133293152, + "num_tokens": 300378008.0, + "step": 7738 + }, + { + "epoch": 0.9844803460119578, + "grad_norm": 2.357426643371582, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8845816254615784, + "num_tokens": 300420367.0, + "step": 7739 + }, + { + "epoch": 0.9846075562905483, + "grad_norm": 2.6523702144622803, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8886298537254333, + "num_tokens": 300452591.0, + "step": 7740 + }, + { + "epoch": 0.9847347665691388, + "grad_norm": 2.6267688274383545, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.880652666091919, + "num_tokens": 300491121.0, + "step": 7741 + }, + { + "epoch": 0.9848619768477292, + "grad_norm": 2.5499181747436523, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8783745765686035, + "num_tokens": 300532339.0, + "step": 7742 + }, + { + "epoch": 0.9849891871263198, + "grad_norm": 2.5862560272216797, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8721568584442139, + "num_tokens": 300570351.0, + "step": 7743 + }, + { + "epoch": 0.9851163974049103, + "grad_norm": 2.361354112625122, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.884367823600769, + "num_tokens": 300613497.0, + "step": 7744 + }, + { + "epoch": 0.9852436076835008, + "grad_norm": 2.7328789234161377, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8868932723999023, + "num_tokens": 300647623.0, + "step": 7745 + }, + { + "epoch": 0.9853708179620914, + "grad_norm": 2.460340976715088, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8856704831123352, + "num_tokens": 300689494.0, + "step": 7746 + }, + { + "epoch": 0.9854980282406819, + "grad_norm": 2.5432233810424805, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8934653401374817, + "num_tokens": 300725390.0, + "step": 7747 + }, + { + "epoch": 0.9856252385192723, + "grad_norm": 2.396941661834717, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8885909914970398, + "num_tokens": 300764414.0, + "step": 7748 + }, + { + "epoch": 0.9857524487978628, + "grad_norm": 2.5012834072113037, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.892773449420929, + "num_tokens": 300802266.0, + "step": 7749 + }, + { + "epoch": 0.9858796590764534, + "grad_norm": 2.5306336879730225, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8721156716346741, + "num_tokens": 300843663.0, + "step": 7750 + }, + { + "epoch": 0.9860068693550439, + "grad_norm": 2.4876420497894287, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8805185556411743, + "num_tokens": 300880871.0, + "step": 7751 + }, + { + "epoch": 0.9861340796336344, + "grad_norm": 2.7934935092926025, + "learning_rate": 1e-06, + "loss": 0.3942, + "mean_token_accuracy": 0.8661729097366333, + "num_tokens": 300917647.0, + "step": 7752 + }, + { + "epoch": 0.986261289912225, + "grad_norm": 2.4963185787200928, + "learning_rate": 1e-06, + "loss": 0.3554, + "mean_token_accuracy": 0.8735880851745605, + "num_tokens": 300958704.0, + "step": 7753 + }, + { + "epoch": 0.9863885001908154, + "grad_norm": 2.4533557891845703, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8819937705993652, + "num_tokens": 300997483.0, + "step": 7754 + }, + { + "epoch": 0.9865157104694059, + "grad_norm": 2.498385429382324, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8731996417045593, + "num_tokens": 301037492.0, + "step": 7755 + }, + { + "epoch": 0.9866429207479964, + "grad_norm": 2.231415033340454, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8963804841041565, + "num_tokens": 301079520.0, + "step": 7756 + }, + { + "epoch": 0.986770131026587, + "grad_norm": 2.463320016860962, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8777313232421875, + "num_tokens": 301120053.0, + "step": 7757 + }, + { + "epoch": 0.9868973413051775, + "grad_norm": 2.4245195388793945, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8841619491577148, + "num_tokens": 301159837.0, + "step": 7758 + }, + { + "epoch": 0.987024551583768, + "grad_norm": 2.5383360385894775, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8859120607376099, + "num_tokens": 301194564.0, + "step": 7759 + }, + { + "epoch": 0.9871517618623584, + "grad_norm": 2.6735024452209473, + "learning_rate": 1e-06, + "loss": 0.3949, + "mean_token_accuracy": 0.8609121441841125, + "num_tokens": 301232003.0, + "step": 7760 + }, + { + "epoch": 0.987278972140949, + "grad_norm": 2.516615152359009, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8838311433792114, + "num_tokens": 301268907.0, + "step": 7761 + }, + { + "epoch": 0.9874061824195395, + "grad_norm": 2.4288578033447266, + "learning_rate": 1e-06, + "loss": 0.3707, + "mean_token_accuracy": 0.8723164796829224, + "num_tokens": 301309259.0, + "step": 7762 + }, + { + "epoch": 0.98753339269813, + "grad_norm": 2.601146697998047, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8898910284042358, + "num_tokens": 301343951.0, + "step": 7763 + }, + { + "epoch": 0.9876606029767205, + "grad_norm": 2.501453399658203, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8883852362632751, + "num_tokens": 301380584.0, + "step": 7764 + }, + { + "epoch": 0.9877878132553111, + "grad_norm": 2.4603755474090576, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8833227157592773, + "num_tokens": 301420415.0, + "step": 7765 + }, + { + "epoch": 0.9879150235339016, + "grad_norm": 2.709160804748535, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8700908422470093, + "num_tokens": 301457953.0, + "step": 7766 + }, + { + "epoch": 0.988042233812492, + "grad_norm": 2.4465057849884033, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8875818252563477, + "num_tokens": 301494645.0, + "step": 7767 + }, + { + "epoch": 0.9881694440910825, + "grad_norm": 2.5562660694122314, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8774074912071228, + "num_tokens": 301532896.0, + "step": 7768 + }, + { + "epoch": 0.9882966543696731, + "grad_norm": 2.6034958362579346, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8925607204437256, + "num_tokens": 301566195.0, + "step": 7769 + }, + { + "epoch": 0.9884238646482636, + "grad_norm": 2.537569522857666, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8867834806442261, + "num_tokens": 301605382.0, + "step": 7770 + }, + { + "epoch": 0.9885510749268541, + "grad_norm": 2.3409292697906494, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8805774450302124, + "num_tokens": 301648798.0, + "step": 7771 + }, + { + "epoch": 0.9886782852054447, + "grad_norm": 2.4505362510681152, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8809592127799988, + "num_tokens": 301690765.0, + "step": 7772 + }, + { + "epoch": 0.9888054954840351, + "grad_norm": 2.3861935138702393, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8820235729217529, + "num_tokens": 301734929.0, + "step": 7773 + }, + { + "epoch": 0.9889327057626256, + "grad_norm": 2.626711130142212, + "learning_rate": 1e-06, + "loss": 0.3657, + "mean_token_accuracy": 0.8724201321601868, + "num_tokens": 301776120.0, + "step": 7774 + }, + { + "epoch": 0.9890599160412161, + "grad_norm": 2.4144668579101562, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8721928596496582, + "num_tokens": 301817102.0, + "step": 7775 + }, + { + "epoch": 0.9891871263198067, + "grad_norm": 2.7559831142425537, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.874306321144104, + "num_tokens": 301853646.0, + "step": 7776 + }, + { + "epoch": 0.9893143365983972, + "grad_norm": 2.7020938396453857, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8890470862388611, + "num_tokens": 301885459.0, + "step": 7777 + }, + { + "epoch": 0.9894415468769877, + "grad_norm": 2.673448085784912, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8814166188240051, + "num_tokens": 301921283.0, + "step": 7778 + }, + { + "epoch": 0.9895687571555781, + "grad_norm": 2.4982001781463623, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.87809157371521, + "num_tokens": 301964405.0, + "step": 7779 + }, + { + "epoch": 0.9896959674341687, + "grad_norm": 2.4316318035125732, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8860377073287964, + "num_tokens": 302004973.0, + "step": 7780 + }, + { + "epoch": 0.9898231777127592, + "grad_norm": 2.355419874191284, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8734740614891052, + "num_tokens": 302049332.0, + "step": 7781 + }, + { + "epoch": 0.9899503879913497, + "grad_norm": 2.2927794456481934, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8956993818283081, + "num_tokens": 302092015.0, + "step": 7782 + }, + { + "epoch": 0.9900775982699402, + "grad_norm": 2.4925808906555176, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8889092206954956, + "num_tokens": 302125278.0, + "step": 7783 + }, + { + "epoch": 0.9902048085485308, + "grad_norm": 2.4318768978118896, + "learning_rate": 1e-06, + "loss": 0.3552, + "mean_token_accuracy": 0.8789554834365845, + "num_tokens": 302169521.0, + "step": 7784 + }, + { + "epoch": 0.9903320188271212, + "grad_norm": 2.6214725971221924, + "learning_rate": 1e-06, + "loss": 0.3644, + "mean_token_accuracy": 0.8722577095031738, + "num_tokens": 302209410.0, + "step": 7785 + }, + { + "epoch": 0.9904592291057117, + "grad_norm": 2.428131580352783, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.879371702671051, + "num_tokens": 302252175.0, + "step": 7786 + }, + { + "epoch": 0.9905864393843022, + "grad_norm": 2.5514397621154785, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8745546340942383, + "num_tokens": 302289488.0, + "step": 7787 + }, + { + "epoch": 0.9907136496628928, + "grad_norm": 2.6683592796325684, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8817216753959656, + "num_tokens": 302321465.0, + "step": 7788 + }, + { + "epoch": 0.9908408599414833, + "grad_norm": 2.4225215911865234, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8818867206573486, + "num_tokens": 302363904.0, + "step": 7789 + }, + { + "epoch": 0.9909680702200738, + "grad_norm": 2.3957929611206055, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8963553309440613, + "num_tokens": 302405103.0, + "step": 7790 + }, + { + "epoch": 0.9910952804986642, + "grad_norm": 2.4376108646392822, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8776193857192993, + "num_tokens": 302450262.0, + "step": 7791 + }, + { + "epoch": 0.9912224907772548, + "grad_norm": 2.517540693283081, + "learning_rate": 1e-06, + "loss": 0.3595, + "mean_token_accuracy": 0.8746519088745117, + "num_tokens": 302493557.0, + "step": 7792 + }, + { + "epoch": 0.9913497010558453, + "grad_norm": 2.5488603115081787, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8812663555145264, + "num_tokens": 302533032.0, + "step": 7793 + }, + { + "epoch": 0.9914769113344358, + "grad_norm": 2.5042154788970947, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8804826736450195, + "num_tokens": 302574089.0, + "step": 7794 + }, + { + "epoch": 0.9916041216130264, + "grad_norm": 2.603142261505127, + "learning_rate": 1e-06, + "loss": 0.3487, + "mean_token_accuracy": 0.8772946000099182, + "num_tokens": 302615069.0, + "step": 7795 + }, + { + "epoch": 0.9917313318916169, + "grad_norm": 2.3371646404266357, + "learning_rate": 1e-06, + "loss": 0.3493, + "mean_token_accuracy": 0.8772521615028381, + "num_tokens": 302657590.0, + "step": 7796 + }, + { + "epoch": 0.9918585421702073, + "grad_norm": 2.5777595043182373, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.887836217880249, + "num_tokens": 302693029.0, + "step": 7797 + }, + { + "epoch": 0.9919857524487978, + "grad_norm": 2.7572641372680664, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8806489706039429, + "num_tokens": 302729787.0, + "step": 7798 + }, + { + "epoch": 0.9921129627273884, + "grad_norm": 2.560004711151123, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8818937540054321, + "num_tokens": 302769362.0, + "step": 7799 + }, + { + "epoch": 0.9922401730059789, + "grad_norm": 2.512049436569214, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8839801549911499, + "num_tokens": 302807150.0, + "step": 7800 + }, + { + "epoch": 0.9923673832845694, + "grad_norm": 2.2615954875946045, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.876287043094635, + "num_tokens": 302855821.0, + "step": 7801 + }, + { + "epoch": 0.9924945935631599, + "grad_norm": 2.636101007461548, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8761767148971558, + "num_tokens": 302893782.0, + "step": 7802 + }, + { + "epoch": 0.9926218038417504, + "grad_norm": 3.0675363540649414, + "learning_rate": 1e-06, + "loss": 0.3882, + "mean_token_accuracy": 0.8669620752334595, + "num_tokens": 302923285.0, + "step": 7803 + }, + { + "epoch": 0.9927490141203409, + "grad_norm": 2.3915188312530518, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.887553870677948, + "num_tokens": 302964677.0, + "step": 7804 + }, + { + "epoch": 0.9928762243989314, + "grad_norm": 2.7109568119049072, + "learning_rate": 1e-06, + "loss": 0.3541, + "mean_token_accuracy": 0.8760250806808472, + "num_tokens": 302999386.0, + "step": 7805 + }, + { + "epoch": 0.993003434677522, + "grad_norm": 2.4585394859313965, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8818458914756775, + "num_tokens": 303040991.0, + "step": 7806 + }, + { + "epoch": 0.9931306449561125, + "grad_norm": 2.998060464859009, + "learning_rate": 1e-06, + "loss": 0.3639, + "mean_token_accuracy": 0.8704402446746826, + "num_tokens": 303073541.0, + "step": 7807 + }, + { + "epoch": 0.993257855234703, + "grad_norm": 2.715996026992798, + "learning_rate": 1e-06, + "loss": 0.4096, + "mean_token_accuracy": 0.8584914803504944, + "num_tokens": 303114627.0, + "step": 7808 + }, + { + "epoch": 0.9933850655132934, + "grad_norm": 2.404348134994507, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8836573362350464, + "num_tokens": 303154724.0, + "step": 7809 + }, + { + "epoch": 0.993512275791884, + "grad_norm": 2.604128837585449, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.88983154296875, + "num_tokens": 303189028.0, + "step": 7810 + }, + { + "epoch": 0.9936394860704745, + "grad_norm": 2.7408621311187744, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8930361866950989, + "num_tokens": 303217593.0, + "step": 7811 + }, + { + "epoch": 0.993766696349065, + "grad_norm": 2.4009850025177, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8761780858039856, + "num_tokens": 303260729.0, + "step": 7812 + }, + { + "epoch": 0.9938939066276555, + "grad_norm": 2.349081516265869, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8822903037071228, + "num_tokens": 303304999.0, + "step": 7813 + }, + { + "epoch": 0.9940211169062461, + "grad_norm": 2.4525890350341797, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8805460333824158, + "num_tokens": 303349847.0, + "step": 7814 + }, + { + "epoch": 0.9941483271848366, + "grad_norm": 2.899787425994873, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8804302215576172, + "num_tokens": 303386506.0, + "step": 7815 + }, + { + "epoch": 0.994275537463427, + "grad_norm": 2.515979051589966, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8862887620925903, + "num_tokens": 303424237.0, + "step": 7816 + }, + { + "epoch": 0.9944027477420175, + "grad_norm": 2.7515130043029785, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.878376841545105, + "num_tokens": 303463657.0, + "step": 7817 + }, + { + "epoch": 0.9945299580206081, + "grad_norm": 2.5178475379943848, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8891966938972473, + "num_tokens": 303501495.0, + "step": 7818 + }, + { + "epoch": 0.9946571682991986, + "grad_norm": 2.7309341430664062, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8747595548629761, + "num_tokens": 303539070.0, + "step": 7819 + }, + { + "epoch": 0.9947843785777891, + "grad_norm": 2.4602930545806885, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8841382265090942, + "num_tokens": 303577457.0, + "step": 7820 + }, + { + "epoch": 0.9949115888563796, + "grad_norm": 2.3691341876983643, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8992741107940674, + "num_tokens": 303617798.0, + "step": 7821 + }, + { + "epoch": 0.9950387991349701, + "grad_norm": 2.579258918762207, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.885021448135376, + "num_tokens": 303653460.0, + "step": 7822 + }, + { + "epoch": 0.9951660094135606, + "grad_norm": 2.4786550998687744, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8794209957122803, + "num_tokens": 303693558.0, + "step": 7823 + }, + { + "epoch": 0.9952932196921511, + "grad_norm": 2.60927677154541, + "learning_rate": 1e-06, + "loss": 0.3706, + "mean_token_accuracy": 0.8702120184898376, + "num_tokens": 303734868.0, + "step": 7824 + }, + { + "epoch": 0.9954204299707416, + "grad_norm": 2.7562403678894043, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8758764863014221, + "num_tokens": 303768414.0, + "step": 7825 + }, + { + "epoch": 0.9955476402493322, + "grad_norm": 2.5263893604278564, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8835098743438721, + "num_tokens": 303804921.0, + "step": 7826 + }, + { + "epoch": 0.9956748505279227, + "grad_norm": 2.2215654850006104, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8805139660835266, + "num_tokens": 303854503.0, + "step": 7827 + }, + { + "epoch": 0.9958020608065131, + "grad_norm": 2.7104556560516357, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8909687995910645, + "num_tokens": 303885207.0, + "step": 7828 + }, + { + "epoch": 0.9959292710851037, + "grad_norm": 2.3351409435272217, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8839839696884155, + "num_tokens": 303927740.0, + "step": 7829 + }, + { + "epoch": 0.9960564813636942, + "grad_norm": 2.500135660171509, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8758395910263062, + "num_tokens": 303968701.0, + "step": 7830 + }, + { + "epoch": 0.9961836916422847, + "grad_norm": 2.53191876411438, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8741816282272339, + "num_tokens": 304010636.0, + "step": 7831 + }, + { + "epoch": 0.9963109019208752, + "grad_norm": 2.5409092903137207, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.884153425693512, + "num_tokens": 304046321.0, + "step": 7832 + }, + { + "epoch": 0.9964381121994658, + "grad_norm": 2.5493698120117188, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8880515098571777, + "num_tokens": 304083334.0, + "step": 7833 + }, + { + "epoch": 0.9965653224780562, + "grad_norm": 2.4605724811553955, + "learning_rate": 1e-06, + "loss": 0.3679, + "mean_token_accuracy": 0.870551586151123, + "num_tokens": 304126952.0, + "step": 7834 + }, + { + "epoch": 0.9966925327566467, + "grad_norm": 2.5049548149108887, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8819921016693115, + "num_tokens": 304166989.0, + "step": 7835 + }, + { + "epoch": 0.9968197430352372, + "grad_norm": 2.5930428504943848, + "learning_rate": 1e-06, + "loss": 0.3547, + "mean_token_accuracy": 0.8752149939537048, + "num_tokens": 304210208.0, + "step": 7836 + }, + { + "epoch": 0.9969469533138278, + "grad_norm": 2.6430835723876953, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8883415460586548, + "num_tokens": 304245122.0, + "step": 7837 + }, + { + "epoch": 0.9970741635924183, + "grad_norm": 2.5309698581695557, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8741086721420288, + "num_tokens": 304285732.0, + "step": 7838 + }, + { + "epoch": 0.9972013738710088, + "grad_norm": 2.629382610321045, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8855158090591431, + "num_tokens": 304322448.0, + "step": 7839 + }, + { + "epoch": 0.9973285841495992, + "grad_norm": 2.571319103240967, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.873089611530304, + "num_tokens": 304361887.0, + "step": 7840 + }, + { + "epoch": 0.9974557944281898, + "grad_norm": 2.6034562587738037, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8883541226387024, + "num_tokens": 304396005.0, + "step": 7841 + }, + { + "epoch": 0.9975830047067803, + "grad_norm": 2.625523328781128, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.874833345413208, + "num_tokens": 304434828.0, + "step": 7842 + }, + { + "epoch": 0.9977102149853708, + "grad_norm": 2.35748028755188, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8949032425880432, + "num_tokens": 304475545.0, + "step": 7843 + }, + { + "epoch": 0.9978374252639614, + "grad_norm": 2.522799491882324, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8823647499084473, + "num_tokens": 304516710.0, + "step": 7844 + }, + { + "epoch": 0.9979646355425519, + "grad_norm": 2.759222984313965, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8821507692337036, + "num_tokens": 304552227.0, + "step": 7845 + }, + { + "epoch": 0.9980918458211423, + "grad_norm": 2.5353503227233887, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8839368224143982, + "num_tokens": 304590342.0, + "step": 7846 + }, + { + "epoch": 0.9982190560997328, + "grad_norm": 2.4987871646881104, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8763049840927124, + "num_tokens": 304632129.0, + "step": 7847 + }, + { + "epoch": 0.9983462663783234, + "grad_norm": 2.736147403717041, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.878676176071167, + "num_tokens": 304665914.0, + "step": 7848 + }, + { + "epoch": 0.9984734766569139, + "grad_norm": 2.6134965419769287, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8905714750289917, + "num_tokens": 304699086.0, + "step": 7849 + }, + { + "epoch": 0.9986006869355044, + "grad_norm": 2.529045581817627, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8823212385177612, + "num_tokens": 304735043.0, + "step": 7850 + }, + { + "epoch": 0.9987278972140949, + "grad_norm": 2.692746877670288, + "learning_rate": 1e-06, + "loss": 0.3622, + "mean_token_accuracy": 0.8740554451942444, + "num_tokens": 304771470.0, + "step": 7851 + }, + { + "epoch": 0.9988551074926854, + "grad_norm": 2.9485509395599365, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8821388483047485, + "num_tokens": 304803268.0, + "step": 7852 + }, + { + "epoch": 0.9989823177712759, + "grad_norm": 2.7112488746643066, + "learning_rate": 1e-06, + "loss": 0.3897, + "mean_token_accuracy": 0.8670710325241089, + "num_tokens": 304840208.0, + "step": 7853 + }, + { + "epoch": 0.9991095280498664, + "grad_norm": 2.4543569087982178, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8960034251213074, + "num_tokens": 304876285.0, + "step": 7854 + }, + { + "epoch": 0.9992367383284569, + "grad_norm": 2.5276236534118652, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8801218271255493, + "num_tokens": 304913111.0, + "step": 7855 + }, + { + "epoch": 0.9993639486070475, + "grad_norm": 2.387044668197632, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8811159729957581, + "num_tokens": 304954807.0, + "step": 7856 + }, + { + "epoch": 0.999491158885638, + "grad_norm": 2.3839075565338135, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8814615607261658, + "num_tokens": 304995307.0, + "step": 7857 + }, + { + "epoch": 0.9996183691642284, + "grad_norm": 2.1531271934509277, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.895550012588501, + "num_tokens": 305036728.0, + "step": 7858 + }, + { + "epoch": 0.9997455794428189, + "grad_norm": 2.3480613231658936, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8745079040527344, + "num_tokens": 305081247.0, + "step": 7859 + }, + { + "epoch": 0.9998727897214095, + "grad_norm": 2.491703987121582, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8777273893356323, + "num_tokens": 305124331.0, + "step": 7860 + }, + { + "epoch": 1.0, + "grad_norm": 2.439756393432617, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8840318322181702, + "num_tokens": 305165232.0, + "step": 7861 + }, + { + "epoch": 1.0001272102785905, + "grad_norm": 2.3663835525512695, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8957567811012268, + "num_tokens": 305201983.0, + "step": 7862 + }, + { + "epoch": 1.000254420557181, + "grad_norm": 2.2674427032470703, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9016526341438293, + "num_tokens": 305240892.0, + "step": 7863 + }, + { + "epoch": 1.0003816308357716, + "grad_norm": 2.6393702030181885, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8829724788665771, + "num_tokens": 305274167.0, + "step": 7864 + }, + { + "epoch": 1.0005088411143621, + "grad_norm": 2.335589647293091, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8859710097312927, + "num_tokens": 305314699.0, + "step": 7865 + }, + { + "epoch": 1.0006360513929526, + "grad_norm": 2.4910857677459717, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8854085803031921, + "num_tokens": 305355364.0, + "step": 7866 + }, + { + "epoch": 1.0007632616715432, + "grad_norm": 2.358555555343628, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8889503479003906, + "num_tokens": 305398623.0, + "step": 7867 + }, + { + "epoch": 1.0008904719501335, + "grad_norm": 2.597661018371582, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8897712230682373, + "num_tokens": 305435815.0, + "step": 7868 + }, + { + "epoch": 1.001017682228724, + "grad_norm": 2.6595592498779297, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.8808170557022095, + "num_tokens": 305475050.0, + "step": 7869 + }, + { + "epoch": 1.0011448925073145, + "grad_norm": 2.5529448986053467, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9027581214904785, + "num_tokens": 305508933.0, + "step": 7870 + }, + { + "epoch": 1.001272102785905, + "grad_norm": 2.6177539825439453, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8816947937011719, + "num_tokens": 305548013.0, + "step": 7871 + }, + { + "epoch": 1.0013993130644956, + "grad_norm": 2.5570297241210938, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8825660943984985, + "num_tokens": 305595767.0, + "step": 7872 + }, + { + "epoch": 1.0015265233430861, + "grad_norm": 2.556291341781616, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8974586725234985, + "num_tokens": 305638418.0, + "step": 7873 + }, + { + "epoch": 1.0016537336216766, + "grad_norm": 2.779414653778076, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8930313587188721, + "num_tokens": 305674129.0, + "step": 7874 + }, + { + "epoch": 1.0017809439002672, + "grad_norm": 2.6371283531188965, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8822159767150879, + "num_tokens": 305713053.0, + "step": 7875 + }, + { + "epoch": 1.0019081541788577, + "grad_norm": 2.726891040802002, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8885740041732788, + "num_tokens": 305751533.0, + "step": 7876 + }, + { + "epoch": 1.0020353644574482, + "grad_norm": 2.930734395980835, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8859943151473999, + "num_tokens": 305786704.0, + "step": 7877 + }, + { + "epoch": 1.0021625747360388, + "grad_norm": 2.9083104133605957, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.890015184879303, + "num_tokens": 305821249.0, + "step": 7878 + }, + { + "epoch": 1.0022897850146293, + "grad_norm": 2.7834222316741943, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8811255097389221, + "num_tokens": 305859003.0, + "step": 7879 + }, + { + "epoch": 1.0024169952932196, + "grad_norm": 2.79630970954895, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8892263174057007, + "num_tokens": 305893722.0, + "step": 7880 + }, + { + "epoch": 1.0025442055718101, + "grad_norm": 2.4528748989105225, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.893652617931366, + "num_tokens": 305935363.0, + "step": 7881 + }, + { + "epoch": 1.0026714158504006, + "grad_norm": 2.451188802719116, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8888067007064819, + "num_tokens": 305977356.0, + "step": 7882 + }, + { + "epoch": 1.0027986261289912, + "grad_norm": 2.6701319217681885, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8865367770195007, + "num_tokens": 306014848.0, + "step": 7883 + }, + { + "epoch": 1.0029258364075817, + "grad_norm": 2.4947509765625, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9118641018867493, + "num_tokens": 306049981.0, + "step": 7884 + }, + { + "epoch": 1.0030530466861722, + "grad_norm": 3.194253921508789, + "learning_rate": 1e-06, + "loss": 0.4125, + "mean_token_accuracy": 0.8592133522033691, + "num_tokens": 306087254.0, + "step": 7885 + }, + { + "epoch": 1.0031802569647628, + "grad_norm": 2.4846253395080566, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.894810676574707, + "num_tokens": 306126480.0, + "step": 7886 + }, + { + "epoch": 1.0033074672433533, + "grad_norm": 2.479996681213379, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.885484516620636, + "num_tokens": 306171830.0, + "step": 7887 + }, + { + "epoch": 1.0034346775219438, + "grad_norm": 2.650618314743042, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.899746298789978, + "num_tokens": 306207115.0, + "step": 7888 + }, + { + "epoch": 1.0035618878005343, + "grad_norm": 2.6438417434692383, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8916155099868774, + "num_tokens": 306247998.0, + "step": 7889 + }, + { + "epoch": 1.0036890980791249, + "grad_norm": 3.024785041809082, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8871925473213196, + "num_tokens": 306280419.0, + "step": 7890 + }, + { + "epoch": 1.0038163083577154, + "grad_norm": 2.759187936782837, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8955233097076416, + "num_tokens": 306317089.0, + "step": 7891 + }, + { + "epoch": 1.0039435186363057, + "grad_norm": 2.554086208343506, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8989683985710144, + "num_tokens": 306355846.0, + "step": 7892 + }, + { + "epoch": 1.0040707289148962, + "grad_norm": 2.631178617477417, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8884872794151306, + "num_tokens": 306393686.0, + "step": 7893 + }, + { + "epoch": 1.0041979391934868, + "grad_norm": 2.8485910892486572, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8881505727767944, + "num_tokens": 306428407.0, + "step": 7894 + }, + { + "epoch": 1.0043251494720773, + "grad_norm": 2.2472290992736816, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9018576145172119, + "num_tokens": 306474424.0, + "step": 7895 + }, + { + "epoch": 1.0044523597506678, + "grad_norm": 2.599212884902954, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8960591554641724, + "num_tokens": 306509039.0, + "step": 7896 + }, + { + "epoch": 1.0045795700292584, + "grad_norm": 2.4728848934173584, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8932346105575562, + "num_tokens": 306550742.0, + "step": 7897 + }, + { + "epoch": 1.0047067803078489, + "grad_norm": 2.322571039199829, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9055039882659912, + "num_tokens": 306591925.0, + "step": 7898 + }, + { + "epoch": 1.0048339905864394, + "grad_norm": 2.782627820968628, + "learning_rate": 1e-06, + "loss": 0.3684, + "mean_token_accuracy": 0.8717724084854126, + "num_tokens": 306635513.0, + "step": 7899 + }, + { + "epoch": 1.00496120086503, + "grad_norm": 2.6533358097076416, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8941683769226074, + "num_tokens": 306669502.0, + "step": 7900 + }, + { + "epoch": 1.0050884111436205, + "grad_norm": 2.6710312366485596, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9003864526748657, + "num_tokens": 306701657.0, + "step": 7901 + }, + { + "epoch": 1.005215621422211, + "grad_norm": 2.5459389686584473, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8914881944656372, + "num_tokens": 306741674.0, + "step": 7902 + }, + { + "epoch": 1.0053428317008015, + "grad_norm": 2.972630739212036, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8995046615600586, + "num_tokens": 306779926.0, + "step": 7903 + }, + { + "epoch": 1.0054700419793918, + "grad_norm": 2.562143564224243, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8873796463012695, + "num_tokens": 306821411.0, + "step": 7904 + }, + { + "epoch": 1.0055972522579824, + "grad_norm": 2.378347635269165, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8937857151031494, + "num_tokens": 306866198.0, + "step": 7905 + }, + { + "epoch": 1.0057244625365729, + "grad_norm": 2.7785234451293945, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8932734131813049, + "num_tokens": 306901558.0, + "step": 7906 + }, + { + "epoch": 1.0058516728151634, + "grad_norm": 2.6223738193511963, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8889977931976318, + "num_tokens": 306940903.0, + "step": 7907 + }, + { + "epoch": 1.005978883093754, + "grad_norm": 2.6204214096069336, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8953441381454468, + "num_tokens": 306980713.0, + "step": 7908 + }, + { + "epoch": 1.0061060933723445, + "grad_norm": 2.8320329189300537, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8849198818206787, + "num_tokens": 307012904.0, + "step": 7909 + }, + { + "epoch": 1.006233303650935, + "grad_norm": 2.5911574363708496, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8920665383338928, + "num_tokens": 307051596.0, + "step": 7910 + }, + { + "epoch": 1.0063605139295255, + "grad_norm": 2.686415672302246, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.9005458354949951, + "num_tokens": 307086516.0, + "step": 7911 + }, + { + "epoch": 1.006487724208116, + "grad_norm": 2.5612077713012695, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8954383134841919, + "num_tokens": 307124801.0, + "step": 7912 + }, + { + "epoch": 1.0066149344867066, + "grad_norm": 2.518993854522705, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8949652314186096, + "num_tokens": 307163520.0, + "step": 7913 + }, + { + "epoch": 1.006742144765297, + "grad_norm": 2.680450201034546, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8917065858840942, + "num_tokens": 307200208.0, + "step": 7914 + }, + { + "epoch": 1.0068693550438876, + "grad_norm": 2.5051331520080566, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8980265855789185, + "num_tokens": 307239899.0, + "step": 7915 + }, + { + "epoch": 1.0069965653224782, + "grad_norm": 2.531341552734375, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9086987972259521, + "num_tokens": 307275664.0, + "step": 7916 + }, + { + "epoch": 1.0071237756010685, + "grad_norm": 2.7327256202697754, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8930854797363281, + "num_tokens": 307314370.0, + "step": 7917 + }, + { + "epoch": 1.007250985879659, + "grad_norm": 2.579340696334839, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8939231634140015, + "num_tokens": 307351788.0, + "step": 7918 + }, + { + "epoch": 1.0073781961582495, + "grad_norm": 2.939807653427124, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8771871328353882, + "num_tokens": 307387680.0, + "step": 7919 + }, + { + "epoch": 1.00750540643684, + "grad_norm": 3.0108392238616943, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8808370232582092, + "num_tokens": 307423185.0, + "step": 7920 + }, + { + "epoch": 1.0076326167154306, + "grad_norm": 2.6996889114379883, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8820334672927856, + "num_tokens": 307464111.0, + "step": 7921 + }, + { + "epoch": 1.0077598269940211, + "grad_norm": 2.6140637397766113, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8941161036491394, + "num_tokens": 307501094.0, + "step": 7922 + }, + { + "epoch": 1.0078870372726116, + "grad_norm": 2.917628526687622, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8849542140960693, + "num_tokens": 307535348.0, + "step": 7923 + }, + { + "epoch": 1.0080142475512022, + "grad_norm": 2.882441520690918, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8916316032409668, + "num_tokens": 307571416.0, + "step": 7924 + }, + { + "epoch": 1.0081414578297927, + "grad_norm": 2.7620794773101807, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.896719217300415, + "num_tokens": 307607305.0, + "step": 7925 + }, + { + "epoch": 1.0082686681083832, + "grad_norm": 2.436258316040039, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8953791856765747, + "num_tokens": 307654439.0, + "step": 7926 + }, + { + "epoch": 1.0083958783869738, + "grad_norm": 2.625488042831421, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8779875636100769, + "num_tokens": 307697206.0, + "step": 7927 + }, + { + "epoch": 1.0085230886655643, + "grad_norm": 2.578810930252075, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8926153182983398, + "num_tokens": 307733826.0, + "step": 7928 + }, + { + "epoch": 1.0086502989441546, + "grad_norm": 2.7105588912963867, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8861602544784546, + "num_tokens": 307770615.0, + "step": 7929 + }, + { + "epoch": 1.0087775092227451, + "grad_norm": 2.6878068447113037, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8820525407791138, + "num_tokens": 307807332.0, + "step": 7930 + }, + { + "epoch": 1.0089047195013356, + "grad_norm": 2.757631540298462, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8925063014030457, + "num_tokens": 307841633.0, + "step": 7931 + }, + { + "epoch": 1.0090319297799262, + "grad_norm": 2.6737020015716553, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8881387710571289, + "num_tokens": 307878435.0, + "step": 7932 + }, + { + "epoch": 1.0091591400585167, + "grad_norm": 2.599273443222046, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8903065323829651, + "num_tokens": 307917958.0, + "step": 7933 + }, + { + "epoch": 1.0092863503371072, + "grad_norm": 2.4882118701934814, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.899955153465271, + "num_tokens": 307956568.0, + "step": 7934 + }, + { + "epoch": 1.0094135606156978, + "grad_norm": 2.494920253753662, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8952354192733765, + "num_tokens": 307998410.0, + "step": 7935 + }, + { + "epoch": 1.0095407708942883, + "grad_norm": 2.420579433441162, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8927290439605713, + "num_tokens": 308040990.0, + "step": 7936 + }, + { + "epoch": 1.0096679811728788, + "grad_norm": 2.381387948989868, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8850530982017517, + "num_tokens": 308087544.0, + "step": 7937 + }, + { + "epoch": 1.0097951914514693, + "grad_norm": 2.550752878189087, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9060958027839661, + "num_tokens": 308125442.0, + "step": 7938 + }, + { + "epoch": 1.0099224017300599, + "grad_norm": 2.7939670085906982, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8921478986740112, + "num_tokens": 308158212.0, + "step": 7939 + }, + { + "epoch": 1.0100496120086504, + "grad_norm": 2.623614549636841, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8914365172386169, + "num_tokens": 308195041.0, + "step": 7940 + }, + { + "epoch": 1.0101768222872407, + "grad_norm": 2.7684648036956787, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8982374668121338, + "num_tokens": 308226302.0, + "step": 7941 + }, + { + "epoch": 1.0103040325658312, + "grad_norm": 2.3464250564575195, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9035634994506836, + "num_tokens": 308266636.0, + "step": 7942 + }, + { + "epoch": 1.0104312428444218, + "grad_norm": 2.5208864212036133, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.884556770324707, + "num_tokens": 308310963.0, + "step": 7943 + }, + { + "epoch": 1.0105584531230123, + "grad_norm": 2.66219425201416, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8812411427497864, + "num_tokens": 308347717.0, + "step": 7944 + }, + { + "epoch": 1.0106856634016028, + "grad_norm": 2.676138401031494, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8845723271369934, + "num_tokens": 308386751.0, + "step": 7945 + }, + { + "epoch": 1.0108128736801933, + "grad_norm": 2.5914218425750732, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8985930681228638, + "num_tokens": 308423146.0, + "step": 7946 + }, + { + "epoch": 1.0109400839587839, + "grad_norm": 2.525315523147583, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8992129564285278, + "num_tokens": 308462030.0, + "step": 7947 + }, + { + "epoch": 1.0110672942373744, + "grad_norm": 2.6339690685272217, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8825339078903198, + "num_tokens": 308502521.0, + "step": 7948 + }, + { + "epoch": 1.011194504515965, + "grad_norm": 2.746290683746338, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8937026858329773, + "num_tokens": 308536252.0, + "step": 7949 + }, + { + "epoch": 1.0113217147945555, + "grad_norm": 2.622685432434082, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.891819953918457, + "num_tokens": 308575385.0, + "step": 7950 + }, + { + "epoch": 1.011448925073146, + "grad_norm": 2.775668144226074, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8791764378547668, + "num_tokens": 308615189.0, + "step": 7951 + }, + { + "epoch": 1.0115761353517365, + "grad_norm": 2.494783401489258, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8950674533843994, + "num_tokens": 308660055.0, + "step": 7952 + }, + { + "epoch": 1.0117033456303268, + "grad_norm": 2.425950050354004, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8965022563934326, + "num_tokens": 308700538.0, + "step": 7953 + }, + { + "epoch": 1.0118305559089174, + "grad_norm": 2.689600944519043, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8841251134872437, + "num_tokens": 308738531.0, + "step": 7954 + }, + { + "epoch": 1.0119577661875079, + "grad_norm": 2.5441806316375732, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8875936269760132, + "num_tokens": 308776707.0, + "step": 7955 + }, + { + "epoch": 1.0120849764660984, + "grad_norm": 2.5592944622039795, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9014610648155212, + "num_tokens": 308814119.0, + "step": 7956 + }, + { + "epoch": 1.012212186744689, + "grad_norm": 2.7256929874420166, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8792750835418701, + "num_tokens": 308851281.0, + "step": 7957 + }, + { + "epoch": 1.0123393970232795, + "grad_norm": 2.3817248344421387, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9008976221084595, + "num_tokens": 308893001.0, + "step": 7958 + }, + { + "epoch": 1.01246660730187, + "grad_norm": 2.559107780456543, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8880699872970581, + "num_tokens": 308933142.0, + "step": 7959 + }, + { + "epoch": 1.0125938175804605, + "grad_norm": 2.3599295616149902, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8913317918777466, + "num_tokens": 308980304.0, + "step": 7960 + }, + { + "epoch": 1.012721027859051, + "grad_norm": 2.4382033348083496, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.8992620706558228, + "num_tokens": 309021557.0, + "step": 7961 + }, + { + "epoch": 1.0128482381376416, + "grad_norm": 2.562375545501709, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8856923580169678, + "num_tokens": 309065032.0, + "step": 7962 + }, + { + "epoch": 1.012975448416232, + "grad_norm": 2.554150104522705, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8896217942237854, + "num_tokens": 309106741.0, + "step": 7963 + }, + { + "epoch": 1.0131026586948226, + "grad_norm": 2.6082053184509277, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8996337056159973, + "num_tokens": 309144081.0, + "step": 7964 + }, + { + "epoch": 1.0132298689734132, + "grad_norm": 2.7513198852539062, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8877278566360474, + "num_tokens": 309180526.0, + "step": 7965 + }, + { + "epoch": 1.0133570792520035, + "grad_norm": 2.826540231704712, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8778901100158691, + "num_tokens": 309217718.0, + "step": 7966 + }, + { + "epoch": 1.013484289530594, + "grad_norm": 2.5206093788146973, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8924146890640259, + "num_tokens": 309260570.0, + "step": 7967 + }, + { + "epoch": 1.0136114998091845, + "grad_norm": 2.7944228649139404, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8776248097419739, + "num_tokens": 309299238.0, + "step": 7968 + }, + { + "epoch": 1.013738710087775, + "grad_norm": 2.7545835971832275, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.8990216255187988, + "num_tokens": 309332340.0, + "step": 7969 + }, + { + "epoch": 1.0138659203663656, + "grad_norm": 2.5430355072021484, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8814431428909302, + "num_tokens": 309377001.0, + "step": 7970 + }, + { + "epoch": 1.013993130644956, + "grad_norm": 2.6565003395080566, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8767549991607666, + "num_tokens": 309418184.0, + "step": 7971 + }, + { + "epoch": 1.0141203409235466, + "grad_norm": 2.4822981357574463, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9017969369888306, + "num_tokens": 309457514.0, + "step": 7972 + }, + { + "epoch": 1.0142475512021372, + "grad_norm": 2.5481550693511963, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8828953504562378, + "num_tokens": 309499980.0, + "step": 7973 + }, + { + "epoch": 1.0143747614807277, + "grad_norm": 2.7409355640411377, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8871663212776184, + "num_tokens": 309537246.0, + "step": 7974 + }, + { + "epoch": 1.0145019717593182, + "grad_norm": 2.807612657546997, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8939012289047241, + "num_tokens": 309571097.0, + "step": 7975 + }, + { + "epoch": 1.0146291820379088, + "grad_norm": 2.704763174057007, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8767081499099731, + "num_tokens": 309613579.0, + "step": 7976 + }, + { + "epoch": 1.0147563923164993, + "grad_norm": 2.5039734840393066, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9016168117523193, + "num_tokens": 309652011.0, + "step": 7977 + }, + { + "epoch": 1.0148836025950896, + "grad_norm": 2.794201374053955, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8856267333030701, + "num_tokens": 309690280.0, + "step": 7978 + }, + { + "epoch": 1.0150108128736801, + "grad_norm": 2.7559401988983154, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.887405276298523, + "num_tokens": 309727946.0, + "step": 7979 + }, + { + "epoch": 1.0151380231522706, + "grad_norm": 2.396446704864502, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8897804021835327, + "num_tokens": 309773375.0, + "step": 7980 + }, + { + "epoch": 1.0152652334308612, + "grad_norm": 2.6506755352020264, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8932568430900574, + "num_tokens": 309808756.0, + "step": 7981 + }, + { + "epoch": 1.0153924437094517, + "grad_norm": 2.633450746536255, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.905113160610199, + "num_tokens": 309842713.0, + "step": 7982 + }, + { + "epoch": 1.0155196539880422, + "grad_norm": 2.8159680366516113, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8936971426010132, + "num_tokens": 309875924.0, + "step": 7983 + }, + { + "epoch": 1.0156468642666328, + "grad_norm": 2.686082601547241, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8772646188735962, + "num_tokens": 309916724.0, + "step": 7984 + }, + { + "epoch": 1.0157740745452233, + "grad_norm": 2.644453287124634, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8932949304580688, + "num_tokens": 309955205.0, + "step": 7985 + }, + { + "epoch": 1.0159012848238138, + "grad_norm": 3.0525479316711426, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8931280374526978, + "num_tokens": 309984415.0, + "step": 7986 + }, + { + "epoch": 1.0160284951024043, + "grad_norm": 2.50701642036438, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9028276205062866, + "num_tokens": 310023630.0, + "step": 7987 + }, + { + "epoch": 1.0161557053809949, + "grad_norm": 2.5487546920776367, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8935074806213379, + "num_tokens": 310064318.0, + "step": 7988 + }, + { + "epoch": 1.0162829156595854, + "grad_norm": 2.755150079727173, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8903625011444092, + "num_tokens": 310096426.0, + "step": 7989 + }, + { + "epoch": 1.0164101259381757, + "grad_norm": 2.609476089477539, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8882580995559692, + "num_tokens": 310138914.0, + "step": 7990 + }, + { + "epoch": 1.0165373362167662, + "grad_norm": 2.7234699726104736, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8847583532333374, + "num_tokens": 310179005.0, + "step": 7991 + }, + { + "epoch": 1.0166645464953568, + "grad_norm": 2.3850059509277344, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8838272094726562, + "num_tokens": 310225366.0, + "step": 7992 + }, + { + "epoch": 1.0167917567739473, + "grad_norm": 2.7972824573516846, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8863140344619751, + "num_tokens": 310258596.0, + "step": 7993 + }, + { + "epoch": 1.0169189670525378, + "grad_norm": 2.9056475162506104, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8770433664321899, + "num_tokens": 310296324.0, + "step": 7994 + }, + { + "epoch": 1.0170461773311283, + "grad_norm": 2.7999355792999268, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8862929344177246, + "num_tokens": 310331810.0, + "step": 7995 + }, + { + "epoch": 1.0171733876097189, + "grad_norm": 2.7877635955810547, + "learning_rate": 1e-06, + "loss": 0.3562, + "mean_token_accuracy": 0.8736664056777954, + "num_tokens": 310372004.0, + "step": 7996 + }, + { + "epoch": 1.0173005978883094, + "grad_norm": 2.576476812362671, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8926353454589844, + "num_tokens": 310410993.0, + "step": 7997 + }, + { + "epoch": 1.0174278081669, + "grad_norm": 2.6023108959198, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8913207650184631, + "num_tokens": 310451213.0, + "step": 7998 + }, + { + "epoch": 1.0175550184454905, + "grad_norm": 2.541590452194214, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8960395455360413, + "num_tokens": 310491824.0, + "step": 7999 + }, + { + "epoch": 1.017682228724081, + "grad_norm": 2.4641213417053223, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8932436108589172, + "num_tokens": 310532952.0, + "step": 8000 + }, + { + "epoch": 1.0178094390026715, + "grad_norm": 2.4152424335479736, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8993222713470459, + "num_tokens": 310576393.0, + "step": 8001 + }, + { + "epoch": 1.0179366492812618, + "grad_norm": 2.765453815460205, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.898379385471344, + "num_tokens": 310607052.0, + "step": 8002 + }, + { + "epoch": 1.0180638595598523, + "grad_norm": 2.6147241592407227, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9006146192550659, + "num_tokens": 310642616.0, + "step": 8003 + }, + { + "epoch": 1.0181910698384429, + "grad_norm": 2.6910181045532227, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8708626627922058, + "num_tokens": 310683672.0, + "step": 8004 + }, + { + "epoch": 1.0183182801170334, + "grad_norm": 2.6256802082061768, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8916051983833313, + "num_tokens": 310719957.0, + "step": 8005 + }, + { + "epoch": 1.018445490395624, + "grad_norm": 2.7344231605529785, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8959329128265381, + "num_tokens": 310757220.0, + "step": 8006 + }, + { + "epoch": 1.0185727006742145, + "grad_norm": 2.6177542209625244, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8821282386779785, + "num_tokens": 310795623.0, + "step": 8007 + }, + { + "epoch": 1.018699910952805, + "grad_norm": 2.950735092163086, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8903864026069641, + "num_tokens": 310829594.0, + "step": 8008 + }, + { + "epoch": 1.0188271212313955, + "grad_norm": 2.5806472301483154, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8919215202331543, + "num_tokens": 310868009.0, + "step": 8009 + }, + { + "epoch": 1.018954331509986, + "grad_norm": 2.551016330718994, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8901064395904541, + "num_tokens": 310910727.0, + "step": 8010 + }, + { + "epoch": 1.0190815417885766, + "grad_norm": 2.9444961547851562, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8937305212020874, + "num_tokens": 310941540.0, + "step": 8011 + }, + { + "epoch": 1.019208752067167, + "grad_norm": 2.6986422538757324, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8995298743247986, + "num_tokens": 310976692.0, + "step": 8012 + }, + { + "epoch": 1.0193359623457576, + "grad_norm": 2.726365089416504, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8867025971412659, + "num_tokens": 311015250.0, + "step": 8013 + }, + { + "epoch": 1.0194631726243482, + "grad_norm": 2.9653568267822266, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.877099871635437, + "num_tokens": 311053681.0, + "step": 8014 + }, + { + "epoch": 1.0195903829029385, + "grad_norm": 2.7601075172424316, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8954107761383057, + "num_tokens": 311088448.0, + "step": 8015 + }, + { + "epoch": 1.019717593181529, + "grad_norm": 2.54414701461792, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.9001570343971252, + "num_tokens": 311126411.0, + "step": 8016 + }, + { + "epoch": 1.0198448034601195, + "grad_norm": 2.644434690475464, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8904558420181274, + "num_tokens": 311162315.0, + "step": 8017 + }, + { + "epoch": 1.01997201373871, + "grad_norm": 2.5346994400024414, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9002243280410767, + "num_tokens": 311200952.0, + "step": 8018 + }, + { + "epoch": 1.0200992240173006, + "grad_norm": 2.4856069087982178, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8999444842338562, + "num_tokens": 311239978.0, + "step": 8019 + }, + { + "epoch": 1.020226434295891, + "grad_norm": 2.3880112171173096, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.902391791343689, + "num_tokens": 311276587.0, + "step": 8020 + }, + { + "epoch": 1.0203536445744816, + "grad_norm": 2.2681984901428223, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8957659006118774, + "num_tokens": 311324351.0, + "step": 8021 + }, + { + "epoch": 1.0204808548530722, + "grad_norm": 2.6492063999176025, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8870714902877808, + "num_tokens": 311363354.0, + "step": 8022 + }, + { + "epoch": 1.0206080651316627, + "grad_norm": 2.630580425262451, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8859977126121521, + "num_tokens": 311400287.0, + "step": 8023 + }, + { + "epoch": 1.0207352754102532, + "grad_norm": 2.664015054702759, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8979538083076477, + "num_tokens": 311437420.0, + "step": 8024 + }, + { + "epoch": 1.0208624856888437, + "grad_norm": 2.605809450149536, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8954390287399292, + "num_tokens": 311473860.0, + "step": 8025 + }, + { + "epoch": 1.0209896959674343, + "grad_norm": 2.7722225189208984, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8917508125305176, + "num_tokens": 311511129.0, + "step": 8026 + }, + { + "epoch": 1.0211169062460246, + "grad_norm": 2.722674608230591, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8876738548278809, + "num_tokens": 311547561.0, + "step": 8027 + }, + { + "epoch": 1.021244116524615, + "grad_norm": 2.5853257179260254, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.883971095085144, + "num_tokens": 311592704.0, + "step": 8028 + }, + { + "epoch": 1.0213713268032056, + "grad_norm": 2.6974635124206543, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8930373191833496, + "num_tokens": 311631608.0, + "step": 8029 + }, + { + "epoch": 1.0214985370817962, + "grad_norm": 2.8406777381896973, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.888856053352356, + "num_tokens": 311666513.0, + "step": 8030 + }, + { + "epoch": 1.0216257473603867, + "grad_norm": 2.637014150619507, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8971753120422363, + "num_tokens": 311703398.0, + "step": 8031 + }, + { + "epoch": 1.0217529576389772, + "grad_norm": 2.618703842163086, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8967129588127136, + "num_tokens": 311740615.0, + "step": 8032 + }, + { + "epoch": 1.0218801679175677, + "grad_norm": 2.470731735229492, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8949872255325317, + "num_tokens": 311781499.0, + "step": 8033 + }, + { + "epoch": 1.0220073781961583, + "grad_norm": 2.5473339557647705, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8966308236122131, + "num_tokens": 311819143.0, + "step": 8034 + }, + { + "epoch": 1.0221345884747488, + "grad_norm": 2.7198195457458496, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8881911039352417, + "num_tokens": 311859000.0, + "step": 8035 + }, + { + "epoch": 1.0222617987533393, + "grad_norm": 2.621047019958496, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8954907655715942, + "num_tokens": 311895864.0, + "step": 8036 + }, + { + "epoch": 1.0223890090319299, + "grad_norm": 2.688868761062622, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8946882486343384, + "num_tokens": 311932846.0, + "step": 8037 + }, + { + "epoch": 1.0225162193105204, + "grad_norm": 2.735748052597046, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.895549476146698, + "num_tokens": 311970401.0, + "step": 8038 + }, + { + "epoch": 1.0226434295891107, + "grad_norm": 2.5508580207824707, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8857014179229736, + "num_tokens": 312012157.0, + "step": 8039 + }, + { + "epoch": 1.0227706398677012, + "grad_norm": 2.6531291007995605, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8967324495315552, + "num_tokens": 312049929.0, + "step": 8040 + }, + { + "epoch": 1.0228978501462918, + "grad_norm": 2.903125047683716, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8883854150772095, + "num_tokens": 312085047.0, + "step": 8041 + }, + { + "epoch": 1.0230250604248823, + "grad_norm": 2.6303184032440186, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.896903932094574, + "num_tokens": 312123836.0, + "step": 8042 + }, + { + "epoch": 1.0231522707034728, + "grad_norm": 2.5834128856658936, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9024176597595215, + "num_tokens": 312161467.0, + "step": 8043 + }, + { + "epoch": 1.0232794809820633, + "grad_norm": 2.512592077255249, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8828616142272949, + "num_tokens": 312208418.0, + "step": 8044 + }, + { + "epoch": 1.0234066912606539, + "grad_norm": 2.443601608276367, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8821210861206055, + "num_tokens": 312256680.0, + "step": 8045 + }, + { + "epoch": 1.0235339015392444, + "grad_norm": 3.106966495513916, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8817310333251953, + "num_tokens": 312290720.0, + "step": 8046 + }, + { + "epoch": 1.023661111817835, + "grad_norm": 2.7044410705566406, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8901352882385254, + "num_tokens": 312327138.0, + "step": 8047 + }, + { + "epoch": 1.0237883220964255, + "grad_norm": 2.6244313716888428, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.889696478843689, + "num_tokens": 312368827.0, + "step": 8048 + }, + { + "epoch": 1.023915532375016, + "grad_norm": 2.7217917442321777, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.885504961013794, + "num_tokens": 312407356.0, + "step": 8049 + }, + { + "epoch": 1.0240427426536065, + "grad_norm": 2.6903717517852783, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8806260228157043, + "num_tokens": 312450676.0, + "step": 8050 + }, + { + "epoch": 1.0241699529321968, + "grad_norm": 2.2814857959747314, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9111384749412537, + "num_tokens": 312493459.0, + "step": 8051 + }, + { + "epoch": 1.0242971632107873, + "grad_norm": 2.8837010860443115, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8869594931602478, + "num_tokens": 312528595.0, + "step": 8052 + }, + { + "epoch": 1.0244243734893779, + "grad_norm": 2.612687349319458, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.8998972773551941, + "num_tokens": 312568121.0, + "step": 8053 + }, + { + "epoch": 1.0245515837679684, + "grad_norm": 2.4607553482055664, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8933620452880859, + "num_tokens": 312607889.0, + "step": 8054 + }, + { + "epoch": 1.024678794046559, + "grad_norm": 2.6041500568389893, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8931615948677063, + "num_tokens": 312646504.0, + "step": 8055 + }, + { + "epoch": 1.0248060043251495, + "grad_norm": 2.546649694442749, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8869433999061584, + "num_tokens": 312683875.0, + "step": 8056 + }, + { + "epoch": 1.02493321460374, + "grad_norm": 2.598989486694336, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.89265376329422, + "num_tokens": 312727432.0, + "step": 8057 + }, + { + "epoch": 1.0250604248823305, + "grad_norm": 2.631481885910034, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.902748703956604, + "num_tokens": 312762717.0, + "step": 8058 + }, + { + "epoch": 1.025187635160921, + "grad_norm": 2.6381945610046387, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8927939534187317, + "num_tokens": 312799902.0, + "step": 8059 + }, + { + "epoch": 1.0253148454395116, + "grad_norm": 2.618242025375366, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8766194581985474, + "num_tokens": 312843526.0, + "step": 8060 + }, + { + "epoch": 1.025442055718102, + "grad_norm": 2.592829465866089, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8952537775039673, + "num_tokens": 312879330.0, + "step": 8061 + }, + { + "epoch": 1.0255692659966926, + "grad_norm": 2.4780845642089844, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8926264047622681, + "num_tokens": 312921362.0, + "step": 8062 + }, + { + "epoch": 1.0256964762752832, + "grad_norm": 2.723750352859497, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8786313533782959, + "num_tokens": 312961299.0, + "step": 8063 + }, + { + "epoch": 1.0258236865538735, + "grad_norm": 2.601801633834839, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8806577920913696, + "num_tokens": 313003687.0, + "step": 8064 + }, + { + "epoch": 1.025950896832464, + "grad_norm": 2.577096700668335, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8823584318161011, + "num_tokens": 313043362.0, + "step": 8065 + }, + { + "epoch": 1.0260781071110545, + "grad_norm": 2.724916934967041, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8829350471496582, + "num_tokens": 313079923.0, + "step": 8066 + }, + { + "epoch": 1.026205317389645, + "grad_norm": 2.5108108520507812, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8938971757888794, + "num_tokens": 313121738.0, + "step": 8067 + }, + { + "epoch": 1.0263325276682356, + "grad_norm": 2.551828145980835, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8853949308395386, + "num_tokens": 313167461.0, + "step": 8068 + }, + { + "epoch": 1.026459737946826, + "grad_norm": 2.8466367721557617, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8790971636772156, + "num_tokens": 313205454.0, + "step": 8069 + }, + { + "epoch": 1.0265869482254166, + "grad_norm": 2.5484323501586914, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8784657716751099, + "num_tokens": 313247600.0, + "step": 8070 + }, + { + "epoch": 1.0267141585040072, + "grad_norm": 2.4095065593719482, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8886339664459229, + "num_tokens": 313293019.0, + "step": 8071 + }, + { + "epoch": 1.0268413687825977, + "grad_norm": 2.516580104827881, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8969593048095703, + "num_tokens": 313335618.0, + "step": 8072 + }, + { + "epoch": 1.0269685790611882, + "grad_norm": 2.7387964725494385, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.896237313747406, + "num_tokens": 313371941.0, + "step": 8073 + }, + { + "epoch": 1.0270957893397787, + "grad_norm": 2.811732053756714, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8932561874389648, + "num_tokens": 313406753.0, + "step": 8074 + }, + { + "epoch": 1.0272229996183693, + "grad_norm": 2.7809841632843018, + "learning_rate": 1e-06, + "loss": 0.3708, + "mean_token_accuracy": 0.872442901134491, + "num_tokens": 313447526.0, + "step": 8075 + }, + { + "epoch": 1.0273502098969596, + "grad_norm": 2.4760067462921143, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8831480741500854, + "num_tokens": 313491769.0, + "step": 8076 + }, + { + "epoch": 1.02747742017555, + "grad_norm": 2.4886608123779297, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8960279822349548, + "num_tokens": 313532971.0, + "step": 8077 + }, + { + "epoch": 1.0276046304541406, + "grad_norm": 2.555457353591919, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9021070003509521, + "num_tokens": 313570002.0, + "step": 8078 + }, + { + "epoch": 1.0277318407327312, + "grad_norm": 2.497943162918091, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8944869041442871, + "num_tokens": 313611337.0, + "step": 8079 + }, + { + "epoch": 1.0278590510113217, + "grad_norm": 2.7499144077301025, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8773022294044495, + "num_tokens": 313652405.0, + "step": 8080 + }, + { + "epoch": 1.0279862612899122, + "grad_norm": 2.5655977725982666, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9003146886825562, + "num_tokens": 313692652.0, + "step": 8081 + }, + { + "epoch": 1.0281134715685027, + "grad_norm": 2.615694999694824, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8957459926605225, + "num_tokens": 313731436.0, + "step": 8082 + }, + { + "epoch": 1.0282406818470933, + "grad_norm": 2.7778573036193848, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8914705514907837, + "num_tokens": 313766397.0, + "step": 8083 + }, + { + "epoch": 1.0283678921256838, + "grad_norm": 2.424302339553833, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8992483615875244, + "num_tokens": 313806804.0, + "step": 8084 + }, + { + "epoch": 1.0284951024042743, + "grad_norm": 2.534449577331543, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8931297063827515, + "num_tokens": 313847870.0, + "step": 8085 + }, + { + "epoch": 1.0286223126828649, + "grad_norm": 2.5920379161834717, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8904823064804077, + "num_tokens": 313888753.0, + "step": 8086 + }, + { + "epoch": 1.0287495229614554, + "grad_norm": 2.6710424423217773, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8851790428161621, + "num_tokens": 313932611.0, + "step": 8087 + }, + { + "epoch": 1.0288767332400457, + "grad_norm": 2.614163875579834, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8812788128852844, + "num_tokens": 313972471.0, + "step": 8088 + }, + { + "epoch": 1.0290039435186362, + "grad_norm": 2.528728723526001, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8864662647247314, + "num_tokens": 314016465.0, + "step": 8089 + }, + { + "epoch": 1.0291311537972267, + "grad_norm": 2.8362207412719727, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8761917948722839, + "num_tokens": 314054993.0, + "step": 8090 + }, + { + "epoch": 1.0292583640758173, + "grad_norm": 2.6364452838897705, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8881438970565796, + "num_tokens": 314092685.0, + "step": 8091 + }, + { + "epoch": 1.0293855743544078, + "grad_norm": 2.5396173000335693, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8887225985527039, + "num_tokens": 314135644.0, + "step": 8092 + }, + { + "epoch": 1.0295127846329983, + "grad_norm": 2.5788395404815674, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8959047198295593, + "num_tokens": 314174208.0, + "step": 8093 + }, + { + "epoch": 1.0296399949115889, + "grad_norm": 2.379275321960449, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8825103640556335, + "num_tokens": 314220867.0, + "step": 8094 + }, + { + "epoch": 1.0297672051901794, + "grad_norm": 2.845734119415283, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8926717042922974, + "num_tokens": 314254276.0, + "step": 8095 + }, + { + "epoch": 1.02989441546877, + "grad_norm": 2.547882080078125, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8927112221717834, + "num_tokens": 314292486.0, + "step": 8096 + }, + { + "epoch": 1.0300216257473604, + "grad_norm": 2.721127986907959, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8843117952346802, + "num_tokens": 314331756.0, + "step": 8097 + }, + { + "epoch": 1.030148836025951, + "grad_norm": 2.525388240814209, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8873722553253174, + "num_tokens": 314374109.0, + "step": 8098 + }, + { + "epoch": 1.0302760463045415, + "grad_norm": 2.3702898025512695, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9089956283569336, + "num_tokens": 314413852.0, + "step": 8099 + }, + { + "epoch": 1.0304032565831318, + "grad_norm": 2.6023547649383545, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8810251355171204, + "num_tokens": 314455505.0, + "step": 8100 + }, + { + "epoch": 1.0305304668617223, + "grad_norm": 2.6737706661224365, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8951746821403503, + "num_tokens": 314494701.0, + "step": 8101 + }, + { + "epoch": 1.0306576771403129, + "grad_norm": 2.5340635776519775, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8943322896957397, + "num_tokens": 314538036.0, + "step": 8102 + }, + { + "epoch": 1.0307848874189034, + "grad_norm": 2.592247486114502, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8955801725387573, + "num_tokens": 314577723.0, + "step": 8103 + }, + { + "epoch": 1.030912097697494, + "grad_norm": 2.697706937789917, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8893303871154785, + "num_tokens": 314616418.0, + "step": 8104 + }, + { + "epoch": 1.0310393079760845, + "grad_norm": 2.6036198139190674, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8929623365402222, + "num_tokens": 314656339.0, + "step": 8105 + }, + { + "epoch": 1.031166518254675, + "grad_norm": 2.4505088329315186, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.897377610206604, + "num_tokens": 314701497.0, + "step": 8106 + }, + { + "epoch": 1.0312937285332655, + "grad_norm": 2.6596250534057617, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8841606974601746, + "num_tokens": 314743249.0, + "step": 8107 + }, + { + "epoch": 1.031420938811856, + "grad_norm": 3.1241981983184814, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8832440376281738, + "num_tokens": 314775587.0, + "step": 8108 + }, + { + "epoch": 1.0315481490904466, + "grad_norm": 2.483020305633545, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.896007776260376, + "num_tokens": 314818945.0, + "step": 8109 + }, + { + "epoch": 1.031675359369037, + "grad_norm": 2.516728401184082, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8958172798156738, + "num_tokens": 314857167.0, + "step": 8110 + }, + { + "epoch": 1.0318025696476276, + "grad_norm": 2.512270212173462, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8971806764602661, + "num_tokens": 314898934.0, + "step": 8111 + }, + { + "epoch": 1.0319297799262181, + "grad_norm": 2.685124635696411, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8846206665039062, + "num_tokens": 314935636.0, + "step": 8112 + }, + { + "epoch": 1.0320569902048085, + "grad_norm": 2.682875394821167, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8842058181762695, + "num_tokens": 314973775.0, + "step": 8113 + }, + { + "epoch": 1.032184200483399, + "grad_norm": 2.663642168045044, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8897135257720947, + "num_tokens": 315012700.0, + "step": 8114 + }, + { + "epoch": 1.0323114107619895, + "grad_norm": 2.8339338302612305, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8952507376670837, + "num_tokens": 315046433.0, + "step": 8115 + }, + { + "epoch": 1.03243862104058, + "grad_norm": 2.916170120239258, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8919554948806763, + "num_tokens": 315077262.0, + "step": 8116 + }, + { + "epoch": 1.0325658313191706, + "grad_norm": 2.5579605102539062, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8838446140289307, + "num_tokens": 315118486.0, + "step": 8117 + }, + { + "epoch": 1.032693041597761, + "grad_norm": 2.8856208324432373, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.8679543733596802, + "num_tokens": 315158714.0, + "step": 8118 + }, + { + "epoch": 1.0328202518763516, + "grad_norm": 2.650289535522461, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8937740325927734, + "num_tokens": 315194987.0, + "step": 8119 + }, + { + "epoch": 1.0329474621549422, + "grad_norm": 2.6642425060272217, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8933960199356079, + "num_tokens": 315230627.0, + "step": 8120 + }, + { + "epoch": 1.0330746724335327, + "grad_norm": 2.60857892036438, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8776835203170776, + "num_tokens": 315275992.0, + "step": 8121 + }, + { + "epoch": 1.0332018827121232, + "grad_norm": 2.9981985092163086, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.889320969581604, + "num_tokens": 315311037.0, + "step": 8122 + }, + { + "epoch": 1.0333290929907137, + "grad_norm": 2.760038375854492, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9006487131118774, + "num_tokens": 315343571.0, + "step": 8123 + }, + { + "epoch": 1.0334563032693043, + "grad_norm": 2.3829290866851807, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8887722492218018, + "num_tokens": 315389000.0, + "step": 8124 + }, + { + "epoch": 1.0335835135478946, + "grad_norm": 2.3606889247894287, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9048107862472534, + "num_tokens": 315432083.0, + "step": 8125 + }, + { + "epoch": 1.033710723826485, + "grad_norm": 2.612534284591675, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8802170753479004, + "num_tokens": 315476497.0, + "step": 8126 + }, + { + "epoch": 1.0338379341050756, + "grad_norm": 2.7152152061462402, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8872158527374268, + "num_tokens": 315514597.0, + "step": 8127 + }, + { + "epoch": 1.0339651443836662, + "grad_norm": 2.6262190341949463, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8912299275398254, + "num_tokens": 315556065.0, + "step": 8128 + }, + { + "epoch": 1.0340923546622567, + "grad_norm": 2.3569114208221436, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8946433067321777, + "num_tokens": 315600406.0, + "step": 8129 + }, + { + "epoch": 1.0342195649408472, + "grad_norm": 2.531970739364624, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8943707942962646, + "num_tokens": 315644597.0, + "step": 8130 + }, + { + "epoch": 1.0343467752194377, + "grad_norm": 3.0043163299560547, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8800381422042847, + "num_tokens": 315677847.0, + "step": 8131 + }, + { + "epoch": 1.0344739854980283, + "grad_norm": 2.6688733100891113, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8783179521560669, + "num_tokens": 315720792.0, + "step": 8132 + }, + { + "epoch": 1.0346011957766188, + "grad_norm": 2.882354259490967, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8766506314277649, + "num_tokens": 315761320.0, + "step": 8133 + }, + { + "epoch": 1.0347284060552093, + "grad_norm": 2.8922178745269775, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8834743499755859, + "num_tokens": 315800172.0, + "step": 8134 + }, + { + "epoch": 1.0348556163337999, + "grad_norm": 2.6492016315460205, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8838655352592468, + "num_tokens": 315837862.0, + "step": 8135 + }, + { + "epoch": 1.0349828266123904, + "grad_norm": 2.385082244873047, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8949746489524841, + "num_tokens": 315882338.0, + "step": 8136 + }, + { + "epoch": 1.0351100368909807, + "grad_norm": 2.449801206588745, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8929368257522583, + "num_tokens": 315925125.0, + "step": 8137 + }, + { + "epoch": 1.0352372471695712, + "grad_norm": 2.9164202213287354, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8930820822715759, + "num_tokens": 315956880.0, + "step": 8138 + }, + { + "epoch": 1.0353644574481617, + "grad_norm": 2.781677722930908, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9075838923454285, + "num_tokens": 315989693.0, + "step": 8139 + }, + { + "epoch": 1.0354916677267523, + "grad_norm": 2.5420331954956055, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8991698026657104, + "num_tokens": 316028269.0, + "step": 8140 + }, + { + "epoch": 1.0356188780053428, + "grad_norm": 2.8556790351867676, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8874934911727905, + "num_tokens": 316066525.0, + "step": 8141 + }, + { + "epoch": 1.0357460882839333, + "grad_norm": 2.7921032905578613, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9018664360046387, + "num_tokens": 316099250.0, + "step": 8142 + }, + { + "epoch": 1.0358732985625239, + "grad_norm": 2.733184576034546, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8950008153915405, + "num_tokens": 316135912.0, + "step": 8143 + }, + { + "epoch": 1.0360005088411144, + "grad_norm": 2.9118518829345703, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8844554424285889, + "num_tokens": 316169710.0, + "step": 8144 + }, + { + "epoch": 1.036127719119705, + "grad_norm": 2.78098726272583, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8959159851074219, + "num_tokens": 316202633.0, + "step": 8145 + }, + { + "epoch": 1.0362549293982954, + "grad_norm": 2.8457581996917725, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8921721577644348, + "num_tokens": 316237818.0, + "step": 8146 + }, + { + "epoch": 1.036382139676886, + "grad_norm": 3.063232421875, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8837523460388184, + "num_tokens": 316272170.0, + "step": 8147 + }, + { + "epoch": 1.0365093499554765, + "grad_norm": 2.4976606369018555, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8944194316864014, + "num_tokens": 316314646.0, + "step": 8148 + }, + { + "epoch": 1.0366365602340668, + "grad_norm": 2.7448346614837646, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.894996166229248, + "num_tokens": 316351989.0, + "step": 8149 + }, + { + "epoch": 1.0367637705126573, + "grad_norm": 2.49660587310791, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8963433504104614, + "num_tokens": 316394348.0, + "step": 8150 + }, + { + "epoch": 1.0368909807912479, + "grad_norm": 2.4807353019714355, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8946565389633179, + "num_tokens": 316440057.0, + "step": 8151 + }, + { + "epoch": 1.0370181910698384, + "grad_norm": 2.6513419151306152, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8846524953842163, + "num_tokens": 316481056.0, + "step": 8152 + }, + { + "epoch": 1.037145401348429, + "grad_norm": 2.8712003231048584, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8853878974914551, + "num_tokens": 316517211.0, + "step": 8153 + }, + { + "epoch": 1.0372726116270194, + "grad_norm": 2.8524625301361084, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8792583346366882, + "num_tokens": 316553358.0, + "step": 8154 + }, + { + "epoch": 1.03739982190561, + "grad_norm": 2.4721107482910156, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9021070003509521, + "num_tokens": 316592432.0, + "step": 8155 + }, + { + "epoch": 1.0375270321842005, + "grad_norm": 2.448322057723999, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8966235518455505, + "num_tokens": 316634052.0, + "step": 8156 + }, + { + "epoch": 1.037654242462791, + "grad_norm": 2.795363426208496, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8912101984024048, + "num_tokens": 316669364.0, + "step": 8157 + }, + { + "epoch": 1.0377814527413816, + "grad_norm": 2.6472127437591553, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8926849365234375, + "num_tokens": 316706670.0, + "step": 8158 + }, + { + "epoch": 1.037908663019972, + "grad_norm": 2.6453309059143066, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8890445232391357, + "num_tokens": 316744961.0, + "step": 8159 + }, + { + "epoch": 1.0380358732985626, + "grad_norm": 2.5645699501037598, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8840086460113525, + "num_tokens": 316785918.0, + "step": 8160 + }, + { + "epoch": 1.0381630835771531, + "grad_norm": 2.7365598678588867, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8844009041786194, + "num_tokens": 316823495.0, + "step": 8161 + }, + { + "epoch": 1.0382902938557435, + "grad_norm": 2.5944933891296387, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8987727165222168, + "num_tokens": 316863458.0, + "step": 8162 + }, + { + "epoch": 1.038417504134334, + "grad_norm": 2.2951064109802246, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9101470708847046, + "num_tokens": 316905276.0, + "step": 8163 + }, + { + "epoch": 1.0385447144129245, + "grad_norm": 2.439803123474121, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8869889974594116, + "num_tokens": 316946386.0, + "step": 8164 + }, + { + "epoch": 1.038671924691515, + "grad_norm": 2.5571329593658447, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.902263343334198, + "num_tokens": 316981549.0, + "step": 8165 + }, + { + "epoch": 1.0387991349701056, + "grad_norm": 2.4886841773986816, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8964333534240723, + "num_tokens": 317025626.0, + "step": 8166 + }, + { + "epoch": 1.038926345248696, + "grad_norm": 2.752619981765747, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8868602514266968, + "num_tokens": 317060421.0, + "step": 8167 + }, + { + "epoch": 1.0390535555272866, + "grad_norm": 2.8067245483398438, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.8650161027908325, + "num_tokens": 317099061.0, + "step": 8168 + }, + { + "epoch": 1.0391807658058771, + "grad_norm": 2.578315019607544, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8982657194137573, + "num_tokens": 317135850.0, + "step": 8169 + }, + { + "epoch": 1.0393079760844677, + "grad_norm": 2.554180860519409, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8975106477737427, + "num_tokens": 317174447.0, + "step": 8170 + }, + { + "epoch": 1.0394351863630582, + "grad_norm": 2.5411462783813477, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8801223039627075, + "num_tokens": 317218533.0, + "step": 8171 + }, + { + "epoch": 1.0395623966416487, + "grad_norm": 2.9032223224639893, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8989914655685425, + "num_tokens": 317246973.0, + "step": 8172 + }, + { + "epoch": 1.0396896069202393, + "grad_norm": 2.6202170848846436, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8961213827133179, + "num_tokens": 317283265.0, + "step": 8173 + }, + { + "epoch": 1.0398168171988296, + "grad_norm": 2.792694568634033, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.89638352394104, + "num_tokens": 317316467.0, + "step": 8174 + }, + { + "epoch": 1.03994402747742, + "grad_norm": 2.6838691234588623, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8821468353271484, + "num_tokens": 317356942.0, + "step": 8175 + }, + { + "epoch": 1.0400712377560106, + "grad_norm": 2.627580165863037, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8800128698348999, + "num_tokens": 317401145.0, + "step": 8176 + }, + { + "epoch": 1.0401984480346012, + "grad_norm": 2.664919376373291, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8946149349212646, + "num_tokens": 317439553.0, + "step": 8177 + }, + { + "epoch": 1.0403256583131917, + "grad_norm": 2.661278009414673, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8744410276412964, + "num_tokens": 317482422.0, + "step": 8178 + }, + { + "epoch": 1.0404528685917822, + "grad_norm": 2.417067289352417, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.897356390953064, + "num_tokens": 317525046.0, + "step": 8179 + }, + { + "epoch": 1.0405800788703727, + "grad_norm": 2.730020523071289, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8942959308624268, + "num_tokens": 317558771.0, + "step": 8180 + }, + { + "epoch": 1.0407072891489633, + "grad_norm": 2.8481316566467285, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8936253190040588, + "num_tokens": 317591802.0, + "step": 8181 + }, + { + "epoch": 1.0408344994275538, + "grad_norm": 2.331758737564087, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.901152491569519, + "num_tokens": 317634820.0, + "step": 8182 + }, + { + "epoch": 1.0409617097061443, + "grad_norm": 2.4596590995788574, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.900101363658905, + "num_tokens": 317677231.0, + "step": 8183 + }, + { + "epoch": 1.0410889199847349, + "grad_norm": 2.652170181274414, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8905701637268066, + "num_tokens": 317715370.0, + "step": 8184 + }, + { + "epoch": 1.0412161302633254, + "grad_norm": 2.7800352573394775, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8892165422439575, + "num_tokens": 317753154.0, + "step": 8185 + }, + { + "epoch": 1.0413433405419157, + "grad_norm": 2.652935266494751, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8957377076148987, + "num_tokens": 317790258.0, + "step": 8186 + }, + { + "epoch": 1.0414705508205062, + "grad_norm": 2.878129482269287, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8887515068054199, + "num_tokens": 317824154.0, + "step": 8187 + }, + { + "epoch": 1.0415977610990967, + "grad_norm": 2.6715567111968994, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8892029523849487, + "num_tokens": 317865087.0, + "step": 8188 + }, + { + "epoch": 1.0417249713776873, + "grad_norm": 2.4138596057891846, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.9034082889556885, + "num_tokens": 317907843.0, + "step": 8189 + }, + { + "epoch": 1.0418521816562778, + "grad_norm": 2.4744436740875244, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9077098369598389, + "num_tokens": 317948384.0, + "step": 8190 + }, + { + "epoch": 1.0419793919348683, + "grad_norm": 2.564880132675171, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8959367275238037, + "num_tokens": 317989891.0, + "step": 8191 + }, + { + "epoch": 1.0421066022134589, + "grad_norm": 2.7036051750183105, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8903824687004089, + "num_tokens": 318025356.0, + "step": 8192 + }, + { + "epoch": 1.0422338124920494, + "grad_norm": 2.6282083988189697, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8897640109062195, + "num_tokens": 318065978.0, + "step": 8193 + }, + { + "epoch": 1.04236102277064, + "grad_norm": 2.5203709602355957, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8982841372489929, + "num_tokens": 318106130.0, + "step": 8194 + }, + { + "epoch": 1.0424882330492304, + "grad_norm": 2.5547807216644287, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9067850708961487, + "num_tokens": 318141626.0, + "step": 8195 + }, + { + "epoch": 1.042615443327821, + "grad_norm": 2.563793420791626, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8956271409988403, + "num_tokens": 318181883.0, + "step": 8196 + }, + { + "epoch": 1.0427426536064115, + "grad_norm": 2.63934063911438, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8903024792671204, + "num_tokens": 318226110.0, + "step": 8197 + }, + { + "epoch": 1.0428698638850018, + "grad_norm": 2.879940986633301, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8851509094238281, + "num_tokens": 318264031.0, + "step": 8198 + }, + { + "epoch": 1.0429970741635923, + "grad_norm": 2.7915866374969482, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8973002433776855, + "num_tokens": 318300015.0, + "step": 8199 + }, + { + "epoch": 1.0431242844421829, + "grad_norm": 2.8830630779266357, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8874715566635132, + "num_tokens": 318337547.0, + "step": 8200 + }, + { + "epoch": 1.0432514947207734, + "grad_norm": 2.8663461208343506, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8862462639808655, + "num_tokens": 318374745.0, + "step": 8201 + }, + { + "epoch": 1.043378704999364, + "grad_norm": 2.583395481109619, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8982938528060913, + "num_tokens": 318413713.0, + "step": 8202 + }, + { + "epoch": 1.0435059152779544, + "grad_norm": 2.692315101623535, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8943369388580322, + "num_tokens": 318452008.0, + "step": 8203 + }, + { + "epoch": 1.043633125556545, + "grad_norm": 3.1391026973724365, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8793286085128784, + "num_tokens": 318487253.0, + "step": 8204 + }, + { + "epoch": 1.0437603358351355, + "grad_norm": 2.6546366214752197, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9029942750930786, + "num_tokens": 318524927.0, + "step": 8205 + }, + { + "epoch": 1.043887546113726, + "grad_norm": 3.1951053142547607, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8792921304702759, + "num_tokens": 318555178.0, + "step": 8206 + }, + { + "epoch": 1.0440147563923166, + "grad_norm": 3.1257896423339844, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8809252381324768, + "num_tokens": 318584599.0, + "step": 8207 + }, + { + "epoch": 1.044141966670907, + "grad_norm": 2.6263267993927, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8876611590385437, + "num_tokens": 318622529.0, + "step": 8208 + }, + { + "epoch": 1.0442691769494976, + "grad_norm": 2.805603504180908, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8910472393035889, + "num_tokens": 318655526.0, + "step": 8209 + }, + { + "epoch": 1.0443963872280881, + "grad_norm": 2.7338247299194336, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.87786465883255, + "num_tokens": 318697149.0, + "step": 8210 + }, + { + "epoch": 1.0445235975066784, + "grad_norm": 2.8318567276000977, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8819212317466736, + "num_tokens": 318734291.0, + "step": 8211 + }, + { + "epoch": 1.044650807785269, + "grad_norm": 2.5850441455841064, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.893195629119873, + "num_tokens": 318770740.0, + "step": 8212 + }, + { + "epoch": 1.0447780180638595, + "grad_norm": 2.410498857498169, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9016507863998413, + "num_tokens": 318810912.0, + "step": 8213 + }, + { + "epoch": 1.04490522834245, + "grad_norm": 2.5671448707580566, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9010924100875854, + "num_tokens": 318850303.0, + "step": 8214 + }, + { + "epoch": 1.0450324386210406, + "grad_norm": 2.4267539978027344, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8893033862113953, + "num_tokens": 318895521.0, + "step": 8215 + }, + { + "epoch": 1.045159648899631, + "grad_norm": 2.6055283546447754, + "learning_rate": 1e-06, + "loss": 0.3548, + "mean_token_accuracy": 0.8711100220680237, + "num_tokens": 318942181.0, + "step": 8216 + }, + { + "epoch": 1.0452868591782216, + "grad_norm": 2.7485482692718506, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9015607833862305, + "num_tokens": 318976299.0, + "step": 8217 + }, + { + "epoch": 1.0454140694568121, + "grad_norm": 2.974752426147461, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8828455209732056, + "num_tokens": 319011945.0, + "step": 8218 + }, + { + "epoch": 1.0455412797354027, + "grad_norm": 2.476175546646118, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9013321995735168, + "num_tokens": 319052624.0, + "step": 8219 + }, + { + "epoch": 1.0456684900139932, + "grad_norm": 2.752391815185547, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8847936391830444, + "num_tokens": 319091714.0, + "step": 8220 + }, + { + "epoch": 1.0457957002925837, + "grad_norm": 2.5651133060455322, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.9112024307250977, + "num_tokens": 319124080.0, + "step": 8221 + }, + { + "epoch": 1.0459229105711743, + "grad_norm": 2.76798939704895, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8883457779884338, + "num_tokens": 319164363.0, + "step": 8222 + }, + { + "epoch": 1.0460501208497646, + "grad_norm": 2.4726414680480957, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8866593241691589, + "num_tokens": 319210080.0, + "step": 8223 + }, + { + "epoch": 1.046177331128355, + "grad_norm": 2.6129517555236816, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.893947958946228, + "num_tokens": 319250906.0, + "step": 8224 + }, + { + "epoch": 1.0463045414069456, + "grad_norm": 2.6780917644500732, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.890454888343811, + "num_tokens": 319291677.0, + "step": 8225 + }, + { + "epoch": 1.0464317516855361, + "grad_norm": 2.546140670776367, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8899335265159607, + "num_tokens": 319333293.0, + "step": 8226 + }, + { + "epoch": 1.0465589619641267, + "grad_norm": 2.6528308391571045, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8883199691772461, + "num_tokens": 319376122.0, + "step": 8227 + }, + { + "epoch": 1.0466861722427172, + "grad_norm": 3.217379570007324, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8871890902519226, + "num_tokens": 319406673.0, + "step": 8228 + }, + { + "epoch": 1.0468133825213077, + "grad_norm": 2.7005746364593506, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8996908664703369, + "num_tokens": 319440837.0, + "step": 8229 + }, + { + "epoch": 1.0469405927998983, + "grad_norm": 2.606426477432251, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8978371620178223, + "num_tokens": 319481173.0, + "step": 8230 + }, + { + "epoch": 1.0470678030784888, + "grad_norm": 2.8846216201782227, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8784353733062744, + "num_tokens": 319517901.0, + "step": 8231 + }, + { + "epoch": 1.0471950133570793, + "grad_norm": 2.2836992740631104, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9034275412559509, + "num_tokens": 319563031.0, + "step": 8232 + }, + { + "epoch": 1.0473222236356698, + "grad_norm": 2.7425055503845215, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8910772800445557, + "num_tokens": 319596415.0, + "step": 8233 + }, + { + "epoch": 1.0474494339142604, + "grad_norm": 2.623641014099121, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8905502557754517, + "num_tokens": 319635309.0, + "step": 8234 + }, + { + "epoch": 1.0475766441928507, + "grad_norm": 2.6314237117767334, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.888512134552002, + "num_tokens": 319676702.0, + "step": 8235 + }, + { + "epoch": 1.0477038544714412, + "grad_norm": 2.416149616241455, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8922240734100342, + "num_tokens": 319721682.0, + "step": 8236 + }, + { + "epoch": 1.0478310647500317, + "grad_norm": 2.5296130180358887, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8999223113059998, + "num_tokens": 319761668.0, + "step": 8237 + }, + { + "epoch": 1.0479582750286223, + "grad_norm": 2.949164390563965, + "learning_rate": 1e-06, + "loss": 0.3535, + "mean_token_accuracy": 0.8733440637588501, + "num_tokens": 319798926.0, + "step": 8238 + }, + { + "epoch": 1.0480854853072128, + "grad_norm": 2.7184507846832275, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8965799808502197, + "num_tokens": 319835407.0, + "step": 8239 + }, + { + "epoch": 1.0482126955858033, + "grad_norm": 2.4487221240997314, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8936423659324646, + "num_tokens": 319876621.0, + "step": 8240 + }, + { + "epoch": 1.0483399058643939, + "grad_norm": 2.51845383644104, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8967433571815491, + "num_tokens": 319917982.0, + "step": 8241 + }, + { + "epoch": 1.0484671161429844, + "grad_norm": 2.545361042022705, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.90043044090271, + "num_tokens": 319958160.0, + "step": 8242 + }, + { + "epoch": 1.048594326421575, + "grad_norm": 2.469787359237671, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8886839151382446, + "num_tokens": 320003526.0, + "step": 8243 + }, + { + "epoch": 1.0487215367001654, + "grad_norm": 2.7081985473632812, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8772562742233276, + "num_tokens": 320044580.0, + "step": 8244 + }, + { + "epoch": 1.048848746978756, + "grad_norm": 2.487828493118286, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8960151672363281, + "num_tokens": 320085950.0, + "step": 8245 + }, + { + "epoch": 1.0489759572573465, + "grad_norm": 2.676786184310913, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.893223226070404, + "num_tokens": 320124389.0, + "step": 8246 + }, + { + "epoch": 1.0491031675359368, + "grad_norm": 2.43385910987854, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8883016705513, + "num_tokens": 320168862.0, + "step": 8247 + }, + { + "epoch": 1.0492303778145273, + "grad_norm": 2.453880786895752, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8972073793411255, + "num_tokens": 320209836.0, + "step": 8248 + }, + { + "epoch": 1.0493575880931179, + "grad_norm": 2.543243885040283, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.886608362197876, + "num_tokens": 320253280.0, + "step": 8249 + }, + { + "epoch": 1.0494847983717084, + "grad_norm": 2.6197903156280518, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8919097781181335, + "num_tokens": 320290452.0, + "step": 8250 + }, + { + "epoch": 1.049612008650299, + "grad_norm": 2.583646059036255, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8873804211616516, + "num_tokens": 320330918.0, + "step": 8251 + }, + { + "epoch": 1.0497392189288894, + "grad_norm": 2.641005277633667, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8893083333969116, + "num_tokens": 320370882.0, + "step": 8252 + }, + { + "epoch": 1.04986642920748, + "grad_norm": 2.4648027420043945, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8993128538131714, + "num_tokens": 320411553.0, + "step": 8253 + }, + { + "epoch": 1.0499936394860705, + "grad_norm": 2.5958991050720215, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8848244547843933, + "num_tokens": 320450873.0, + "step": 8254 + }, + { + "epoch": 1.050120849764661, + "grad_norm": 2.751262664794922, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8930456638336182, + "num_tokens": 320483787.0, + "step": 8255 + }, + { + "epoch": 1.0502480600432516, + "grad_norm": 2.5942797660827637, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8993842601776123, + "num_tokens": 320524914.0, + "step": 8256 + }, + { + "epoch": 1.050375270321842, + "grad_norm": 2.869373083114624, + "learning_rate": 1e-06, + "loss": 0.3814, + "mean_token_accuracy": 0.8673783540725708, + "num_tokens": 320565003.0, + "step": 8257 + }, + { + "epoch": 1.0505024806004326, + "grad_norm": 2.7133421897888184, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.880614161491394, + "num_tokens": 320603529.0, + "step": 8258 + }, + { + "epoch": 1.0506296908790231, + "grad_norm": 2.5594308376312256, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8905319571495056, + "num_tokens": 320643229.0, + "step": 8259 + }, + { + "epoch": 1.0507569011576134, + "grad_norm": 2.6359140872955322, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8808647394180298, + "num_tokens": 320685617.0, + "step": 8260 + }, + { + "epoch": 1.050884111436204, + "grad_norm": 2.5948193073272705, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8877453804016113, + "num_tokens": 320728438.0, + "step": 8261 + }, + { + "epoch": 1.0510113217147945, + "grad_norm": 2.5608155727386475, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.894110918045044, + "num_tokens": 320766456.0, + "step": 8262 + }, + { + "epoch": 1.051138531993385, + "grad_norm": 2.8666248321533203, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8924142718315125, + "num_tokens": 320798199.0, + "step": 8263 + }, + { + "epoch": 1.0512657422719756, + "grad_norm": 2.5978198051452637, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.897065281867981, + "num_tokens": 320835399.0, + "step": 8264 + }, + { + "epoch": 1.051392952550566, + "grad_norm": 2.6003923416137695, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8868139982223511, + "num_tokens": 320874708.0, + "step": 8265 + }, + { + "epoch": 1.0515201628291566, + "grad_norm": 2.747417688369751, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8883005976676941, + "num_tokens": 320913546.0, + "step": 8266 + }, + { + "epoch": 1.0516473731077471, + "grad_norm": 2.784877300262451, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8891198635101318, + "num_tokens": 320951608.0, + "step": 8267 + }, + { + "epoch": 1.0517745833863377, + "grad_norm": 2.621152877807617, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8763002753257751, + "num_tokens": 320996712.0, + "step": 8268 + }, + { + "epoch": 1.0519017936649282, + "grad_norm": 2.7891488075256348, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8932738900184631, + "num_tokens": 321030507.0, + "step": 8269 + }, + { + "epoch": 1.0520290039435187, + "grad_norm": 2.638014078140259, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8870447874069214, + "num_tokens": 321070748.0, + "step": 8270 + }, + { + "epoch": 1.0521562142221093, + "grad_norm": 2.6613712310791016, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8874942660331726, + "num_tokens": 321106837.0, + "step": 8271 + }, + { + "epoch": 1.0522834245006996, + "grad_norm": 2.4934163093566895, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9066094756126404, + "num_tokens": 321145299.0, + "step": 8272 + }, + { + "epoch": 1.05241063477929, + "grad_norm": 2.699871063232422, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8810873627662659, + "num_tokens": 321182294.0, + "step": 8273 + }, + { + "epoch": 1.0525378450578806, + "grad_norm": 2.546229839324951, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8776702284812927, + "num_tokens": 321223152.0, + "step": 8274 + }, + { + "epoch": 1.0526650553364711, + "grad_norm": 2.5397050380706787, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8954942226409912, + "num_tokens": 321262364.0, + "step": 8275 + }, + { + "epoch": 1.0527922656150617, + "grad_norm": 2.6326236724853516, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9027426242828369, + "num_tokens": 321299004.0, + "step": 8276 + }, + { + "epoch": 1.0529194758936522, + "grad_norm": 2.4729011058807373, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8957818746566772, + "num_tokens": 321341182.0, + "step": 8277 + }, + { + "epoch": 1.0530466861722427, + "grad_norm": 2.4389681816101074, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8886302709579468, + "num_tokens": 321382490.0, + "step": 8278 + }, + { + "epoch": 1.0531738964508333, + "grad_norm": 2.59263014793396, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8954716324806213, + "num_tokens": 321419641.0, + "step": 8279 + }, + { + "epoch": 1.0533011067294238, + "grad_norm": 2.7750115394592285, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8836506605148315, + "num_tokens": 321456121.0, + "step": 8280 + }, + { + "epoch": 1.0534283170080143, + "grad_norm": 2.380770444869995, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9041696190834045, + "num_tokens": 321497414.0, + "step": 8281 + }, + { + "epoch": 1.0535555272866048, + "grad_norm": 2.7424726486206055, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8882583379745483, + "num_tokens": 321535654.0, + "step": 8282 + }, + { + "epoch": 1.0536827375651954, + "grad_norm": 2.636568307876587, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8934468030929565, + "num_tokens": 321575460.0, + "step": 8283 + }, + { + "epoch": 1.0538099478437857, + "grad_norm": 2.9657135009765625, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8868880271911621, + "num_tokens": 321609273.0, + "step": 8284 + }, + { + "epoch": 1.0539371581223762, + "grad_norm": 2.5176548957824707, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.903156042098999, + "num_tokens": 321648743.0, + "step": 8285 + }, + { + "epoch": 1.0540643684009667, + "grad_norm": 2.689152717590332, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.888888955116272, + "num_tokens": 321686387.0, + "step": 8286 + }, + { + "epoch": 1.0541915786795573, + "grad_norm": 2.8292810916900635, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9052280783653259, + "num_tokens": 321722597.0, + "step": 8287 + }, + { + "epoch": 1.0543187889581478, + "grad_norm": 2.762446165084839, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8891966342926025, + "num_tokens": 321757225.0, + "step": 8288 + }, + { + "epoch": 1.0544459992367383, + "grad_norm": 2.7380003929138184, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8934078812599182, + "num_tokens": 321792528.0, + "step": 8289 + }, + { + "epoch": 1.0545732095153288, + "grad_norm": 2.7091734409332275, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8925788998603821, + "num_tokens": 321828741.0, + "step": 8290 + }, + { + "epoch": 1.0547004197939194, + "grad_norm": 2.510693073272705, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8884151577949524, + "num_tokens": 321871802.0, + "step": 8291 + }, + { + "epoch": 1.05482763007251, + "grad_norm": 2.622609853744507, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8858969807624817, + "num_tokens": 321909856.0, + "step": 8292 + }, + { + "epoch": 1.0549548403511004, + "grad_norm": 2.628406047821045, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8904790878295898, + "num_tokens": 321947119.0, + "step": 8293 + }, + { + "epoch": 1.055082050629691, + "grad_norm": 2.9545586109161377, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8761295080184937, + "num_tokens": 321982920.0, + "step": 8294 + }, + { + "epoch": 1.0552092609082815, + "grad_norm": 2.541219472885132, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.9044881463050842, + "num_tokens": 322021343.0, + "step": 8295 + }, + { + "epoch": 1.0553364711868718, + "grad_norm": 2.836763858795166, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8786634206771851, + "num_tokens": 322056194.0, + "step": 8296 + }, + { + "epoch": 1.0554636814654623, + "grad_norm": 2.5181262493133545, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.900667667388916, + "num_tokens": 322094991.0, + "step": 8297 + }, + { + "epoch": 1.0555908917440529, + "grad_norm": 2.2755110263824463, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8939865827560425, + "num_tokens": 322142819.0, + "step": 8298 + }, + { + "epoch": 1.0557181020226434, + "grad_norm": 2.792069673538208, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8850880861282349, + "num_tokens": 322181713.0, + "step": 8299 + }, + { + "epoch": 1.055845312301234, + "grad_norm": 2.8319575786590576, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8775020837783813, + "num_tokens": 322219412.0, + "step": 8300 + }, + { + "epoch": 1.0559725225798244, + "grad_norm": 2.4784138202667236, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8978195786476135, + "num_tokens": 322259032.0, + "step": 8301 + }, + { + "epoch": 1.056099732858415, + "grad_norm": 2.7212867736816406, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8856209516525269, + "num_tokens": 322296888.0, + "step": 8302 + }, + { + "epoch": 1.0562269431370055, + "grad_norm": 2.6088361740112305, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9026647210121155, + "num_tokens": 322333740.0, + "step": 8303 + }, + { + "epoch": 1.056354153415596, + "grad_norm": 2.8454880714416504, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8881564140319824, + "num_tokens": 322364235.0, + "step": 8304 + }, + { + "epoch": 1.0564813636941865, + "grad_norm": 2.7525277137756348, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8822532892227173, + "num_tokens": 322398669.0, + "step": 8305 + }, + { + "epoch": 1.056608573972777, + "grad_norm": 2.489832639694214, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.891717255115509, + "num_tokens": 322441217.0, + "step": 8306 + }, + { + "epoch": 1.0567357842513676, + "grad_norm": 2.445923089981079, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8837745189666748, + "num_tokens": 322487685.0, + "step": 8307 + }, + { + "epoch": 1.0568629945299581, + "grad_norm": 2.510200023651123, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9014732837677002, + "num_tokens": 322527752.0, + "step": 8308 + }, + { + "epoch": 1.0569902048085484, + "grad_norm": 2.650923013687134, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8896505832672119, + "num_tokens": 322565158.0, + "step": 8309 + }, + { + "epoch": 1.057117415087139, + "grad_norm": 2.6905782222747803, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.890518307685852, + "num_tokens": 322602203.0, + "step": 8310 + }, + { + "epoch": 1.0572446253657295, + "grad_norm": 2.818878412246704, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8901312351226807, + "num_tokens": 322639255.0, + "step": 8311 + }, + { + "epoch": 1.05737183564432, + "grad_norm": 2.630969524383545, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8870476484298706, + "num_tokens": 322678743.0, + "step": 8312 + }, + { + "epoch": 1.0574990459229106, + "grad_norm": 2.5801854133605957, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.888454794883728, + "num_tokens": 322718572.0, + "step": 8313 + }, + { + "epoch": 1.057626256201501, + "grad_norm": 2.64176344871521, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8910425901412964, + "num_tokens": 322757125.0, + "step": 8314 + }, + { + "epoch": 1.0577534664800916, + "grad_norm": 2.825291633605957, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8755893111228943, + "num_tokens": 322796061.0, + "step": 8315 + }, + { + "epoch": 1.0578806767586821, + "grad_norm": 2.521341323852539, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8951416611671448, + "num_tokens": 322837285.0, + "step": 8316 + }, + { + "epoch": 1.0580078870372727, + "grad_norm": 2.393695831298828, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9027906656265259, + "num_tokens": 322879435.0, + "step": 8317 + }, + { + "epoch": 1.0581350973158632, + "grad_norm": 2.9294707775115967, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8890329599380493, + "num_tokens": 322915536.0, + "step": 8318 + }, + { + "epoch": 1.0582623075944537, + "grad_norm": 2.6038124561309814, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8849939107894897, + "num_tokens": 322957511.0, + "step": 8319 + }, + { + "epoch": 1.058389517873044, + "grad_norm": 2.653210401535034, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8866258263587952, + "num_tokens": 322997977.0, + "step": 8320 + }, + { + "epoch": 1.0585167281516346, + "grad_norm": 2.4520246982574463, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.892453134059906, + "num_tokens": 323037215.0, + "step": 8321 + }, + { + "epoch": 1.058643938430225, + "grad_norm": 2.623751640319824, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8897743225097656, + "num_tokens": 323076507.0, + "step": 8322 + }, + { + "epoch": 1.0587711487088156, + "grad_norm": 2.5506691932678223, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8873466849327087, + "num_tokens": 323114191.0, + "step": 8323 + }, + { + "epoch": 1.0588983589874061, + "grad_norm": 2.725364923477173, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.898330569267273, + "num_tokens": 323150246.0, + "step": 8324 + }, + { + "epoch": 1.0590255692659967, + "grad_norm": 2.3352580070495605, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9028772115707397, + "num_tokens": 323195357.0, + "step": 8325 + }, + { + "epoch": 1.0591527795445872, + "grad_norm": 2.730678081512451, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8870906233787537, + "num_tokens": 323232604.0, + "step": 8326 + }, + { + "epoch": 1.0592799898231777, + "grad_norm": 2.711078405380249, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8815097212791443, + "num_tokens": 323271487.0, + "step": 8327 + }, + { + "epoch": 1.0594072001017683, + "grad_norm": 2.674868583679199, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8948742151260376, + "num_tokens": 323308939.0, + "step": 8328 + }, + { + "epoch": 1.0595344103803588, + "grad_norm": 2.6550652980804443, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8945457935333252, + "num_tokens": 323344681.0, + "step": 8329 + }, + { + "epoch": 1.0596616206589493, + "grad_norm": 2.66316819190979, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8923859596252441, + "num_tokens": 323381031.0, + "step": 8330 + }, + { + "epoch": 1.0597888309375398, + "grad_norm": 2.5357608795166016, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8985384106636047, + "num_tokens": 323420260.0, + "step": 8331 + }, + { + "epoch": 1.0599160412161304, + "grad_norm": 2.744769334793091, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8879669904708862, + "num_tokens": 323454813.0, + "step": 8332 + }, + { + "epoch": 1.0600432514947207, + "grad_norm": 3.0801942348480225, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8794360160827637, + "num_tokens": 323485118.0, + "step": 8333 + }, + { + "epoch": 1.0601704617733112, + "grad_norm": 2.737072229385376, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8856829404830933, + "num_tokens": 323522351.0, + "step": 8334 + }, + { + "epoch": 1.0602976720519017, + "grad_norm": 2.9281015396118164, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8890806436538696, + "num_tokens": 323554864.0, + "step": 8335 + }, + { + "epoch": 1.0604248823304923, + "grad_norm": 2.6634984016418457, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8950009942054749, + "num_tokens": 323588783.0, + "step": 8336 + }, + { + "epoch": 1.0605520926090828, + "grad_norm": 2.615842819213867, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9001601934432983, + "num_tokens": 323623423.0, + "step": 8337 + }, + { + "epoch": 1.0606793028876733, + "grad_norm": 2.6718404293060303, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8781759142875671, + "num_tokens": 323662392.0, + "step": 8338 + }, + { + "epoch": 1.0608065131662638, + "grad_norm": 2.452043056488037, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8905940055847168, + "num_tokens": 323706780.0, + "step": 8339 + }, + { + "epoch": 1.0609337234448544, + "grad_norm": 2.6687324047088623, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8824355006217957, + "num_tokens": 323746130.0, + "step": 8340 + }, + { + "epoch": 1.061060933723445, + "grad_norm": 2.691631317138672, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8878470659255981, + "num_tokens": 323788561.0, + "step": 8341 + }, + { + "epoch": 1.0611881440020354, + "grad_norm": 2.5900497436523438, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8855457901954651, + "num_tokens": 323830991.0, + "step": 8342 + }, + { + "epoch": 1.061315354280626, + "grad_norm": 2.61592435836792, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8912516832351685, + "num_tokens": 323870547.0, + "step": 8343 + }, + { + "epoch": 1.0614425645592165, + "grad_norm": 2.7047455310821533, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8837233781814575, + "num_tokens": 323912067.0, + "step": 8344 + }, + { + "epoch": 1.0615697748378068, + "grad_norm": 3.1534736156463623, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8833576440811157, + "num_tokens": 323944284.0, + "step": 8345 + }, + { + "epoch": 1.0616969851163973, + "grad_norm": 2.5866799354553223, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9006221294403076, + "num_tokens": 323981730.0, + "step": 8346 + }, + { + "epoch": 1.0618241953949878, + "grad_norm": 2.486224412918091, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8937185406684875, + "num_tokens": 324023968.0, + "step": 8347 + }, + { + "epoch": 1.0619514056735784, + "grad_norm": 2.6582019329071045, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8853758573532104, + "num_tokens": 324064483.0, + "step": 8348 + }, + { + "epoch": 1.062078615952169, + "grad_norm": 2.413133144378662, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8944453001022339, + "num_tokens": 324107727.0, + "step": 8349 + }, + { + "epoch": 1.0622058262307594, + "grad_norm": 2.500251054763794, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8903054594993591, + "num_tokens": 324153106.0, + "step": 8350 + }, + { + "epoch": 1.06233303650935, + "grad_norm": 2.8322036266326904, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8917204737663269, + "num_tokens": 324186302.0, + "step": 8351 + }, + { + "epoch": 1.0624602467879405, + "grad_norm": 2.9544379711151123, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8852863311767578, + "num_tokens": 324227988.0, + "step": 8352 + }, + { + "epoch": 1.062587457066531, + "grad_norm": 2.53690767288208, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8992699384689331, + "num_tokens": 324266342.0, + "step": 8353 + }, + { + "epoch": 1.0627146673451215, + "grad_norm": 2.5519723892211914, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8856410384178162, + "num_tokens": 324309335.0, + "step": 8354 + }, + { + "epoch": 1.062841877623712, + "grad_norm": 2.6109752655029297, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.889755368232727, + "num_tokens": 324350518.0, + "step": 8355 + }, + { + "epoch": 1.0629690879023026, + "grad_norm": 2.6289620399475098, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.9007519483566284, + "num_tokens": 324386155.0, + "step": 8356 + }, + { + "epoch": 1.0630962981808931, + "grad_norm": 2.587071418762207, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8963436484336853, + "num_tokens": 324424378.0, + "step": 8357 + }, + { + "epoch": 1.0632235084594834, + "grad_norm": 2.728353977203369, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.890002429485321, + "num_tokens": 324459876.0, + "step": 8358 + }, + { + "epoch": 1.063350718738074, + "grad_norm": 3.110224485397339, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8702024221420288, + "num_tokens": 324492344.0, + "step": 8359 + }, + { + "epoch": 1.0634779290166645, + "grad_norm": 2.7526605129241943, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8914195895195007, + "num_tokens": 324528716.0, + "step": 8360 + }, + { + "epoch": 1.063605139295255, + "grad_norm": 2.843806505203247, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8776583671569824, + "num_tokens": 324566113.0, + "step": 8361 + }, + { + "epoch": 1.0637323495738455, + "grad_norm": 2.692505121231079, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8873146772384644, + "num_tokens": 324603833.0, + "step": 8362 + }, + { + "epoch": 1.063859559852436, + "grad_norm": 2.337651491165161, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8907418847084045, + "num_tokens": 324650656.0, + "step": 8363 + }, + { + "epoch": 1.0639867701310266, + "grad_norm": 2.7239933013916016, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.889000654220581, + "num_tokens": 324686852.0, + "step": 8364 + }, + { + "epoch": 1.0641139804096171, + "grad_norm": 2.5252845287323, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8967361450195312, + "num_tokens": 324724943.0, + "step": 8365 + }, + { + "epoch": 1.0642411906882077, + "grad_norm": 2.5671916007995605, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8953583836555481, + "num_tokens": 324763340.0, + "step": 8366 + }, + { + "epoch": 1.0643684009667982, + "grad_norm": 2.66186261177063, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8812909126281738, + "num_tokens": 324806855.0, + "step": 8367 + }, + { + "epoch": 1.0644956112453887, + "grad_norm": 2.886569023132324, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8817579746246338, + "num_tokens": 324846862.0, + "step": 8368 + }, + { + "epoch": 1.064622821523979, + "grad_norm": 2.6730525493621826, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8823667764663696, + "num_tokens": 324886028.0, + "step": 8369 + }, + { + "epoch": 1.0647500318025696, + "grad_norm": 3.1372272968292236, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8806036710739136, + "num_tokens": 324919032.0, + "step": 8370 + }, + { + "epoch": 1.06487724208116, + "grad_norm": 2.585413694381714, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8780377507209778, + "num_tokens": 324963422.0, + "step": 8371 + }, + { + "epoch": 1.0650044523597506, + "grad_norm": 2.6533091068267822, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8833863139152527, + "num_tokens": 325003079.0, + "step": 8372 + }, + { + "epoch": 1.0651316626383411, + "grad_norm": 2.2702088356018066, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9057689309120178, + "num_tokens": 325044569.0, + "step": 8373 + }, + { + "epoch": 1.0652588729169317, + "grad_norm": 2.702831506729126, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8889641165733337, + "num_tokens": 325080948.0, + "step": 8374 + }, + { + "epoch": 1.0653860831955222, + "grad_norm": 2.5619008541107178, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8925002813339233, + "num_tokens": 325119113.0, + "step": 8375 + }, + { + "epoch": 1.0655132934741127, + "grad_norm": 2.7383310794830322, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.903266191482544, + "num_tokens": 325149225.0, + "step": 8376 + }, + { + "epoch": 1.0656405037527032, + "grad_norm": 2.6365020275115967, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8938960433006287, + "num_tokens": 325185151.0, + "step": 8377 + }, + { + "epoch": 1.0657677140312938, + "grad_norm": 2.837676763534546, + "learning_rate": 1e-06, + "loss": 0.3464, + "mean_token_accuracy": 0.8790559768676758, + "num_tokens": 325224334.0, + "step": 8378 + }, + { + "epoch": 1.0658949243098843, + "grad_norm": 2.6982920169830322, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8868207335472107, + "num_tokens": 325264650.0, + "step": 8379 + }, + { + "epoch": 1.0660221345884748, + "grad_norm": 2.8057057857513428, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.888073205947876, + "num_tokens": 325301816.0, + "step": 8380 + }, + { + "epoch": 1.0661493448670654, + "grad_norm": 2.6700799465179443, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8952094912528992, + "num_tokens": 325342961.0, + "step": 8381 + }, + { + "epoch": 1.0662765551456557, + "grad_norm": 2.9969022274017334, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8826573491096497, + "num_tokens": 325378385.0, + "step": 8382 + }, + { + "epoch": 1.0664037654242462, + "grad_norm": 2.57035231590271, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8769549131393433, + "num_tokens": 325425423.0, + "step": 8383 + }, + { + "epoch": 1.0665309757028367, + "grad_norm": 2.9417171478271484, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.8747434616088867, + "num_tokens": 325459829.0, + "step": 8384 + }, + { + "epoch": 1.0666581859814273, + "grad_norm": 2.460378646850586, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9041483998298645, + "num_tokens": 325499273.0, + "step": 8385 + }, + { + "epoch": 1.0667853962600178, + "grad_norm": 2.743928909301758, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8756852746009827, + "num_tokens": 325540519.0, + "step": 8386 + }, + { + "epoch": 1.0669126065386083, + "grad_norm": 2.4346020221710205, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9061374664306641, + "num_tokens": 325579372.0, + "step": 8387 + }, + { + "epoch": 1.0670398168171988, + "grad_norm": 2.4740166664123535, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8854420781135559, + "num_tokens": 325622359.0, + "step": 8388 + }, + { + "epoch": 1.0671670270957894, + "grad_norm": 2.7212305068969727, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9016507863998413, + "num_tokens": 325655431.0, + "step": 8389 + }, + { + "epoch": 1.06729423737438, + "grad_norm": 2.7794809341430664, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.890418529510498, + "num_tokens": 325691399.0, + "step": 8390 + }, + { + "epoch": 1.0674214476529704, + "grad_norm": 2.789449453353882, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8871070146560669, + "num_tokens": 325729423.0, + "step": 8391 + }, + { + "epoch": 1.067548657931561, + "grad_norm": 2.6127824783325195, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9052450656890869, + "num_tokens": 325767234.0, + "step": 8392 + }, + { + "epoch": 1.0676758682101515, + "grad_norm": 2.6961116790771484, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8962249755859375, + "num_tokens": 325805609.0, + "step": 8393 + }, + { + "epoch": 1.0678030784887418, + "grad_norm": 2.8906548023223877, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8835328817367554, + "num_tokens": 325843438.0, + "step": 8394 + }, + { + "epoch": 1.0679302887673323, + "grad_norm": 2.551456928253174, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8907883763313293, + "num_tokens": 325888273.0, + "step": 8395 + }, + { + "epoch": 1.0680574990459228, + "grad_norm": 2.6038267612457275, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8984977006912231, + "num_tokens": 325924507.0, + "step": 8396 + }, + { + "epoch": 1.0681847093245134, + "grad_norm": 2.808215856552124, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9012188911437988, + "num_tokens": 325954816.0, + "step": 8397 + }, + { + "epoch": 1.068311919603104, + "grad_norm": 2.4860997200012207, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8865013122558594, + "num_tokens": 326001190.0, + "step": 8398 + }, + { + "epoch": 1.0684391298816944, + "grad_norm": 2.9774937629699707, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8699162602424622, + "num_tokens": 326037924.0, + "step": 8399 + }, + { + "epoch": 1.068566340160285, + "grad_norm": 2.7004759311676025, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8944017291069031, + "num_tokens": 326075534.0, + "step": 8400 + }, + { + "epoch": 1.0686935504388755, + "grad_norm": 2.8366408348083496, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8765720725059509, + "num_tokens": 326114030.0, + "step": 8401 + }, + { + "epoch": 1.068820760717466, + "grad_norm": 2.518679141998291, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8943420648574829, + "num_tokens": 326154670.0, + "step": 8402 + }, + { + "epoch": 1.0689479709960565, + "grad_norm": 2.800943374633789, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8946068286895752, + "num_tokens": 326193143.0, + "step": 8403 + }, + { + "epoch": 1.069075181274647, + "grad_norm": 2.81752610206604, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8827283382415771, + "num_tokens": 326232476.0, + "step": 8404 + }, + { + "epoch": 1.0692023915532376, + "grad_norm": 2.9463701248168945, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8877056837081909, + "num_tokens": 326266141.0, + "step": 8405 + }, + { + "epoch": 1.0693296018318281, + "grad_norm": 2.81390643119812, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8885859251022339, + "num_tokens": 326304998.0, + "step": 8406 + }, + { + "epoch": 1.0694568121104184, + "grad_norm": 2.5183258056640625, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8928496241569519, + "num_tokens": 326351134.0, + "step": 8407 + }, + { + "epoch": 1.069584022389009, + "grad_norm": 2.7701494693756104, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.895377516746521, + "num_tokens": 326385207.0, + "step": 8408 + }, + { + "epoch": 1.0697112326675995, + "grad_norm": 2.8008625507354736, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8770502209663391, + "num_tokens": 326424687.0, + "step": 8409 + }, + { + "epoch": 1.06983844294619, + "grad_norm": 2.927208662033081, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8900785446166992, + "num_tokens": 326453198.0, + "step": 8410 + }, + { + "epoch": 1.0699656532247805, + "grad_norm": 2.3903796672821045, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8944743871688843, + "num_tokens": 326496734.0, + "step": 8411 + }, + { + "epoch": 1.070092863503371, + "grad_norm": 2.9571287631988525, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8908861875534058, + "num_tokens": 326529763.0, + "step": 8412 + }, + { + "epoch": 1.0702200737819616, + "grad_norm": 2.6815154552459717, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8810912370681763, + "num_tokens": 326566707.0, + "step": 8413 + }, + { + "epoch": 1.0703472840605521, + "grad_norm": 2.765064239501953, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9061648845672607, + "num_tokens": 326597137.0, + "step": 8414 + }, + { + "epoch": 1.0704744943391427, + "grad_norm": 2.61014723777771, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.8995259404182434, + "num_tokens": 326634432.0, + "step": 8415 + }, + { + "epoch": 1.0706017046177332, + "grad_norm": 2.363976240158081, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9037231802940369, + "num_tokens": 326674277.0, + "step": 8416 + }, + { + "epoch": 1.0707289148963237, + "grad_norm": 2.8541905879974365, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8936115503311157, + "num_tokens": 326708424.0, + "step": 8417 + }, + { + "epoch": 1.070856125174914, + "grad_norm": 3.0380232334136963, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8856875896453857, + "num_tokens": 326739005.0, + "step": 8418 + }, + { + "epoch": 1.0709833354535045, + "grad_norm": 2.381417989730835, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9007216095924377, + "num_tokens": 326781767.0, + "step": 8419 + }, + { + "epoch": 1.071110545732095, + "grad_norm": 2.969531536102295, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8842425346374512, + "num_tokens": 326814106.0, + "step": 8420 + }, + { + "epoch": 1.0712377560106856, + "grad_norm": 2.58286714553833, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8844488263130188, + "num_tokens": 326856527.0, + "step": 8421 + }, + { + "epoch": 1.0713649662892761, + "grad_norm": 2.9711549282073975, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8869555592536926, + "num_tokens": 326889305.0, + "step": 8422 + }, + { + "epoch": 1.0714921765678667, + "grad_norm": 2.619675397872925, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8892042636871338, + "num_tokens": 326931126.0, + "step": 8423 + }, + { + "epoch": 1.0716193868464572, + "grad_norm": 2.7699830532073975, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8851062059402466, + "num_tokens": 326969852.0, + "step": 8424 + }, + { + "epoch": 1.0717465971250477, + "grad_norm": 2.8477623462677, + "learning_rate": 1e-06, + "loss": 0.3561, + "mean_token_accuracy": 0.872317373752594, + "num_tokens": 327009500.0, + "step": 8425 + }, + { + "epoch": 1.0718738074036382, + "grad_norm": 2.6666839122772217, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8878475427627563, + "num_tokens": 327048947.0, + "step": 8426 + }, + { + "epoch": 1.0720010176822288, + "grad_norm": 2.5359480381011963, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.897649884223938, + "num_tokens": 327091992.0, + "step": 8427 + }, + { + "epoch": 1.0721282279608193, + "grad_norm": 2.672952175140381, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8904237747192383, + "num_tokens": 327132223.0, + "step": 8428 + }, + { + "epoch": 1.0722554382394098, + "grad_norm": 2.548412561416626, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8931934833526611, + "num_tokens": 327175772.0, + "step": 8429 + }, + { + "epoch": 1.0723826485180004, + "grad_norm": 2.628603935241699, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.8997503519058228, + "num_tokens": 327215268.0, + "step": 8430 + }, + { + "epoch": 1.0725098587965907, + "grad_norm": 2.9195690155029297, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8950831890106201, + "num_tokens": 327245755.0, + "step": 8431 + }, + { + "epoch": 1.0726370690751812, + "grad_norm": 2.572974920272827, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9005637168884277, + "num_tokens": 327282166.0, + "step": 8432 + }, + { + "epoch": 1.0727642793537717, + "grad_norm": 2.696934700012207, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8897638320922852, + "num_tokens": 327319604.0, + "step": 8433 + }, + { + "epoch": 1.0728914896323622, + "grad_norm": 2.646242141723633, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8915132284164429, + "num_tokens": 327359561.0, + "step": 8434 + }, + { + "epoch": 1.0730186999109528, + "grad_norm": 2.7972638607025146, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8874236345291138, + "num_tokens": 327396453.0, + "step": 8435 + }, + { + "epoch": 1.0731459101895433, + "grad_norm": 2.7766804695129395, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8873308897018433, + "num_tokens": 327431643.0, + "step": 8436 + }, + { + "epoch": 1.0732731204681338, + "grad_norm": 2.5972418785095215, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8761407732963562, + "num_tokens": 327474139.0, + "step": 8437 + }, + { + "epoch": 1.0734003307467244, + "grad_norm": 2.9520740509033203, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8933110237121582, + "num_tokens": 327507111.0, + "step": 8438 + }, + { + "epoch": 1.073527541025315, + "grad_norm": 2.889183282852173, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8736269474029541, + "num_tokens": 327544893.0, + "step": 8439 + }, + { + "epoch": 1.0736547513039054, + "grad_norm": 2.7873942852020264, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8830187320709229, + "num_tokens": 327586873.0, + "step": 8440 + }, + { + "epoch": 1.073781961582496, + "grad_norm": 2.7273199558258057, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8986674547195435, + "num_tokens": 327622347.0, + "step": 8441 + }, + { + "epoch": 1.0739091718610865, + "grad_norm": 2.8730432987213135, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8992974162101746, + "num_tokens": 327653080.0, + "step": 8442 + }, + { + "epoch": 1.0740363821396768, + "grad_norm": 2.7272071838378906, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8917653560638428, + "num_tokens": 327689431.0, + "step": 8443 + }, + { + "epoch": 1.0741635924182673, + "grad_norm": 3.042377471923828, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8851058483123779, + "num_tokens": 327722698.0, + "step": 8444 + }, + { + "epoch": 1.0742908026968578, + "grad_norm": 2.4163267612457275, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9012889862060547, + "num_tokens": 327763375.0, + "step": 8445 + }, + { + "epoch": 1.0744180129754484, + "grad_norm": 2.593175172805786, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8910857439041138, + "num_tokens": 327804017.0, + "step": 8446 + }, + { + "epoch": 1.074545223254039, + "grad_norm": 2.780343532562256, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.887118399143219, + "num_tokens": 327843322.0, + "step": 8447 + }, + { + "epoch": 1.0746724335326294, + "grad_norm": 2.7789273262023926, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8917418718338013, + "num_tokens": 327879024.0, + "step": 8448 + }, + { + "epoch": 1.07479964381122, + "grad_norm": 2.73893141746521, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8872941732406616, + "num_tokens": 327915434.0, + "step": 8449 + }, + { + "epoch": 1.0749268540898105, + "grad_norm": 2.765778064727783, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8995108008384705, + "num_tokens": 327952183.0, + "step": 8450 + }, + { + "epoch": 1.075054064368401, + "grad_norm": 2.7367732524871826, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8937463760375977, + "num_tokens": 327990098.0, + "step": 8451 + }, + { + "epoch": 1.0751812746469915, + "grad_norm": 2.4693286418914795, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8907584547996521, + "num_tokens": 328035118.0, + "step": 8452 + }, + { + "epoch": 1.075308484925582, + "grad_norm": 2.49687123298645, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8969846963882446, + "num_tokens": 328079479.0, + "step": 8453 + }, + { + "epoch": 1.0754356952041726, + "grad_norm": 2.6120710372924805, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8840659856796265, + "num_tokens": 328121975.0, + "step": 8454 + }, + { + "epoch": 1.0755629054827631, + "grad_norm": 2.563363552093506, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8955338001251221, + "num_tokens": 328161861.0, + "step": 8455 + }, + { + "epoch": 1.0756901157613534, + "grad_norm": 2.9758307933807373, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.881344735622406, + "num_tokens": 328196650.0, + "step": 8456 + }, + { + "epoch": 1.075817326039944, + "grad_norm": 2.6041510105133057, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8910655975341797, + "num_tokens": 328241443.0, + "step": 8457 + }, + { + "epoch": 1.0759445363185345, + "grad_norm": 2.59562349319458, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8951932191848755, + "num_tokens": 328278425.0, + "step": 8458 + }, + { + "epoch": 1.076071746597125, + "grad_norm": 2.647282600402832, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8981854915618896, + "num_tokens": 328319827.0, + "step": 8459 + }, + { + "epoch": 1.0761989568757155, + "grad_norm": 2.705258846282959, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8915055990219116, + "num_tokens": 328355652.0, + "step": 8460 + }, + { + "epoch": 1.076326167154306, + "grad_norm": 2.784677028656006, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8734759092330933, + "num_tokens": 328394873.0, + "step": 8461 + }, + { + "epoch": 1.0764533774328966, + "grad_norm": 2.2803852558135986, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8957977294921875, + "num_tokens": 328444496.0, + "step": 8462 + }, + { + "epoch": 1.0765805877114871, + "grad_norm": 2.6850690841674805, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8921380043029785, + "num_tokens": 328480948.0, + "step": 8463 + }, + { + "epoch": 1.0767077979900777, + "grad_norm": 2.7153029441833496, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8948094844818115, + "num_tokens": 328516138.0, + "step": 8464 + }, + { + "epoch": 1.0768350082686682, + "grad_norm": 2.590052366256714, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8943266868591309, + "num_tokens": 328556171.0, + "step": 8465 + }, + { + "epoch": 1.0769622185472587, + "grad_norm": 2.5446386337280273, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8953960537910461, + "num_tokens": 328595585.0, + "step": 8466 + }, + { + "epoch": 1.077089428825849, + "grad_norm": 2.863455057144165, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8768541812896729, + "num_tokens": 328629458.0, + "step": 8467 + }, + { + "epoch": 1.0772166391044395, + "grad_norm": 2.811422348022461, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8791142106056213, + "num_tokens": 328670122.0, + "step": 8468 + }, + { + "epoch": 1.07734384938303, + "grad_norm": 2.6276350021362305, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8973824977874756, + "num_tokens": 328706624.0, + "step": 8469 + }, + { + "epoch": 1.0774710596616206, + "grad_norm": 2.513500213623047, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8972253799438477, + "num_tokens": 328747029.0, + "step": 8470 + }, + { + "epoch": 1.0775982699402111, + "grad_norm": 2.686713218688965, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.893638014793396, + "num_tokens": 328783864.0, + "step": 8471 + }, + { + "epoch": 1.0777254802188017, + "grad_norm": 2.6973583698272705, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8831778764724731, + "num_tokens": 328821642.0, + "step": 8472 + }, + { + "epoch": 1.0778526904973922, + "grad_norm": 2.7072699069976807, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.890790581703186, + "num_tokens": 328862066.0, + "step": 8473 + }, + { + "epoch": 1.0779799007759827, + "grad_norm": 2.9451136589050293, + "learning_rate": 1e-06, + "loss": 0.348, + "mean_token_accuracy": 0.8761920928955078, + "num_tokens": 328896279.0, + "step": 8474 + }, + { + "epoch": 1.0781071110545732, + "grad_norm": 2.6388161182403564, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8991184830665588, + "num_tokens": 328932421.0, + "step": 8475 + }, + { + "epoch": 1.0782343213331638, + "grad_norm": 2.6645853519439697, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8977674841880798, + "num_tokens": 328970004.0, + "step": 8476 + }, + { + "epoch": 1.0783615316117543, + "grad_norm": 2.6753649711608887, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8969711065292358, + "num_tokens": 329010095.0, + "step": 8477 + }, + { + "epoch": 1.0784887418903448, + "grad_norm": 2.6579248905181885, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8843656778335571, + "num_tokens": 329049610.0, + "step": 8478 + }, + { + "epoch": 1.0786159521689354, + "grad_norm": 2.4317023754119873, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9038382768630981, + "num_tokens": 329089710.0, + "step": 8479 + }, + { + "epoch": 1.0787431624475257, + "grad_norm": 2.869135618209839, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8768024444580078, + "num_tokens": 329126669.0, + "step": 8480 + }, + { + "epoch": 1.0788703727261162, + "grad_norm": 2.844674587249756, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8884366154670715, + "num_tokens": 329161470.0, + "step": 8481 + }, + { + "epoch": 1.0789975830047067, + "grad_norm": 2.7121660709381104, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8818143606185913, + "num_tokens": 329202143.0, + "step": 8482 + }, + { + "epoch": 1.0791247932832972, + "grad_norm": 2.5044891834259033, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.897807240486145, + "num_tokens": 329242700.0, + "step": 8483 + }, + { + "epoch": 1.0792520035618878, + "grad_norm": 2.6450467109680176, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8954903483390808, + "num_tokens": 329281541.0, + "step": 8484 + }, + { + "epoch": 1.0793792138404783, + "grad_norm": 2.6603546142578125, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8967453241348267, + "num_tokens": 329315265.0, + "step": 8485 + }, + { + "epoch": 1.0795064241190688, + "grad_norm": 2.617166757583618, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8895936012268066, + "num_tokens": 329356193.0, + "step": 8486 + }, + { + "epoch": 1.0796336343976594, + "grad_norm": 2.517158269882202, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8898881673812866, + "num_tokens": 329398178.0, + "step": 8487 + }, + { + "epoch": 1.0797608446762499, + "grad_norm": 2.668480157852173, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8829967379570007, + "num_tokens": 329441960.0, + "step": 8488 + }, + { + "epoch": 1.0798880549548404, + "grad_norm": 2.8234941959381104, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9077044129371643, + "num_tokens": 329476849.0, + "step": 8489 + }, + { + "epoch": 1.080015265233431, + "grad_norm": 2.8642337322235107, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8826045989990234, + "num_tokens": 329513951.0, + "step": 8490 + }, + { + "epoch": 1.0801424755120215, + "grad_norm": 2.4061930179595947, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8888344764709473, + "num_tokens": 329563158.0, + "step": 8491 + }, + { + "epoch": 1.0802696857906118, + "grad_norm": 2.574641466140747, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8924358487129211, + "num_tokens": 329605885.0, + "step": 8492 + }, + { + "epoch": 1.0803968960692023, + "grad_norm": 2.5696375370025635, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.907734215259552, + "num_tokens": 329641197.0, + "step": 8493 + }, + { + "epoch": 1.0805241063477928, + "grad_norm": 2.4796035289764404, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8908901810646057, + "num_tokens": 329683050.0, + "step": 8494 + }, + { + "epoch": 1.0806513166263834, + "grad_norm": 2.3734354972839355, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9030928015708923, + "num_tokens": 329724644.0, + "step": 8495 + }, + { + "epoch": 1.080778526904974, + "grad_norm": 3.0117335319519043, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8983091115951538, + "num_tokens": 329755772.0, + "step": 8496 + }, + { + "epoch": 1.0809057371835644, + "grad_norm": 2.9201531410217285, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8901183605194092, + "num_tokens": 329788757.0, + "step": 8497 + }, + { + "epoch": 1.081032947462155, + "grad_norm": 2.699906587600708, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.899739682674408, + "num_tokens": 329825379.0, + "step": 8498 + }, + { + "epoch": 1.0811601577407455, + "grad_norm": 2.6389853954315186, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8896408677101135, + "num_tokens": 329864486.0, + "step": 8499 + }, + { + "epoch": 1.081287368019336, + "grad_norm": 2.603931188583374, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8913136720657349, + "num_tokens": 329904810.0, + "step": 8500 + }, + { + "epoch": 1.0814145782979265, + "grad_norm": 2.4222400188446045, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8922333717346191, + "num_tokens": 329948455.0, + "step": 8501 + }, + { + "epoch": 1.081541788576517, + "grad_norm": 2.678513526916504, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8855600357055664, + "num_tokens": 329989292.0, + "step": 8502 + }, + { + "epoch": 1.0816689988551076, + "grad_norm": 2.429713010787964, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8975635170936584, + "num_tokens": 330030623.0, + "step": 8503 + }, + { + "epoch": 1.0817962091336981, + "grad_norm": 2.7784087657928467, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8915342092514038, + "num_tokens": 330070904.0, + "step": 8504 + }, + { + "epoch": 1.0819234194122884, + "grad_norm": 2.610379219055176, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8944758176803589, + "num_tokens": 330105949.0, + "step": 8505 + }, + { + "epoch": 1.082050629690879, + "grad_norm": 2.556001663208008, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8978210687637329, + "num_tokens": 330141090.0, + "step": 8506 + }, + { + "epoch": 1.0821778399694695, + "grad_norm": 2.758704423904419, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8905996084213257, + "num_tokens": 330176701.0, + "step": 8507 + }, + { + "epoch": 1.08230505024806, + "grad_norm": 2.6376447677612305, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8802618980407715, + "num_tokens": 330218254.0, + "step": 8508 + }, + { + "epoch": 1.0824322605266505, + "grad_norm": 2.835585832595825, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8935723304748535, + "num_tokens": 330255508.0, + "step": 8509 + }, + { + "epoch": 1.082559470805241, + "grad_norm": 2.3540213108062744, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8994107246398926, + "num_tokens": 330300184.0, + "step": 8510 + }, + { + "epoch": 1.0826866810838316, + "grad_norm": 2.623814344406128, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8804135918617249, + "num_tokens": 330345524.0, + "step": 8511 + }, + { + "epoch": 1.0828138913624221, + "grad_norm": 2.6557509899139404, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.887078046798706, + "num_tokens": 330386013.0, + "step": 8512 + }, + { + "epoch": 1.0829411016410126, + "grad_norm": 2.181757926940918, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8950967788696289, + "num_tokens": 330439716.0, + "step": 8513 + }, + { + "epoch": 1.0830683119196032, + "grad_norm": 2.8377230167388916, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8937602043151855, + "num_tokens": 330474345.0, + "step": 8514 + }, + { + "epoch": 1.0831955221981937, + "grad_norm": 2.732243299484253, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8926150798797607, + "num_tokens": 330506882.0, + "step": 8515 + }, + { + "epoch": 1.083322732476784, + "grad_norm": 2.6356587409973145, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8834106922149658, + "num_tokens": 330549905.0, + "step": 8516 + }, + { + "epoch": 1.0834499427553745, + "grad_norm": 2.6147055625915527, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8895093202590942, + "num_tokens": 330587285.0, + "step": 8517 + }, + { + "epoch": 1.083577153033965, + "grad_norm": 2.5759706497192383, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8851490020751953, + "num_tokens": 330631857.0, + "step": 8518 + }, + { + "epoch": 1.0837043633125556, + "grad_norm": 2.7985520362854004, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8963825702667236, + "num_tokens": 330670626.0, + "step": 8519 + }, + { + "epoch": 1.0838315735911461, + "grad_norm": 2.706418991088867, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8844364881515503, + "num_tokens": 330711078.0, + "step": 8520 + }, + { + "epoch": 1.0839587838697367, + "grad_norm": 2.804340124130249, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8913440704345703, + "num_tokens": 330745621.0, + "step": 8521 + }, + { + "epoch": 1.0840859941483272, + "grad_norm": 2.956916570663452, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8769677877426147, + "num_tokens": 330778814.0, + "step": 8522 + }, + { + "epoch": 1.0842132044269177, + "grad_norm": 2.4826838970184326, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8929895758628845, + "num_tokens": 330820713.0, + "step": 8523 + }, + { + "epoch": 1.0843404147055082, + "grad_norm": 2.8125133514404297, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8879423141479492, + "num_tokens": 330858330.0, + "step": 8524 + }, + { + "epoch": 1.0844676249840988, + "grad_norm": 2.7057762145996094, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8854925632476807, + "num_tokens": 330897365.0, + "step": 8525 + }, + { + "epoch": 1.0845948352626893, + "grad_norm": 2.615828514099121, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8833770751953125, + "num_tokens": 330937210.0, + "step": 8526 + }, + { + "epoch": 1.0847220455412798, + "grad_norm": 2.8039751052856445, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8876327276229858, + "num_tokens": 330977544.0, + "step": 8527 + }, + { + "epoch": 1.0848492558198704, + "grad_norm": 2.54960298538208, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8831130266189575, + "num_tokens": 331020498.0, + "step": 8528 + }, + { + "epoch": 1.0849764660984607, + "grad_norm": 2.8179330825805664, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8781931400299072, + "num_tokens": 331057349.0, + "step": 8529 + }, + { + "epoch": 1.0851036763770512, + "grad_norm": 2.6478891372680664, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9025236964225769, + "num_tokens": 331093522.0, + "step": 8530 + }, + { + "epoch": 1.0852308866556417, + "grad_norm": 2.6521222591400146, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8848373889923096, + "num_tokens": 331134782.0, + "step": 8531 + }, + { + "epoch": 1.0853580969342322, + "grad_norm": 2.677285671234131, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9024590253829956, + "num_tokens": 331169402.0, + "step": 8532 + }, + { + "epoch": 1.0854853072128228, + "grad_norm": 2.702819347381592, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8793487548828125, + "num_tokens": 331209480.0, + "step": 8533 + }, + { + "epoch": 1.0856125174914133, + "grad_norm": 2.6966452598571777, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8833847045898438, + "num_tokens": 331250831.0, + "step": 8534 + }, + { + "epoch": 1.0857397277700038, + "grad_norm": 2.716053009033203, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8959470987319946, + "num_tokens": 331288428.0, + "step": 8535 + }, + { + "epoch": 1.0858669380485944, + "grad_norm": 2.7961840629577637, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8940609097480774, + "num_tokens": 331321637.0, + "step": 8536 + }, + { + "epoch": 1.0859941483271849, + "grad_norm": 2.718078136444092, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8858294486999512, + "num_tokens": 331360834.0, + "step": 8537 + }, + { + "epoch": 1.0861213586057754, + "grad_norm": 2.6688895225524902, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8969913721084595, + "num_tokens": 331397238.0, + "step": 8538 + }, + { + "epoch": 1.086248568884366, + "grad_norm": 2.765474319458008, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9031153917312622, + "num_tokens": 331427766.0, + "step": 8539 + }, + { + "epoch": 1.0863757791629565, + "grad_norm": 2.8992955684661865, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8757796287536621, + "num_tokens": 331463470.0, + "step": 8540 + }, + { + "epoch": 1.0865029894415468, + "grad_norm": 2.617832660675049, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8886934518814087, + "num_tokens": 331505704.0, + "step": 8541 + }, + { + "epoch": 1.0866301997201373, + "grad_norm": 2.705122232437134, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8829097747802734, + "num_tokens": 331545284.0, + "step": 8542 + }, + { + "epoch": 1.0867574099987278, + "grad_norm": 2.482830286026001, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9043192863464355, + "num_tokens": 331583736.0, + "step": 8543 + }, + { + "epoch": 1.0868846202773184, + "grad_norm": 2.471876621246338, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8962973356246948, + "num_tokens": 331623780.0, + "step": 8544 + }, + { + "epoch": 1.0870118305559089, + "grad_norm": 2.7526416778564453, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8845192193984985, + "num_tokens": 331662582.0, + "step": 8545 + }, + { + "epoch": 1.0871390408344994, + "grad_norm": 2.6506712436676025, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8903417587280273, + "num_tokens": 331701900.0, + "step": 8546 + }, + { + "epoch": 1.08726625111309, + "grad_norm": 2.583937883377075, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8947970867156982, + "num_tokens": 331743315.0, + "step": 8547 + }, + { + "epoch": 1.0873934613916805, + "grad_norm": 2.535979747772217, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8903591632843018, + "num_tokens": 331783317.0, + "step": 8548 + }, + { + "epoch": 1.087520671670271, + "grad_norm": 2.59649395942688, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8833807110786438, + "num_tokens": 331825518.0, + "step": 8549 + }, + { + "epoch": 1.0876478819488615, + "grad_norm": 2.6285815238952637, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8817270398139954, + "num_tokens": 331865861.0, + "step": 8550 + }, + { + "epoch": 1.087775092227452, + "grad_norm": 2.725527763366699, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8867661356925964, + "num_tokens": 331904412.0, + "step": 8551 + }, + { + "epoch": 1.0879023025060426, + "grad_norm": 2.4906299114227295, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8926401138305664, + "num_tokens": 331948123.0, + "step": 8552 + }, + { + "epoch": 1.0880295127846331, + "grad_norm": 2.931595802307129, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8870579600334167, + "num_tokens": 331981883.0, + "step": 8553 + }, + { + "epoch": 1.0881567230632234, + "grad_norm": 3.059415817260742, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8863286972045898, + "num_tokens": 332013597.0, + "step": 8554 + }, + { + "epoch": 1.088283933341814, + "grad_norm": 2.8583927154541016, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8721752166748047, + "num_tokens": 332058224.0, + "step": 8555 + }, + { + "epoch": 1.0884111436204045, + "grad_norm": 2.648641347885132, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9010471105575562, + "num_tokens": 332093962.0, + "step": 8556 + }, + { + "epoch": 1.088538353898995, + "grad_norm": 2.7481513023376465, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8849771618843079, + "num_tokens": 332131239.0, + "step": 8557 + }, + { + "epoch": 1.0886655641775855, + "grad_norm": 2.7041239738464355, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8769062161445618, + "num_tokens": 332174035.0, + "step": 8558 + }, + { + "epoch": 1.088792774456176, + "grad_norm": 2.444674491882324, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8986935615539551, + "num_tokens": 332218704.0, + "step": 8559 + }, + { + "epoch": 1.0889199847347666, + "grad_norm": 2.651273012161255, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8820978403091431, + "num_tokens": 332263763.0, + "step": 8560 + }, + { + "epoch": 1.0890471950133571, + "grad_norm": 2.890035629272461, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8748956918716431, + "num_tokens": 332302122.0, + "step": 8561 + }, + { + "epoch": 1.0891744052919476, + "grad_norm": 2.86289381980896, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8816202282905579, + "num_tokens": 332337649.0, + "step": 8562 + }, + { + "epoch": 1.0893016155705382, + "grad_norm": 2.733290910720825, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8876588344573975, + "num_tokens": 332376154.0, + "step": 8563 + }, + { + "epoch": 1.0894288258491287, + "grad_norm": 2.3964688777923584, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8954143524169922, + "num_tokens": 332423196.0, + "step": 8564 + }, + { + "epoch": 1.089556036127719, + "grad_norm": 2.6486639976501465, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8891164064407349, + "num_tokens": 332460189.0, + "step": 8565 + }, + { + "epoch": 1.0896832464063095, + "grad_norm": 2.5648865699768066, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8881051540374756, + "num_tokens": 332503149.0, + "step": 8566 + }, + { + "epoch": 1.0898104566849, + "grad_norm": 2.60768461227417, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8973098993301392, + "num_tokens": 332539754.0, + "step": 8567 + }, + { + "epoch": 1.0899376669634906, + "grad_norm": 2.617753505706787, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8932774066925049, + "num_tokens": 332578343.0, + "step": 8568 + }, + { + "epoch": 1.0900648772420811, + "grad_norm": 2.7609493732452393, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.891400933265686, + "num_tokens": 332613626.0, + "step": 8569 + }, + { + "epoch": 1.0901920875206716, + "grad_norm": 2.470139980316162, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9050317406654358, + "num_tokens": 332655513.0, + "step": 8570 + }, + { + "epoch": 1.0903192977992622, + "grad_norm": 2.707758665084839, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8860353231430054, + "num_tokens": 332693739.0, + "step": 8571 + }, + { + "epoch": 1.0904465080778527, + "grad_norm": 2.4878032207489014, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8972395658493042, + "num_tokens": 332734264.0, + "step": 8572 + }, + { + "epoch": 1.0905737183564432, + "grad_norm": 3.0360207557678223, + "learning_rate": 1e-06, + "loss": 0.4027, + "mean_token_accuracy": 0.8635806441307068, + "num_tokens": 332774616.0, + "step": 8573 + }, + { + "epoch": 1.0907009286350338, + "grad_norm": 2.944298267364502, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8833348751068115, + "num_tokens": 332807849.0, + "step": 8574 + }, + { + "epoch": 1.0908281389136243, + "grad_norm": 2.7955312728881836, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8940016627311707, + "num_tokens": 332846789.0, + "step": 8575 + }, + { + "epoch": 1.0909553491922148, + "grad_norm": 2.510930299758911, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8916337490081787, + "num_tokens": 332890779.0, + "step": 8576 + }, + { + "epoch": 1.0910825594708053, + "grad_norm": 2.4679484367370605, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.8982150554656982, + "num_tokens": 332930700.0, + "step": 8577 + }, + { + "epoch": 1.0912097697493957, + "grad_norm": 2.607097625732422, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8960720300674438, + "num_tokens": 332969279.0, + "step": 8578 + }, + { + "epoch": 1.0913369800279862, + "grad_norm": 2.6092309951782227, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8759266138076782, + "num_tokens": 333009465.0, + "step": 8579 + }, + { + "epoch": 1.0914641903065767, + "grad_norm": 2.4192721843719482, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9085050225257874, + "num_tokens": 333046823.0, + "step": 8580 + }, + { + "epoch": 1.0915914005851672, + "grad_norm": 2.624570846557617, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8897067904472351, + "num_tokens": 333086426.0, + "step": 8581 + }, + { + "epoch": 1.0917186108637578, + "grad_norm": 2.3769397735595703, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9083918333053589, + "num_tokens": 333124077.0, + "step": 8582 + }, + { + "epoch": 1.0918458211423483, + "grad_norm": 2.458207130432129, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9032930731773376, + "num_tokens": 333162048.0, + "step": 8583 + }, + { + "epoch": 1.0919730314209388, + "grad_norm": 2.6316914558410645, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8765758872032166, + "num_tokens": 333204280.0, + "step": 8584 + }, + { + "epoch": 1.0921002416995294, + "grad_norm": 2.5869340896606445, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8967021107673645, + "num_tokens": 333241812.0, + "step": 8585 + }, + { + "epoch": 1.0922274519781199, + "grad_norm": 2.563542366027832, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8915508985519409, + "num_tokens": 333281253.0, + "step": 8586 + }, + { + "epoch": 1.0923546622567104, + "grad_norm": 2.7577290534973145, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.884109616279602, + "num_tokens": 333319026.0, + "step": 8587 + }, + { + "epoch": 1.092481872535301, + "grad_norm": 2.771153450012207, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.880582332611084, + "num_tokens": 333360249.0, + "step": 8588 + }, + { + "epoch": 1.0926090828138915, + "grad_norm": 2.814476490020752, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8783818483352661, + "num_tokens": 333399370.0, + "step": 8589 + }, + { + "epoch": 1.0927362930924818, + "grad_norm": 2.796400308609009, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8840662240982056, + "num_tokens": 333438580.0, + "step": 8590 + }, + { + "epoch": 1.0928635033710723, + "grad_norm": 2.4228830337524414, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9059555530548096, + "num_tokens": 333478283.0, + "step": 8591 + }, + { + "epoch": 1.0929907136496628, + "grad_norm": 2.9156806468963623, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8816254734992981, + "num_tokens": 333518157.0, + "step": 8592 + }, + { + "epoch": 1.0931179239282534, + "grad_norm": 2.7356789112091064, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8775445222854614, + "num_tokens": 333560367.0, + "step": 8593 + }, + { + "epoch": 1.0932451342068439, + "grad_norm": 2.8620030879974365, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8915826082229614, + "num_tokens": 333599504.0, + "step": 8594 + }, + { + "epoch": 1.0933723444854344, + "grad_norm": 2.7657387256622314, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8914942741394043, + "num_tokens": 333636074.0, + "step": 8595 + }, + { + "epoch": 1.093499554764025, + "grad_norm": 2.7332024574279785, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8829209804534912, + "num_tokens": 333677672.0, + "step": 8596 + }, + { + "epoch": 1.0936267650426155, + "grad_norm": 2.7017576694488525, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8967699408531189, + "num_tokens": 333714939.0, + "step": 8597 + }, + { + "epoch": 1.093753975321206, + "grad_norm": 2.7797048091888428, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8925020694732666, + "num_tokens": 333752190.0, + "step": 8598 + }, + { + "epoch": 1.0938811855997965, + "grad_norm": 2.8490679264068604, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8877538442611694, + "num_tokens": 333789126.0, + "step": 8599 + }, + { + "epoch": 1.094008395878387, + "grad_norm": 2.612713575363159, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8963423371315002, + "num_tokens": 333828009.0, + "step": 8600 + }, + { + "epoch": 1.0941356061569776, + "grad_norm": 2.4894907474517822, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8847044706344604, + "num_tokens": 333871563.0, + "step": 8601 + }, + { + "epoch": 1.094262816435568, + "grad_norm": 2.7269673347473145, + "learning_rate": 1e-06, + "loss": 0.3625, + "mean_token_accuracy": 0.8720167875289917, + "num_tokens": 333911232.0, + "step": 8602 + }, + { + "epoch": 1.0943900267141584, + "grad_norm": 2.5217320919036865, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8906446695327759, + "num_tokens": 333953748.0, + "step": 8603 + }, + { + "epoch": 1.094517236992749, + "grad_norm": 2.670109510421753, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8934558629989624, + "num_tokens": 333988881.0, + "step": 8604 + }, + { + "epoch": 1.0946444472713395, + "grad_norm": 2.8357648849487305, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.882171094417572, + "num_tokens": 334027865.0, + "step": 8605 + }, + { + "epoch": 1.09477165754993, + "grad_norm": 2.4928786754608154, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8958596587181091, + "num_tokens": 334068126.0, + "step": 8606 + }, + { + "epoch": 1.0948988678285205, + "grad_norm": 2.764331579208374, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8831503987312317, + "num_tokens": 334106231.0, + "step": 8607 + }, + { + "epoch": 1.095026078107111, + "grad_norm": 2.6970996856689453, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.890259325504303, + "num_tokens": 334143912.0, + "step": 8608 + }, + { + "epoch": 1.0951532883857016, + "grad_norm": 2.6342551708221436, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8918964266777039, + "num_tokens": 334182703.0, + "step": 8609 + }, + { + "epoch": 1.0952804986642921, + "grad_norm": 3.187159299850464, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8744573593139648, + "num_tokens": 334211953.0, + "step": 8610 + }, + { + "epoch": 1.0954077089428826, + "grad_norm": 2.710721492767334, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8824157118797302, + "num_tokens": 334253215.0, + "step": 8611 + }, + { + "epoch": 1.0955349192214732, + "grad_norm": 2.7682318687438965, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.887661337852478, + "num_tokens": 334288131.0, + "step": 8612 + }, + { + "epoch": 1.0956621295000637, + "grad_norm": 2.670074939727783, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8849336504936218, + "num_tokens": 334329413.0, + "step": 8613 + }, + { + "epoch": 1.095789339778654, + "grad_norm": 2.845791816711426, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8853654265403748, + "num_tokens": 334362639.0, + "step": 8614 + }, + { + "epoch": 1.0959165500572445, + "grad_norm": 2.470808267593384, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8989289402961731, + "num_tokens": 334403224.0, + "step": 8615 + }, + { + "epoch": 1.096043760335835, + "grad_norm": 2.892604112625122, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8930409550666809, + "num_tokens": 334434934.0, + "step": 8616 + }, + { + "epoch": 1.0961709706144256, + "grad_norm": 2.5516061782836914, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9021857976913452, + "num_tokens": 334473276.0, + "step": 8617 + }, + { + "epoch": 1.0962981808930161, + "grad_norm": 2.968125104904175, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8861559629440308, + "num_tokens": 334507327.0, + "step": 8618 + }, + { + "epoch": 1.0964253911716066, + "grad_norm": 2.5800952911376953, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.893927812576294, + "num_tokens": 334553812.0, + "step": 8619 + }, + { + "epoch": 1.0965526014501972, + "grad_norm": 2.6355819702148438, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8917475342750549, + "num_tokens": 334595324.0, + "step": 8620 + }, + { + "epoch": 1.0966798117287877, + "grad_norm": 3.0695247650146484, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.884976863861084, + "num_tokens": 334629464.0, + "step": 8621 + }, + { + "epoch": 1.0968070220073782, + "grad_norm": 2.542781114578247, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8961752653121948, + "num_tokens": 334667860.0, + "step": 8622 + }, + { + "epoch": 1.0969342322859688, + "grad_norm": 2.778388023376465, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8991438150405884, + "num_tokens": 334703966.0, + "step": 8623 + }, + { + "epoch": 1.0970614425645593, + "grad_norm": 2.6627144813537598, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8951084613800049, + "num_tokens": 334742856.0, + "step": 8624 + }, + { + "epoch": 1.0971886528431498, + "grad_norm": 2.4943904876708984, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8975573182106018, + "num_tokens": 334784300.0, + "step": 8625 + }, + { + "epoch": 1.0973158631217403, + "grad_norm": 2.529690980911255, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9006912708282471, + "num_tokens": 334824571.0, + "step": 8626 + }, + { + "epoch": 1.0974430734003306, + "grad_norm": 2.566441297531128, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.901206910610199, + "num_tokens": 334863574.0, + "step": 8627 + }, + { + "epoch": 1.0975702836789212, + "grad_norm": 2.561699151992798, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8968523144721985, + "num_tokens": 334902596.0, + "step": 8628 + }, + { + "epoch": 1.0976974939575117, + "grad_norm": 2.617614269256592, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.888222336769104, + "num_tokens": 334945835.0, + "step": 8629 + }, + { + "epoch": 1.0978247042361022, + "grad_norm": 2.4718687534332275, + "learning_rate": 1e-06, + "loss": 0.2369, + "mean_token_accuracy": 0.9130789041519165, + "num_tokens": 334983087.0, + "step": 8630 + }, + { + "epoch": 1.0979519145146928, + "grad_norm": 2.92802357673645, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8758931159973145, + "num_tokens": 335020514.0, + "step": 8631 + }, + { + "epoch": 1.0980791247932833, + "grad_norm": 2.5842161178588867, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8983980417251587, + "num_tokens": 335060647.0, + "step": 8632 + }, + { + "epoch": 1.0982063350718738, + "grad_norm": 2.529181718826294, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8849139213562012, + "num_tokens": 335106771.0, + "step": 8633 + }, + { + "epoch": 1.0983335453504643, + "grad_norm": 2.6689021587371826, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8925251960754395, + "num_tokens": 335143151.0, + "step": 8634 + }, + { + "epoch": 1.0984607556290549, + "grad_norm": 2.847731351852417, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8794988989830017, + "num_tokens": 335182301.0, + "step": 8635 + }, + { + "epoch": 1.0985879659076454, + "grad_norm": 2.5305464267730713, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8892455101013184, + "num_tokens": 335225420.0, + "step": 8636 + }, + { + "epoch": 1.098715176186236, + "grad_norm": 2.8073220252990723, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8711644411087036, + "num_tokens": 335268019.0, + "step": 8637 + }, + { + "epoch": 1.0988423864648265, + "grad_norm": 2.645318031311035, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8934046626091003, + "num_tokens": 335309261.0, + "step": 8638 + }, + { + "epoch": 1.0989695967434168, + "grad_norm": 2.771949291229248, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8792495131492615, + "num_tokens": 335348647.0, + "step": 8639 + }, + { + "epoch": 1.0990968070220073, + "grad_norm": 2.430722713470459, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8922233581542969, + "num_tokens": 335392892.0, + "step": 8640 + }, + { + "epoch": 1.0992240173005978, + "grad_norm": 2.7517595291137695, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8872656226158142, + "num_tokens": 335429627.0, + "step": 8641 + }, + { + "epoch": 1.0993512275791884, + "grad_norm": 2.8266072273254395, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8725904226303101, + "num_tokens": 335470185.0, + "step": 8642 + }, + { + "epoch": 1.0994784378577789, + "grad_norm": 2.4043238162994385, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8932400941848755, + "num_tokens": 335512817.0, + "step": 8643 + }, + { + "epoch": 1.0996056481363694, + "grad_norm": 2.5988776683807373, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8898550271987915, + "num_tokens": 335551915.0, + "step": 8644 + }, + { + "epoch": 1.09973285841496, + "grad_norm": 2.6035735607147217, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.885710597038269, + "num_tokens": 335594257.0, + "step": 8645 + }, + { + "epoch": 1.0998600686935505, + "grad_norm": 2.5308310985565186, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8908778429031372, + "num_tokens": 335637193.0, + "step": 8646 + }, + { + "epoch": 1.099987278972141, + "grad_norm": 2.431519031524658, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.898617684841156, + "num_tokens": 335676257.0, + "step": 8647 + }, + { + "epoch": 1.1001144892507315, + "grad_norm": 2.5443806648254395, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8808715343475342, + "num_tokens": 335720230.0, + "step": 8648 + }, + { + "epoch": 1.100241699529322, + "grad_norm": 2.338841438293457, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9037394523620605, + "num_tokens": 335759822.0, + "step": 8649 + }, + { + "epoch": 1.1003689098079126, + "grad_norm": 2.872176170349121, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8898414373397827, + "num_tokens": 335793368.0, + "step": 8650 + }, + { + "epoch": 1.100496120086503, + "grad_norm": 2.7187540531158447, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8856825828552246, + "num_tokens": 335833412.0, + "step": 8651 + }, + { + "epoch": 1.1006233303650934, + "grad_norm": 2.7011377811431885, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8929125070571899, + "num_tokens": 335869682.0, + "step": 8652 + }, + { + "epoch": 1.100750540643684, + "grad_norm": 2.9732954502105713, + "learning_rate": 1e-06, + "loss": 0.3647, + "mean_token_accuracy": 0.8684415221214294, + "num_tokens": 335908231.0, + "step": 8653 + }, + { + "epoch": 1.1008777509222745, + "grad_norm": 2.412907600402832, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8931448459625244, + "num_tokens": 335953424.0, + "step": 8654 + }, + { + "epoch": 1.101004961200865, + "grad_norm": 2.600968360900879, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.888983964920044, + "num_tokens": 335991242.0, + "step": 8655 + }, + { + "epoch": 1.1011321714794555, + "grad_norm": 2.7931671142578125, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8868043422698975, + "num_tokens": 336028217.0, + "step": 8656 + }, + { + "epoch": 1.101259381758046, + "grad_norm": 2.788094997406006, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.889593243598938, + "num_tokens": 336061816.0, + "step": 8657 + }, + { + "epoch": 1.1013865920366366, + "grad_norm": 2.6775784492492676, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9018231630325317, + "num_tokens": 336096724.0, + "step": 8658 + }, + { + "epoch": 1.101513802315227, + "grad_norm": 2.852597951889038, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8894896507263184, + "num_tokens": 336132018.0, + "step": 8659 + }, + { + "epoch": 1.1016410125938176, + "grad_norm": 2.821758270263672, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8900240063667297, + "num_tokens": 336168103.0, + "step": 8660 + }, + { + "epoch": 1.1017682228724082, + "grad_norm": 2.659139394760132, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8949888348579407, + "num_tokens": 336206709.0, + "step": 8661 + }, + { + "epoch": 1.1018954331509987, + "grad_norm": 2.356146812438965, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8929835557937622, + "num_tokens": 336253147.0, + "step": 8662 + }, + { + "epoch": 1.102022643429589, + "grad_norm": 2.4684948921203613, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.8949927091598511, + "num_tokens": 336296637.0, + "step": 8663 + }, + { + "epoch": 1.1021498537081795, + "grad_norm": 2.7085249423980713, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8864516019821167, + "num_tokens": 336337401.0, + "step": 8664 + }, + { + "epoch": 1.10227706398677, + "grad_norm": 3.1278512477874756, + "learning_rate": 1e-06, + "loss": 0.39, + "mean_token_accuracy": 0.8610503673553467, + "num_tokens": 336373511.0, + "step": 8665 + }, + { + "epoch": 1.1024042742653606, + "grad_norm": 2.8069162368774414, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8975128531455994, + "num_tokens": 336410288.0, + "step": 8666 + }, + { + "epoch": 1.1025314845439511, + "grad_norm": 2.8038809299468994, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8829476833343506, + "num_tokens": 336451322.0, + "step": 8667 + }, + { + "epoch": 1.1026586948225416, + "grad_norm": 2.7552318572998047, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.883514940738678, + "num_tokens": 336488965.0, + "step": 8668 + }, + { + "epoch": 1.1027859051011322, + "grad_norm": 2.547229290008545, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8955392837524414, + "num_tokens": 336532407.0, + "step": 8669 + }, + { + "epoch": 1.1029131153797227, + "grad_norm": 2.738205671310425, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8838652968406677, + "num_tokens": 336568742.0, + "step": 8670 + }, + { + "epoch": 1.1030403256583132, + "grad_norm": 2.5343313217163086, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8908263444900513, + "num_tokens": 336614201.0, + "step": 8671 + }, + { + "epoch": 1.1031675359369038, + "grad_norm": 2.533768892288208, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8981034755706787, + "num_tokens": 336654517.0, + "step": 8672 + }, + { + "epoch": 1.1032947462154943, + "grad_norm": 2.7763242721557617, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8843392133712769, + "num_tokens": 336691043.0, + "step": 8673 + }, + { + "epoch": 1.1034219564940848, + "grad_norm": 2.7309398651123047, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9005075693130493, + "num_tokens": 336723233.0, + "step": 8674 + }, + { + "epoch": 1.1035491667726753, + "grad_norm": 2.426605701446533, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9127439856529236, + "num_tokens": 336762942.0, + "step": 8675 + }, + { + "epoch": 1.1036763770512656, + "grad_norm": 2.675485610961914, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8876163959503174, + "num_tokens": 336802864.0, + "step": 8676 + }, + { + "epoch": 1.1038035873298562, + "grad_norm": 2.636885643005371, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9028612375259399, + "num_tokens": 336836604.0, + "step": 8677 + }, + { + "epoch": 1.1039307976084467, + "grad_norm": 2.426593780517578, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8850911855697632, + "num_tokens": 336882374.0, + "step": 8678 + }, + { + "epoch": 1.1040580078870372, + "grad_norm": 2.4791452884674072, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8909837007522583, + "num_tokens": 336926039.0, + "step": 8679 + }, + { + "epoch": 1.1041852181656278, + "grad_norm": 2.868333101272583, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9006450176239014, + "num_tokens": 336956566.0, + "step": 8680 + }, + { + "epoch": 1.1043124284442183, + "grad_norm": 2.6928491592407227, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.889169454574585, + "num_tokens": 336996977.0, + "step": 8681 + }, + { + "epoch": 1.1044396387228088, + "grad_norm": 2.570840358734131, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.89737868309021, + "num_tokens": 337035638.0, + "step": 8682 + }, + { + "epoch": 1.1045668490013993, + "grad_norm": 2.7272374629974365, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8946205973625183, + "num_tokens": 337074646.0, + "step": 8683 + }, + { + "epoch": 1.1046940592799899, + "grad_norm": 2.5662481784820557, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8976091146469116, + "num_tokens": 337115951.0, + "step": 8684 + }, + { + "epoch": 1.1048212695585804, + "grad_norm": 2.5997183322906494, + "learning_rate": 1e-06, + "loss": 0.3624, + "mean_token_accuracy": 0.8742947578430176, + "num_tokens": 337162414.0, + "step": 8685 + }, + { + "epoch": 1.104948479837171, + "grad_norm": 2.7385849952697754, + "learning_rate": 1e-06, + "loss": 0.3722, + "mean_token_accuracy": 0.8688915371894836, + "num_tokens": 337205629.0, + "step": 8686 + }, + { + "epoch": 1.1050756901157615, + "grad_norm": 2.5640158653259277, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8964928984642029, + "num_tokens": 337246518.0, + "step": 8687 + }, + { + "epoch": 1.1052029003943518, + "grad_norm": 2.8884778022766113, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8995819091796875, + "num_tokens": 337281327.0, + "step": 8688 + }, + { + "epoch": 1.1053301106729423, + "grad_norm": 2.550934314727783, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8903012275695801, + "num_tokens": 337327778.0, + "step": 8689 + }, + { + "epoch": 1.1054573209515328, + "grad_norm": 2.8550257682800293, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8957952260971069, + "num_tokens": 337362940.0, + "step": 8690 + }, + { + "epoch": 1.1055845312301233, + "grad_norm": 2.7741875648498535, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8942458629608154, + "num_tokens": 337398746.0, + "step": 8691 + }, + { + "epoch": 1.1057117415087139, + "grad_norm": 2.4193122386932373, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8961348533630371, + "num_tokens": 337442658.0, + "step": 8692 + }, + { + "epoch": 1.1058389517873044, + "grad_norm": 3.0338339805603027, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8896420001983643, + "num_tokens": 337477575.0, + "step": 8693 + }, + { + "epoch": 1.105966162065895, + "grad_norm": 2.6637277603149414, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.881784200668335, + "num_tokens": 337519654.0, + "step": 8694 + }, + { + "epoch": 1.1060933723444855, + "grad_norm": 2.646104335784912, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8893989324569702, + "num_tokens": 337556769.0, + "step": 8695 + }, + { + "epoch": 1.106220582623076, + "grad_norm": 2.8531508445739746, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.885970413684845, + "num_tokens": 337593152.0, + "step": 8696 + }, + { + "epoch": 1.1063477929016665, + "grad_norm": 2.6747028827667236, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9018440246582031, + "num_tokens": 337629534.0, + "step": 8697 + }, + { + "epoch": 1.106475003180257, + "grad_norm": 2.7818164825439453, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8922257423400879, + "num_tokens": 337664337.0, + "step": 8698 + }, + { + "epoch": 1.1066022134588476, + "grad_norm": 2.6336417198181152, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.894039511680603, + "num_tokens": 337699602.0, + "step": 8699 + }, + { + "epoch": 1.106729423737438, + "grad_norm": 3.1022696495056152, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8903611302375793, + "num_tokens": 337730167.0, + "step": 8700 + }, + { + "epoch": 1.1068566340160284, + "grad_norm": 2.636744260787964, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8896873593330383, + "num_tokens": 337771500.0, + "step": 8701 + }, + { + "epoch": 1.106983844294619, + "grad_norm": 2.880410671234131, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8659672737121582, + "num_tokens": 337810656.0, + "step": 8702 + }, + { + "epoch": 1.1071110545732095, + "grad_norm": 2.796502113342285, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8859987258911133, + "num_tokens": 337846067.0, + "step": 8703 + }, + { + "epoch": 1.1072382648518, + "grad_norm": 2.7430813312530518, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8934964537620544, + "num_tokens": 337884992.0, + "step": 8704 + }, + { + "epoch": 1.1073654751303905, + "grad_norm": 2.657067060470581, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8917850255966187, + "num_tokens": 337924400.0, + "step": 8705 + }, + { + "epoch": 1.107492685408981, + "grad_norm": 2.8795204162597656, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8869364261627197, + "num_tokens": 337959253.0, + "step": 8706 + }, + { + "epoch": 1.1076198956875716, + "grad_norm": 2.668752431869507, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8888431787490845, + "num_tokens": 337997094.0, + "step": 8707 + }, + { + "epoch": 1.107747105966162, + "grad_norm": 2.623695135116577, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8851718902587891, + "num_tokens": 338039248.0, + "step": 8708 + }, + { + "epoch": 1.1078743162447526, + "grad_norm": 2.653270721435547, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8901122808456421, + "num_tokens": 338078521.0, + "step": 8709 + }, + { + "epoch": 1.1080015265233432, + "grad_norm": 2.6949141025543213, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8989766836166382, + "num_tokens": 338110821.0, + "step": 8710 + }, + { + "epoch": 1.1081287368019337, + "grad_norm": 2.390737771987915, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8954913020133972, + "num_tokens": 338153773.0, + "step": 8711 + }, + { + "epoch": 1.108255947080524, + "grad_norm": 2.563340902328491, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.896053671836853, + "num_tokens": 338194802.0, + "step": 8712 + }, + { + "epoch": 1.1083831573591145, + "grad_norm": 2.849217414855957, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8854532837867737, + "num_tokens": 338231605.0, + "step": 8713 + }, + { + "epoch": 1.108510367637705, + "grad_norm": 2.8381264209747314, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8864149451255798, + "num_tokens": 338269567.0, + "step": 8714 + }, + { + "epoch": 1.1086375779162956, + "grad_norm": 2.4357221126556396, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8953651189804077, + "num_tokens": 338310982.0, + "step": 8715 + }, + { + "epoch": 1.108764788194886, + "grad_norm": 2.608867645263672, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8839449286460876, + "num_tokens": 338352538.0, + "step": 8716 + }, + { + "epoch": 1.1088919984734766, + "grad_norm": 2.6525473594665527, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8926550149917603, + "num_tokens": 338390793.0, + "step": 8717 + }, + { + "epoch": 1.1090192087520672, + "grad_norm": 2.91367244720459, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8697971105575562, + "num_tokens": 338429423.0, + "step": 8718 + }, + { + "epoch": 1.1091464190306577, + "grad_norm": 2.7043979167938232, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8881071209907532, + "num_tokens": 338466499.0, + "step": 8719 + }, + { + "epoch": 1.1092736293092482, + "grad_norm": 2.7242488861083984, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8978753089904785, + "num_tokens": 338502838.0, + "step": 8720 + }, + { + "epoch": 1.1094008395878387, + "grad_norm": 2.868809938430786, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.875615119934082, + "num_tokens": 338540001.0, + "step": 8721 + }, + { + "epoch": 1.1095280498664293, + "grad_norm": 2.797227382659912, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.896217942237854, + "num_tokens": 338576986.0, + "step": 8722 + }, + { + "epoch": 1.1096552601450198, + "grad_norm": 2.533630132675171, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8941675424575806, + "num_tokens": 338618511.0, + "step": 8723 + }, + { + "epoch": 1.1097824704236103, + "grad_norm": 3.290752649307251, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8834453821182251, + "num_tokens": 338650654.0, + "step": 8724 + }, + { + "epoch": 1.1099096807022006, + "grad_norm": 2.9828450679779053, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8858662247657776, + "num_tokens": 338683748.0, + "step": 8725 + }, + { + "epoch": 1.1100368909807912, + "grad_norm": 2.5511951446533203, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8884027600288391, + "num_tokens": 338726446.0, + "step": 8726 + }, + { + "epoch": 1.1101641012593817, + "grad_norm": 3.005145311355591, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8749269247055054, + "num_tokens": 338767884.0, + "step": 8727 + }, + { + "epoch": 1.1102913115379722, + "grad_norm": 2.553502082824707, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8983848094940186, + "num_tokens": 338807953.0, + "step": 8728 + }, + { + "epoch": 1.1104185218165628, + "grad_norm": 2.8663673400878906, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8826508522033691, + "num_tokens": 338848075.0, + "step": 8729 + }, + { + "epoch": 1.1105457320951533, + "grad_norm": 2.9658496379852295, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8868622779846191, + "num_tokens": 338881331.0, + "step": 8730 + }, + { + "epoch": 1.1106729423737438, + "grad_norm": 2.5002341270446777, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.894376277923584, + "num_tokens": 338925120.0, + "step": 8731 + }, + { + "epoch": 1.1108001526523343, + "grad_norm": 2.533806085586548, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8935074806213379, + "num_tokens": 338966281.0, + "step": 8732 + }, + { + "epoch": 1.1109273629309249, + "grad_norm": 2.7323482036590576, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8937649726867676, + "num_tokens": 339001796.0, + "step": 8733 + }, + { + "epoch": 1.1110545732095154, + "grad_norm": 2.629392147064209, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8976523876190186, + "num_tokens": 339040983.0, + "step": 8734 + }, + { + "epoch": 1.111181783488106, + "grad_norm": 2.404642105102539, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9050232172012329, + "num_tokens": 339081838.0, + "step": 8735 + }, + { + "epoch": 1.1113089937666965, + "grad_norm": 2.5524914264678955, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8848296999931335, + "num_tokens": 339123724.0, + "step": 8736 + }, + { + "epoch": 1.1114362040452868, + "grad_norm": 2.69171404838562, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8882923722267151, + "num_tokens": 339161976.0, + "step": 8737 + }, + { + "epoch": 1.1115634143238773, + "grad_norm": 2.6378443241119385, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8949341773986816, + "num_tokens": 339199132.0, + "step": 8738 + }, + { + "epoch": 1.1116906246024678, + "grad_norm": 2.7786006927490234, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.89634108543396, + "num_tokens": 339232240.0, + "step": 8739 + }, + { + "epoch": 1.1118178348810583, + "grad_norm": 2.7063605785369873, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8933919668197632, + "num_tokens": 339269807.0, + "step": 8740 + }, + { + "epoch": 1.1119450451596489, + "grad_norm": 2.765347480773926, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.889897346496582, + "num_tokens": 339307503.0, + "step": 8741 + }, + { + "epoch": 1.1120722554382394, + "grad_norm": 2.844402551651001, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8762403726577759, + "num_tokens": 339345592.0, + "step": 8742 + }, + { + "epoch": 1.11219946571683, + "grad_norm": 2.625830888748169, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8951712846755981, + "num_tokens": 339384704.0, + "step": 8743 + }, + { + "epoch": 1.1123266759954205, + "grad_norm": 2.7081923484802246, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8840500116348267, + "num_tokens": 339424341.0, + "step": 8744 + }, + { + "epoch": 1.112453886274011, + "grad_norm": 2.6181092262268066, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8906885981559753, + "num_tokens": 339462455.0, + "step": 8745 + }, + { + "epoch": 1.1125810965526015, + "grad_norm": 2.882155179977417, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8848176002502441, + "num_tokens": 339500152.0, + "step": 8746 + }, + { + "epoch": 1.112708306831192, + "grad_norm": 2.495002031326294, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8860921859741211, + "num_tokens": 339547722.0, + "step": 8747 + }, + { + "epoch": 1.1128355171097826, + "grad_norm": 2.746025323867798, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8910422325134277, + "num_tokens": 339582230.0, + "step": 8748 + }, + { + "epoch": 1.112962727388373, + "grad_norm": 2.6206161975860596, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9002469182014465, + "num_tokens": 339617277.0, + "step": 8749 + }, + { + "epoch": 1.1130899376669634, + "grad_norm": 2.610654354095459, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.896827220916748, + "num_tokens": 339655956.0, + "step": 8750 + }, + { + "epoch": 1.113217147945554, + "grad_norm": 2.932904005050659, + "learning_rate": 1e-06, + "loss": 0.3485, + "mean_token_accuracy": 0.8767067193984985, + "num_tokens": 339692895.0, + "step": 8751 + }, + { + "epoch": 1.1133443582241445, + "grad_norm": 2.7776942253112793, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8922227621078491, + "num_tokens": 339728986.0, + "step": 8752 + }, + { + "epoch": 1.113471568502735, + "grad_norm": 2.6325478553771973, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8915000557899475, + "num_tokens": 339769601.0, + "step": 8753 + }, + { + "epoch": 1.1135987787813255, + "grad_norm": 2.694230079650879, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8992292284965515, + "num_tokens": 339803444.0, + "step": 8754 + }, + { + "epoch": 1.113725989059916, + "grad_norm": 2.933763027191162, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8908647894859314, + "num_tokens": 339835889.0, + "step": 8755 + }, + { + "epoch": 1.1138531993385066, + "grad_norm": 2.736145257949829, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.89322829246521, + "num_tokens": 339873911.0, + "step": 8756 + }, + { + "epoch": 1.113980409617097, + "grad_norm": 2.5285937786102295, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8886725306510925, + "num_tokens": 339919322.0, + "step": 8757 + }, + { + "epoch": 1.1141076198956876, + "grad_norm": 2.668625831604004, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8879441618919373, + "num_tokens": 339959738.0, + "step": 8758 + }, + { + "epoch": 1.1142348301742782, + "grad_norm": 2.656745433807373, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8922473192214966, + "num_tokens": 339997091.0, + "step": 8759 + }, + { + "epoch": 1.1143620404528687, + "grad_norm": 2.703514575958252, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8881412148475647, + "num_tokens": 340035754.0, + "step": 8760 + }, + { + "epoch": 1.114489250731459, + "grad_norm": 2.606832981109619, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8975909948348999, + "num_tokens": 340073484.0, + "step": 8761 + }, + { + "epoch": 1.1146164610100495, + "grad_norm": 2.716660737991333, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8887961506843567, + "num_tokens": 340112489.0, + "step": 8762 + }, + { + "epoch": 1.11474367128864, + "grad_norm": 2.4599320888519287, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8924040794372559, + "num_tokens": 340154450.0, + "step": 8763 + }, + { + "epoch": 1.1148708815672306, + "grad_norm": 2.7266738414764404, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8944234848022461, + "num_tokens": 340190308.0, + "step": 8764 + }, + { + "epoch": 1.114998091845821, + "grad_norm": 2.5622901916503906, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8897607922554016, + "num_tokens": 340231443.0, + "step": 8765 + }, + { + "epoch": 1.1151253021244116, + "grad_norm": 2.6309733390808105, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8940988779067993, + "num_tokens": 340268847.0, + "step": 8766 + }, + { + "epoch": 1.1152525124030022, + "grad_norm": 2.507946014404297, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8813169002532959, + "num_tokens": 340314125.0, + "step": 8767 + }, + { + "epoch": 1.1153797226815927, + "grad_norm": 2.5932304859161377, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8982734680175781, + "num_tokens": 340353237.0, + "step": 8768 + }, + { + "epoch": 1.1155069329601832, + "grad_norm": 2.6691231727600098, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9034759998321533, + "num_tokens": 340385388.0, + "step": 8769 + }, + { + "epoch": 1.1156341432387737, + "grad_norm": 2.5140860080718994, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8919229507446289, + "num_tokens": 340427150.0, + "step": 8770 + }, + { + "epoch": 1.1157613535173643, + "grad_norm": 2.53806734085083, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8877445459365845, + "num_tokens": 340469355.0, + "step": 8771 + }, + { + "epoch": 1.1158885637959548, + "grad_norm": 2.577540397644043, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.887211799621582, + "num_tokens": 340511916.0, + "step": 8772 + }, + { + "epoch": 1.1160157740745453, + "grad_norm": 2.561182737350464, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8901354074478149, + "num_tokens": 340554208.0, + "step": 8773 + }, + { + "epoch": 1.1161429843531356, + "grad_norm": 2.424055814743042, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9118958711624146, + "num_tokens": 340594669.0, + "step": 8774 + }, + { + "epoch": 1.1162701946317262, + "grad_norm": 2.635050058364868, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8885663747787476, + "num_tokens": 340634480.0, + "step": 8775 + }, + { + "epoch": 1.1163974049103167, + "grad_norm": 2.6853950023651123, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8956416845321655, + "num_tokens": 340672963.0, + "step": 8776 + }, + { + "epoch": 1.1165246151889072, + "grad_norm": 3.043268918991089, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.893283486366272, + "num_tokens": 340706049.0, + "step": 8777 + }, + { + "epoch": 1.1166518254674977, + "grad_norm": 2.967827558517456, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8750147223472595, + "num_tokens": 340746819.0, + "step": 8778 + }, + { + "epoch": 1.1167790357460883, + "grad_norm": 2.830353021621704, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8759077191352844, + "num_tokens": 340789362.0, + "step": 8779 + }, + { + "epoch": 1.1169062460246788, + "grad_norm": 2.6792852878570557, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.891015887260437, + "num_tokens": 340827765.0, + "step": 8780 + }, + { + "epoch": 1.1170334563032693, + "grad_norm": 2.673997163772583, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9061741828918457, + "num_tokens": 340862028.0, + "step": 8781 + }, + { + "epoch": 1.1171606665818599, + "grad_norm": 2.9414381980895996, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8812036514282227, + "num_tokens": 340900377.0, + "step": 8782 + }, + { + "epoch": 1.1172878768604504, + "grad_norm": 2.9941678047180176, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8757137060165405, + "num_tokens": 340938300.0, + "step": 8783 + }, + { + "epoch": 1.117415087139041, + "grad_norm": 2.789992094039917, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8848823308944702, + "num_tokens": 340977279.0, + "step": 8784 + }, + { + "epoch": 1.1175422974176314, + "grad_norm": 2.8029861450195312, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8847436904907227, + "num_tokens": 341014724.0, + "step": 8785 + }, + { + "epoch": 1.1176695076962218, + "grad_norm": 2.46693754196167, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8870857954025269, + "num_tokens": 341056509.0, + "step": 8786 + }, + { + "epoch": 1.1177967179748123, + "grad_norm": 2.8943235874176025, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8803791403770447, + "num_tokens": 341096213.0, + "step": 8787 + }, + { + "epoch": 1.1179239282534028, + "grad_norm": 2.5846660137176514, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8997950553894043, + "num_tokens": 341137168.0, + "step": 8788 + }, + { + "epoch": 1.1180511385319933, + "grad_norm": 2.8262038230895996, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8929498791694641, + "num_tokens": 341169084.0, + "step": 8789 + }, + { + "epoch": 1.1181783488105839, + "grad_norm": 2.826552391052246, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8776805400848389, + "num_tokens": 341204150.0, + "step": 8790 + }, + { + "epoch": 1.1183055590891744, + "grad_norm": 2.5244014263153076, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8962795734405518, + "num_tokens": 341244640.0, + "step": 8791 + }, + { + "epoch": 1.118432769367765, + "grad_norm": 2.6305623054504395, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.890421450138092, + "num_tokens": 341283922.0, + "step": 8792 + }, + { + "epoch": 1.1185599796463555, + "grad_norm": 2.623004674911499, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.898145318031311, + "num_tokens": 341319419.0, + "step": 8793 + }, + { + "epoch": 1.118687189924946, + "grad_norm": 2.3344383239746094, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8950198292732239, + "num_tokens": 341366151.0, + "step": 8794 + }, + { + "epoch": 1.1188144002035365, + "grad_norm": 2.5905656814575195, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8972911834716797, + "num_tokens": 341407009.0, + "step": 8795 + }, + { + "epoch": 1.118941610482127, + "grad_norm": 2.6741206645965576, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8973972797393799, + "num_tokens": 341445675.0, + "step": 8796 + }, + { + "epoch": 1.1190688207607176, + "grad_norm": 2.535844326019287, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8986077308654785, + "num_tokens": 341483408.0, + "step": 8797 + }, + { + "epoch": 1.119196031039308, + "grad_norm": 2.6838619709014893, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8835558295249939, + "num_tokens": 341522287.0, + "step": 8798 + }, + { + "epoch": 1.1193232413178984, + "grad_norm": 2.749356746673584, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8860417008399963, + "num_tokens": 341561518.0, + "step": 8799 + }, + { + "epoch": 1.119450451596489, + "grad_norm": 2.7347002029418945, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8770136833190918, + "num_tokens": 341605743.0, + "step": 8800 + }, + { + "epoch": 1.1195776618750795, + "grad_norm": 2.524563789367676, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8942133188247681, + "num_tokens": 341648829.0, + "step": 8801 + }, + { + "epoch": 1.11970487215367, + "grad_norm": 2.724738359451294, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8951175212860107, + "num_tokens": 341686019.0, + "step": 8802 + }, + { + "epoch": 1.1198320824322605, + "grad_norm": 2.775151014328003, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8864712715148926, + "num_tokens": 341724190.0, + "step": 8803 + }, + { + "epoch": 1.119959292710851, + "grad_norm": 2.7629570960998535, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8884603381156921, + "num_tokens": 341762215.0, + "step": 8804 + }, + { + "epoch": 1.1200865029894416, + "grad_norm": 2.8429691791534424, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.891311764717102, + "num_tokens": 341796405.0, + "step": 8805 + }, + { + "epoch": 1.120213713268032, + "grad_norm": 2.8231310844421387, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8886029720306396, + "num_tokens": 341832502.0, + "step": 8806 + }, + { + "epoch": 1.1203409235466226, + "grad_norm": 2.742401361465454, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.890558123588562, + "num_tokens": 341869297.0, + "step": 8807 + }, + { + "epoch": 1.1204681338252132, + "grad_norm": 2.7349724769592285, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8809816241264343, + "num_tokens": 341909704.0, + "step": 8808 + }, + { + "epoch": 1.1205953441038037, + "grad_norm": 2.600278615951538, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8916294574737549, + "num_tokens": 341952097.0, + "step": 8809 + }, + { + "epoch": 1.120722554382394, + "grad_norm": 2.8518176078796387, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8864122629165649, + "num_tokens": 341986358.0, + "step": 8810 + }, + { + "epoch": 1.1208497646609845, + "grad_norm": 2.639014959335327, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8915582299232483, + "num_tokens": 342029082.0, + "step": 8811 + }, + { + "epoch": 1.120976974939575, + "grad_norm": 2.7203385829925537, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8962225317955017, + "num_tokens": 342071879.0, + "step": 8812 + }, + { + "epoch": 1.1211041852181656, + "grad_norm": 2.8349268436431885, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8889010548591614, + "num_tokens": 342107524.0, + "step": 8813 + }, + { + "epoch": 1.121231395496756, + "grad_norm": 2.7786378860473633, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.888076901435852, + "num_tokens": 342144087.0, + "step": 8814 + }, + { + "epoch": 1.1213586057753466, + "grad_norm": 2.4710500240325928, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9044443368911743, + "num_tokens": 342184984.0, + "step": 8815 + }, + { + "epoch": 1.1214858160539372, + "grad_norm": 2.569601535797119, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8928970098495483, + "num_tokens": 342224817.0, + "step": 8816 + }, + { + "epoch": 1.1216130263325277, + "grad_norm": 2.4424431324005127, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8873997926712036, + "num_tokens": 342270955.0, + "step": 8817 + }, + { + "epoch": 1.1217402366111182, + "grad_norm": 2.6160731315612793, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.888075053691864, + "num_tokens": 342310370.0, + "step": 8818 + }, + { + "epoch": 1.1218674468897087, + "grad_norm": 2.4628093242645264, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9055065512657166, + "num_tokens": 342351047.0, + "step": 8819 + }, + { + "epoch": 1.1219946571682993, + "grad_norm": 2.6265952587127686, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8988058567047119, + "num_tokens": 342388719.0, + "step": 8820 + }, + { + "epoch": 1.1221218674468898, + "grad_norm": 2.44143009185791, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8990222811698914, + "num_tokens": 342430804.0, + "step": 8821 + }, + { + "epoch": 1.1222490777254803, + "grad_norm": 2.8690197467803955, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8915047645568848, + "num_tokens": 342466251.0, + "step": 8822 + }, + { + "epoch": 1.1223762880040706, + "grad_norm": 2.7265732288360596, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8884923458099365, + "num_tokens": 342502167.0, + "step": 8823 + }, + { + "epoch": 1.1225034982826612, + "grad_norm": 2.766545295715332, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8846308588981628, + "num_tokens": 342541039.0, + "step": 8824 + }, + { + "epoch": 1.1226307085612517, + "grad_norm": 2.716252326965332, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8831595182418823, + "num_tokens": 342584503.0, + "step": 8825 + }, + { + "epoch": 1.1227579188398422, + "grad_norm": 2.634953498840332, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8897528052330017, + "num_tokens": 342624479.0, + "step": 8826 + }, + { + "epoch": 1.1228851291184327, + "grad_norm": 2.816494941711426, + "learning_rate": 1e-06, + "loss": 0.3651, + "mean_token_accuracy": 0.869844377040863, + "num_tokens": 342666070.0, + "step": 8827 + }, + { + "epoch": 1.1230123393970233, + "grad_norm": 2.698760747909546, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8926883935928345, + "num_tokens": 342704728.0, + "step": 8828 + }, + { + "epoch": 1.1231395496756138, + "grad_norm": 2.762049674987793, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8907886147499084, + "num_tokens": 342740161.0, + "step": 8829 + }, + { + "epoch": 1.1232667599542043, + "grad_norm": 2.487889051437378, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8917551040649414, + "num_tokens": 342784394.0, + "step": 8830 + }, + { + "epoch": 1.1233939702327949, + "grad_norm": 2.6632587909698486, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8937872648239136, + "num_tokens": 342823870.0, + "step": 8831 + }, + { + "epoch": 1.1235211805113854, + "grad_norm": 2.7007944583892822, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8945658206939697, + "num_tokens": 342858107.0, + "step": 8832 + }, + { + "epoch": 1.123648390789976, + "grad_norm": 2.429439067840576, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8970088958740234, + "num_tokens": 342900799.0, + "step": 8833 + }, + { + "epoch": 1.1237756010685664, + "grad_norm": 2.789565324783325, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9046446084976196, + "num_tokens": 342932454.0, + "step": 8834 + }, + { + "epoch": 1.1239028113471567, + "grad_norm": 2.678144693374634, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8807734847068787, + "num_tokens": 342972408.0, + "step": 8835 + }, + { + "epoch": 1.1240300216257473, + "grad_norm": 2.542415142059326, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8983898162841797, + "num_tokens": 343012510.0, + "step": 8836 + }, + { + "epoch": 1.1241572319043378, + "grad_norm": 2.6433281898498535, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8948363065719604, + "num_tokens": 343050609.0, + "step": 8837 + }, + { + "epoch": 1.1242844421829283, + "grad_norm": 2.4526920318603516, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8966325521469116, + "num_tokens": 343093200.0, + "step": 8838 + }, + { + "epoch": 1.1244116524615189, + "grad_norm": 2.407740592956543, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8963658809661865, + "num_tokens": 343136898.0, + "step": 8839 + }, + { + "epoch": 1.1245388627401094, + "grad_norm": 2.5796825885772705, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8823032379150391, + "num_tokens": 343182250.0, + "step": 8840 + }, + { + "epoch": 1.1246660730187, + "grad_norm": 2.8682045936584473, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8943405151367188, + "num_tokens": 343217781.0, + "step": 8841 + }, + { + "epoch": 1.1247932832972904, + "grad_norm": 2.383077383041382, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9007492661476135, + "num_tokens": 343258569.0, + "step": 8842 + }, + { + "epoch": 1.124920493575881, + "grad_norm": 2.425795316696167, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8940931558609009, + "num_tokens": 343304029.0, + "step": 8843 + }, + { + "epoch": 1.1250477038544715, + "grad_norm": 2.629253387451172, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8837523460388184, + "num_tokens": 343346379.0, + "step": 8844 + }, + { + "epoch": 1.125174914133062, + "grad_norm": 3.004153251647949, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8706701993942261, + "num_tokens": 343384400.0, + "step": 8845 + }, + { + "epoch": 1.1253021244116526, + "grad_norm": 2.5518736839294434, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8941673040390015, + "num_tokens": 343422144.0, + "step": 8846 + }, + { + "epoch": 1.125429334690243, + "grad_norm": 2.459437370300293, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9052983522415161, + "num_tokens": 343461260.0, + "step": 8847 + }, + { + "epoch": 1.1255565449688334, + "grad_norm": 2.6605935096740723, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8959428071975708, + "num_tokens": 343499819.0, + "step": 8848 + }, + { + "epoch": 1.125683755247424, + "grad_norm": 2.5515341758728027, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8964601755142212, + "num_tokens": 343542210.0, + "step": 8849 + }, + { + "epoch": 1.1258109655260145, + "grad_norm": 2.5780813694000244, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8935074806213379, + "num_tokens": 343581538.0, + "step": 8850 + }, + { + "epoch": 1.125938175804605, + "grad_norm": 2.633927822113037, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8930019736289978, + "num_tokens": 343620923.0, + "step": 8851 + }, + { + "epoch": 1.1260653860831955, + "grad_norm": 2.6351985931396484, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8878906965255737, + "num_tokens": 343662148.0, + "step": 8852 + }, + { + "epoch": 1.126192596361786, + "grad_norm": 2.664337396621704, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8909720182418823, + "num_tokens": 343704892.0, + "step": 8853 + }, + { + "epoch": 1.1263198066403766, + "grad_norm": 2.3439722061157227, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9045835733413696, + "num_tokens": 343747530.0, + "step": 8854 + }, + { + "epoch": 1.126447016918967, + "grad_norm": 2.7636423110961914, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8890711665153503, + "num_tokens": 343784911.0, + "step": 8855 + }, + { + "epoch": 1.1265742271975576, + "grad_norm": 2.431870460510254, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8965312838554382, + "num_tokens": 343829111.0, + "step": 8856 + }, + { + "epoch": 1.1267014374761481, + "grad_norm": 3.004293918609619, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8903207778930664, + "num_tokens": 343860533.0, + "step": 8857 + }, + { + "epoch": 1.1268286477547387, + "grad_norm": 2.820586681365967, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8857215642929077, + "num_tokens": 343894838.0, + "step": 8858 + }, + { + "epoch": 1.126955858033329, + "grad_norm": 2.6053545475006104, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8878424167633057, + "num_tokens": 343934346.0, + "step": 8859 + }, + { + "epoch": 1.1270830683119195, + "grad_norm": 2.77024245262146, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8918174505233765, + "num_tokens": 343971319.0, + "step": 8860 + }, + { + "epoch": 1.12721027859051, + "grad_norm": 2.7483835220336914, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8874934911727905, + "num_tokens": 344008190.0, + "step": 8861 + }, + { + "epoch": 1.1273374888691006, + "grad_norm": 2.856649875640869, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8893840909004211, + "num_tokens": 344042530.0, + "step": 8862 + }, + { + "epoch": 1.127464699147691, + "grad_norm": 2.408273935317993, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9097610116004944, + "num_tokens": 344079611.0, + "step": 8863 + }, + { + "epoch": 1.1275919094262816, + "grad_norm": 2.4443423748016357, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9030182361602783, + "num_tokens": 344119248.0, + "step": 8864 + }, + { + "epoch": 1.1277191197048722, + "grad_norm": 3.0508956909179688, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8841939568519592, + "num_tokens": 344153286.0, + "step": 8865 + }, + { + "epoch": 1.1278463299834627, + "grad_norm": 2.7402682304382324, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8977807760238647, + "num_tokens": 344188259.0, + "step": 8866 + }, + { + "epoch": 1.1279735402620532, + "grad_norm": 2.46159029006958, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8989295959472656, + "num_tokens": 344229620.0, + "step": 8867 + }, + { + "epoch": 1.1281007505406437, + "grad_norm": 2.725374460220337, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8980010747909546, + "num_tokens": 344264668.0, + "step": 8868 + }, + { + "epoch": 1.1282279608192343, + "grad_norm": 2.7519023418426514, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8871461153030396, + "num_tokens": 344303221.0, + "step": 8869 + }, + { + "epoch": 1.1283551710978248, + "grad_norm": 2.6766538619995117, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8873929977416992, + "num_tokens": 344343280.0, + "step": 8870 + }, + { + "epoch": 1.1284823813764153, + "grad_norm": 2.8518519401550293, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8685051798820496, + "num_tokens": 344383066.0, + "step": 8871 + }, + { + "epoch": 1.1286095916550056, + "grad_norm": 2.6805574893951416, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8931025266647339, + "num_tokens": 344423159.0, + "step": 8872 + }, + { + "epoch": 1.1287368019335962, + "grad_norm": 2.6918599605560303, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8870669603347778, + "num_tokens": 344460265.0, + "step": 8873 + }, + { + "epoch": 1.1288640122121867, + "grad_norm": 2.710805892944336, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8774107694625854, + "num_tokens": 344501197.0, + "step": 8874 + }, + { + "epoch": 1.1289912224907772, + "grad_norm": 2.692863941192627, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8901357650756836, + "num_tokens": 344537891.0, + "step": 8875 + }, + { + "epoch": 1.1291184327693677, + "grad_norm": 2.6748955249786377, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8913576602935791, + "num_tokens": 344572792.0, + "step": 8876 + }, + { + "epoch": 1.1292456430479583, + "grad_norm": 2.5403881072998047, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8881523609161377, + "num_tokens": 344613816.0, + "step": 8877 + }, + { + "epoch": 1.1293728533265488, + "grad_norm": 2.440974712371826, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8920031189918518, + "num_tokens": 344658622.0, + "step": 8878 + }, + { + "epoch": 1.1295000636051393, + "grad_norm": 2.5369908809661865, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9001685380935669, + "num_tokens": 344696991.0, + "step": 8879 + }, + { + "epoch": 1.1296272738837299, + "grad_norm": 2.828045606613159, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8875545859336853, + "num_tokens": 344735255.0, + "step": 8880 + }, + { + "epoch": 1.1297544841623204, + "grad_norm": 2.5262043476104736, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8984017372131348, + "num_tokens": 344773483.0, + "step": 8881 + }, + { + "epoch": 1.129881694440911, + "grad_norm": 2.647367000579834, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8876452445983887, + "num_tokens": 344812649.0, + "step": 8882 + }, + { + "epoch": 1.1300089047195012, + "grad_norm": 2.423644542694092, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.9000568985939026, + "num_tokens": 344855106.0, + "step": 8883 + }, + { + "epoch": 1.1301361149980917, + "grad_norm": 2.6127724647521973, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8819308280944824, + "num_tokens": 344897404.0, + "step": 8884 + }, + { + "epoch": 1.1302633252766823, + "grad_norm": 2.971595525741577, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8712261915206909, + "num_tokens": 344936434.0, + "step": 8885 + }, + { + "epoch": 1.1303905355552728, + "grad_norm": 2.7195427417755127, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8952454924583435, + "num_tokens": 344973976.0, + "step": 8886 + }, + { + "epoch": 1.1305177458338633, + "grad_norm": 2.6368792057037354, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8977396488189697, + "num_tokens": 345011198.0, + "step": 8887 + }, + { + "epoch": 1.1306449561124539, + "grad_norm": 2.466531991958618, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.901026725769043, + "num_tokens": 345051104.0, + "step": 8888 + }, + { + "epoch": 1.1307721663910444, + "grad_norm": 2.481370687484741, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.893357515335083, + "num_tokens": 345092965.0, + "step": 8889 + }, + { + "epoch": 1.130899376669635, + "grad_norm": 2.64870023727417, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8880678415298462, + "num_tokens": 345131001.0, + "step": 8890 + }, + { + "epoch": 1.1310265869482254, + "grad_norm": 2.7847254276275635, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8871340155601501, + "num_tokens": 345165312.0, + "step": 8891 + }, + { + "epoch": 1.131153797226816, + "grad_norm": 2.64795184135437, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8895107507705688, + "num_tokens": 345205600.0, + "step": 8892 + }, + { + "epoch": 1.1312810075054065, + "grad_norm": 2.6524291038513184, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9015673398971558, + "num_tokens": 345243981.0, + "step": 8893 + }, + { + "epoch": 1.131408217783997, + "grad_norm": 2.7146997451782227, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8926007747650146, + "num_tokens": 345278896.0, + "step": 8894 + }, + { + "epoch": 1.1315354280625876, + "grad_norm": 2.489530086517334, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8959729671478271, + "num_tokens": 345322175.0, + "step": 8895 + }, + { + "epoch": 1.131662638341178, + "grad_norm": 2.7977497577667236, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8843820095062256, + "num_tokens": 345360463.0, + "step": 8896 + }, + { + "epoch": 1.1317898486197684, + "grad_norm": 2.7427048683166504, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8900223970413208, + "num_tokens": 345395842.0, + "step": 8897 + }, + { + "epoch": 1.131917058898359, + "grad_norm": 2.584765911102295, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8833519220352173, + "num_tokens": 345438088.0, + "step": 8898 + }, + { + "epoch": 1.1320442691769494, + "grad_norm": 2.2264115810394287, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9067480564117432, + "num_tokens": 345483437.0, + "step": 8899 + }, + { + "epoch": 1.13217147945554, + "grad_norm": 2.9371397495269775, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8759441375732422, + "num_tokens": 345522595.0, + "step": 8900 + }, + { + "epoch": 1.1322986897341305, + "grad_norm": 2.9585983753204346, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8933693766593933, + "num_tokens": 345554283.0, + "step": 8901 + }, + { + "epoch": 1.132425900012721, + "grad_norm": 2.7091197967529297, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8874800801277161, + "num_tokens": 345593841.0, + "step": 8902 + }, + { + "epoch": 1.1325531102913116, + "grad_norm": 3.002380847930908, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8859288692474365, + "num_tokens": 345627645.0, + "step": 8903 + }, + { + "epoch": 1.132680320569902, + "grad_norm": 2.757176637649536, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8997083902359009, + "num_tokens": 345664053.0, + "step": 8904 + }, + { + "epoch": 1.1328075308484926, + "grad_norm": 2.701551914215088, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8991466760635376, + "num_tokens": 345701448.0, + "step": 8905 + }, + { + "epoch": 1.1329347411270831, + "grad_norm": 2.585247039794922, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.895095705986023, + "num_tokens": 345743217.0, + "step": 8906 + }, + { + "epoch": 1.1330619514056737, + "grad_norm": 2.477475643157959, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9020262956619263, + "num_tokens": 345780914.0, + "step": 8907 + }, + { + "epoch": 1.133189161684264, + "grad_norm": 2.50337553024292, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9035553336143494, + "num_tokens": 345820840.0, + "step": 8908 + }, + { + "epoch": 1.1333163719628545, + "grad_norm": 2.361135721206665, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8941512107849121, + "num_tokens": 345866685.0, + "step": 8909 + }, + { + "epoch": 1.133443582241445, + "grad_norm": 2.755962610244751, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8936328887939453, + "num_tokens": 345905467.0, + "step": 8910 + }, + { + "epoch": 1.1335707925200356, + "grad_norm": 2.7126033306121826, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8933680057525635, + "num_tokens": 345941226.0, + "step": 8911 + }, + { + "epoch": 1.133698002798626, + "grad_norm": 2.590545415878296, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8971047401428223, + "num_tokens": 345976363.0, + "step": 8912 + }, + { + "epoch": 1.1338252130772166, + "grad_norm": 2.638258457183838, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.8726962208747864, + "num_tokens": 346023888.0, + "step": 8913 + }, + { + "epoch": 1.1339524233558071, + "grad_norm": 2.6657350063323975, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8813463449478149, + "num_tokens": 346068191.0, + "step": 8914 + }, + { + "epoch": 1.1340796336343977, + "grad_norm": 2.581037759780884, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8756884336471558, + "num_tokens": 346113744.0, + "step": 8915 + }, + { + "epoch": 1.1342068439129882, + "grad_norm": 2.6904165744781494, + "learning_rate": 1e-06, + "loss": 0.3393, + "mean_token_accuracy": 0.8821938633918762, + "num_tokens": 346155554.0, + "step": 8916 + }, + { + "epoch": 1.1343340541915787, + "grad_norm": 2.6954007148742676, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8768563270568848, + "num_tokens": 346197818.0, + "step": 8917 + }, + { + "epoch": 1.1344612644701693, + "grad_norm": 2.5837583541870117, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9112704992294312, + "num_tokens": 346231273.0, + "step": 8918 + }, + { + "epoch": 1.1345884747487598, + "grad_norm": 2.6557321548461914, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8834031224250793, + "num_tokens": 346268865.0, + "step": 8919 + }, + { + "epoch": 1.1347156850273503, + "grad_norm": 2.6888725757598877, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8819395303726196, + "num_tokens": 346308942.0, + "step": 8920 + }, + { + "epoch": 1.1348428953059406, + "grad_norm": 2.838132619857788, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.886327862739563, + "num_tokens": 346346789.0, + "step": 8921 + }, + { + "epoch": 1.1349701055845312, + "grad_norm": 2.949171304702759, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8803117275238037, + "num_tokens": 346382422.0, + "step": 8922 + }, + { + "epoch": 1.1350973158631217, + "grad_norm": 2.6610703468322754, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8882275223731995, + "num_tokens": 346425837.0, + "step": 8923 + }, + { + "epoch": 1.1352245261417122, + "grad_norm": 2.6132652759552, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8823695778846741, + "num_tokens": 346468860.0, + "step": 8924 + }, + { + "epoch": 1.1353517364203027, + "grad_norm": 2.7308568954467773, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8802132606506348, + "num_tokens": 346506484.0, + "step": 8925 + }, + { + "epoch": 1.1354789466988933, + "grad_norm": 2.8792922496795654, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8905326724052429, + "num_tokens": 346539389.0, + "step": 8926 + }, + { + "epoch": 1.1356061569774838, + "grad_norm": 2.5494613647460938, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8831990957260132, + "num_tokens": 346583008.0, + "step": 8927 + }, + { + "epoch": 1.1357333672560743, + "grad_norm": 2.581495523452759, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8847430944442749, + "num_tokens": 346624928.0, + "step": 8928 + }, + { + "epoch": 1.1358605775346649, + "grad_norm": 2.417919874191284, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9027293920516968, + "num_tokens": 346665792.0, + "step": 8929 + }, + { + "epoch": 1.1359877878132554, + "grad_norm": 2.5570263862609863, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9005715847015381, + "num_tokens": 346704875.0, + "step": 8930 + }, + { + "epoch": 1.136114998091846, + "grad_norm": 2.246145725250244, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8920685648918152, + "num_tokens": 346753157.0, + "step": 8931 + }, + { + "epoch": 1.1362422083704362, + "grad_norm": 2.687856912612915, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8944554328918457, + "num_tokens": 346794298.0, + "step": 8932 + }, + { + "epoch": 1.1363694186490267, + "grad_norm": 2.7205190658569336, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8882745504379272, + "num_tokens": 346834839.0, + "step": 8933 + }, + { + "epoch": 1.1364966289276173, + "grad_norm": 2.6807479858398438, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8932583332061768, + "num_tokens": 346876005.0, + "step": 8934 + }, + { + "epoch": 1.1366238392062078, + "grad_norm": 2.4372968673706055, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8887121081352234, + "num_tokens": 346922720.0, + "step": 8935 + }, + { + "epoch": 1.1367510494847983, + "grad_norm": 2.666154384613037, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8923996686935425, + "num_tokens": 346958426.0, + "step": 8936 + }, + { + "epoch": 1.1368782597633889, + "grad_norm": 2.5594897270202637, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8909409046173096, + "num_tokens": 346997696.0, + "step": 8937 + }, + { + "epoch": 1.1370054700419794, + "grad_norm": 3.0267903804779053, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8801262974739075, + "num_tokens": 347032526.0, + "step": 8938 + }, + { + "epoch": 1.13713268032057, + "grad_norm": 2.6304681301116943, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8870997428894043, + "num_tokens": 347076538.0, + "step": 8939 + }, + { + "epoch": 1.1372598905991604, + "grad_norm": 2.4691755771636963, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9079060554504395, + "num_tokens": 347117479.0, + "step": 8940 + }, + { + "epoch": 1.137387100877751, + "grad_norm": 2.4610681533813477, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.905192494392395, + "num_tokens": 347162915.0, + "step": 8941 + }, + { + "epoch": 1.1375143111563415, + "grad_norm": 2.631063461303711, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9050084352493286, + "num_tokens": 347197805.0, + "step": 8942 + }, + { + "epoch": 1.137641521434932, + "grad_norm": 2.553194999694824, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8926457166671753, + "num_tokens": 347239514.0, + "step": 8943 + }, + { + "epoch": 1.1377687317135226, + "grad_norm": 3.057392120361328, + "learning_rate": 1e-06, + "loss": 0.3436, + "mean_token_accuracy": 0.8795852661132812, + "num_tokens": 347273670.0, + "step": 8944 + }, + { + "epoch": 1.137895941992113, + "grad_norm": 2.616086006164551, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8957796692848206, + "num_tokens": 347313457.0, + "step": 8945 + }, + { + "epoch": 1.1380231522707034, + "grad_norm": 2.7836475372314453, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8853724598884583, + "num_tokens": 347349247.0, + "step": 8946 + }, + { + "epoch": 1.138150362549294, + "grad_norm": 2.639092445373535, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.89595627784729, + "num_tokens": 347384280.0, + "step": 8947 + }, + { + "epoch": 1.1382775728278844, + "grad_norm": 2.5904486179351807, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8972863554954529, + "num_tokens": 347421059.0, + "step": 8948 + }, + { + "epoch": 1.138404783106475, + "grad_norm": 2.535529613494873, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.9002552032470703, + "num_tokens": 347461539.0, + "step": 8949 + }, + { + "epoch": 1.1385319933850655, + "grad_norm": 2.8238697052001953, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8880177736282349, + "num_tokens": 347496190.0, + "step": 8950 + }, + { + "epoch": 1.138659203663656, + "grad_norm": 2.8067400455474854, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8899291157722473, + "num_tokens": 347531165.0, + "step": 8951 + }, + { + "epoch": 1.1387864139422466, + "grad_norm": 2.6616275310516357, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8821710348129272, + "num_tokens": 347573000.0, + "step": 8952 + }, + { + "epoch": 1.138913624220837, + "grad_norm": 2.872420310974121, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8838984966278076, + "num_tokens": 347605575.0, + "step": 8953 + }, + { + "epoch": 1.1390408344994276, + "grad_norm": 2.7098891735076904, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8807768225669861, + "num_tokens": 347645634.0, + "step": 8954 + }, + { + "epoch": 1.1391680447780181, + "grad_norm": 2.383054256439209, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9025972485542297, + "num_tokens": 347686541.0, + "step": 8955 + }, + { + "epoch": 1.1392952550566087, + "grad_norm": 2.798835039138794, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8918206691741943, + "num_tokens": 347722997.0, + "step": 8956 + }, + { + "epoch": 1.139422465335199, + "grad_norm": 2.46866774559021, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8849788904190063, + "num_tokens": 347770031.0, + "step": 8957 + }, + { + "epoch": 1.1395496756137895, + "grad_norm": 2.814863681793213, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8857516050338745, + "num_tokens": 347806416.0, + "step": 8958 + }, + { + "epoch": 1.13967688589238, + "grad_norm": 2.6791272163391113, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9006825685501099, + "num_tokens": 347842343.0, + "step": 8959 + }, + { + "epoch": 1.1398040961709706, + "grad_norm": 2.732729911804199, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8897089958190918, + "num_tokens": 347880858.0, + "step": 8960 + }, + { + "epoch": 1.139931306449561, + "grad_norm": 3.0197057723999023, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8961081504821777, + "num_tokens": 347911894.0, + "step": 8961 + }, + { + "epoch": 1.1400585167281516, + "grad_norm": 2.713435649871826, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8853465914726257, + "num_tokens": 347952516.0, + "step": 8962 + }, + { + "epoch": 1.1401857270067421, + "grad_norm": 2.7399308681488037, + "learning_rate": 1e-06, + "loss": 0.3797, + "mean_token_accuracy": 0.8668456077575684, + "num_tokens": 347997836.0, + "step": 8963 + }, + { + "epoch": 1.1403129372853327, + "grad_norm": 2.950404644012451, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8801127672195435, + "num_tokens": 348033733.0, + "step": 8964 + }, + { + "epoch": 1.1404401475639232, + "grad_norm": 2.3934195041656494, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8939885497093201, + "num_tokens": 348076122.0, + "step": 8965 + }, + { + "epoch": 1.1405673578425137, + "grad_norm": 2.5922443866729736, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8910952210426331, + "num_tokens": 348115249.0, + "step": 8966 + }, + { + "epoch": 1.1406945681211043, + "grad_norm": 2.547624349594116, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8971250057220459, + "num_tokens": 348155019.0, + "step": 8967 + }, + { + "epoch": 1.1408217783996948, + "grad_norm": 2.500809907913208, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8965243101119995, + "num_tokens": 348196226.0, + "step": 8968 + }, + { + "epoch": 1.1409489886782853, + "grad_norm": 2.626707077026367, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8938331604003906, + "num_tokens": 348234431.0, + "step": 8969 + }, + { + "epoch": 1.1410761989568756, + "grad_norm": 2.5361504554748535, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9026589393615723, + "num_tokens": 348272841.0, + "step": 8970 + }, + { + "epoch": 1.1412034092354661, + "grad_norm": 2.8291075229644775, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8751943111419678, + "num_tokens": 348314799.0, + "step": 8971 + }, + { + "epoch": 1.1413306195140567, + "grad_norm": 2.7233026027679443, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9056674242019653, + "num_tokens": 348346448.0, + "step": 8972 + }, + { + "epoch": 1.1414578297926472, + "grad_norm": 2.5141613483428955, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9016997814178467, + "num_tokens": 348383881.0, + "step": 8973 + }, + { + "epoch": 1.1415850400712377, + "grad_norm": 2.6187925338745117, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8871191740036011, + "num_tokens": 348427766.0, + "step": 8974 + }, + { + "epoch": 1.1417122503498283, + "grad_norm": 2.689532995223999, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8927778005599976, + "num_tokens": 348462934.0, + "step": 8975 + }, + { + "epoch": 1.1418394606284188, + "grad_norm": 2.722895860671997, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8916373252868652, + "num_tokens": 348498245.0, + "step": 8976 + }, + { + "epoch": 1.1419666709070093, + "grad_norm": 2.5389010906219482, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8987979292869568, + "num_tokens": 348540794.0, + "step": 8977 + }, + { + "epoch": 1.1420938811855998, + "grad_norm": 2.751493215560913, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8785886168479919, + "num_tokens": 348582485.0, + "step": 8978 + }, + { + "epoch": 1.1422210914641904, + "grad_norm": 2.9656989574432373, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8768284916877747, + "num_tokens": 348618983.0, + "step": 8979 + }, + { + "epoch": 1.142348301742781, + "grad_norm": 2.5790603160858154, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8960127234458923, + "num_tokens": 348658972.0, + "step": 8980 + }, + { + "epoch": 1.1424755120213712, + "grad_norm": 2.752629280090332, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8830666542053223, + "num_tokens": 348701111.0, + "step": 8981 + }, + { + "epoch": 1.1426027222999617, + "grad_norm": 2.7784698009490967, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8763159513473511, + "num_tokens": 348740789.0, + "step": 8982 + }, + { + "epoch": 1.1427299325785523, + "grad_norm": 2.9014034271240234, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8809607625007629, + "num_tokens": 348775915.0, + "step": 8983 + }, + { + "epoch": 1.1428571428571428, + "grad_norm": 2.560356855392456, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8930332660675049, + "num_tokens": 348814892.0, + "step": 8984 + }, + { + "epoch": 1.1429843531357333, + "grad_norm": 2.759483814239502, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8902351260185242, + "num_tokens": 348850295.0, + "step": 8985 + }, + { + "epoch": 1.1431115634143239, + "grad_norm": 2.5315330028533936, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8942916393280029, + "num_tokens": 348893538.0, + "step": 8986 + }, + { + "epoch": 1.1432387736929144, + "grad_norm": 2.507190465927124, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9014338254928589, + "num_tokens": 348930211.0, + "step": 8987 + }, + { + "epoch": 1.143365983971505, + "grad_norm": 2.6145336627960205, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8988991379737854, + "num_tokens": 348966657.0, + "step": 8988 + }, + { + "epoch": 1.1434931942500954, + "grad_norm": 2.8335957527160645, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8899179697036743, + "num_tokens": 349000370.0, + "step": 8989 + }, + { + "epoch": 1.143620404528686, + "grad_norm": 2.7088115215301514, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8953402042388916, + "num_tokens": 349036650.0, + "step": 8990 + }, + { + "epoch": 1.1437476148072765, + "grad_norm": 2.722306728363037, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8747156858444214, + "num_tokens": 349076841.0, + "step": 8991 + }, + { + "epoch": 1.143874825085867, + "grad_norm": 2.6535329818725586, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8843368291854858, + "num_tokens": 349121227.0, + "step": 8992 + }, + { + "epoch": 1.1440020353644575, + "grad_norm": 3.0385758876800537, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8918410539627075, + "num_tokens": 349153095.0, + "step": 8993 + }, + { + "epoch": 1.144129245643048, + "grad_norm": 2.4071388244628906, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9036958813667297, + "num_tokens": 349195759.0, + "step": 8994 + }, + { + "epoch": 1.1442564559216384, + "grad_norm": 2.6388487815856934, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.9009521007537842, + "num_tokens": 349230765.0, + "step": 8995 + }, + { + "epoch": 1.144383666200229, + "grad_norm": 2.6410186290740967, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8946957588195801, + "num_tokens": 349268399.0, + "step": 8996 + }, + { + "epoch": 1.1445108764788194, + "grad_norm": 2.7124576568603516, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8868052959442139, + "num_tokens": 349305504.0, + "step": 8997 + }, + { + "epoch": 1.14463808675741, + "grad_norm": 2.8733251094818115, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8678933382034302, + "num_tokens": 349344760.0, + "step": 8998 + }, + { + "epoch": 1.1447652970360005, + "grad_norm": 2.451098918914795, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8988728523254395, + "num_tokens": 349388401.0, + "step": 8999 + }, + { + "epoch": 1.144892507314591, + "grad_norm": 2.574199676513672, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8894979357719421, + "num_tokens": 349426547.0, + "step": 9000 + }, + { + "epoch": 1.1450197175931816, + "grad_norm": 2.5274813175201416, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.8984352946281433, + "num_tokens": 349466845.0, + "step": 9001 + }, + { + "epoch": 1.145146927871772, + "grad_norm": 2.4619033336639404, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8952552080154419, + "num_tokens": 349506665.0, + "step": 9002 + }, + { + "epoch": 1.1452741381503626, + "grad_norm": 2.5169758796691895, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8842411041259766, + "num_tokens": 349546418.0, + "step": 9003 + }, + { + "epoch": 1.1454013484289531, + "grad_norm": 2.6338858604431152, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8878923654556274, + "num_tokens": 349587657.0, + "step": 9004 + }, + { + "epoch": 1.1455285587075437, + "grad_norm": 2.6012330055236816, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8939819931983948, + "num_tokens": 349628334.0, + "step": 9005 + }, + { + "epoch": 1.145655768986134, + "grad_norm": 2.8116700649261475, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8870124220848083, + "num_tokens": 349665908.0, + "step": 9006 + }, + { + "epoch": 1.1457829792647245, + "grad_norm": 2.934250831604004, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8894997835159302, + "num_tokens": 349701720.0, + "step": 9007 + }, + { + "epoch": 1.145910189543315, + "grad_norm": 2.7615818977355957, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.876205325126648, + "num_tokens": 349738747.0, + "step": 9008 + }, + { + "epoch": 1.1460373998219056, + "grad_norm": 2.7313663959503174, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8784301280975342, + "num_tokens": 349777626.0, + "step": 9009 + }, + { + "epoch": 1.146164610100496, + "grad_norm": 2.4764814376831055, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8975207805633545, + "num_tokens": 349821056.0, + "step": 9010 + }, + { + "epoch": 1.1462918203790866, + "grad_norm": 2.9290056228637695, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8831839561462402, + "num_tokens": 349854170.0, + "step": 9011 + }, + { + "epoch": 1.1464190306576771, + "grad_norm": 2.8157765865325928, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8758975863456726, + "num_tokens": 349892929.0, + "step": 9012 + }, + { + "epoch": 1.1465462409362677, + "grad_norm": 2.7774085998535156, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8905870318412781, + "num_tokens": 349931449.0, + "step": 9013 + }, + { + "epoch": 1.1466734512148582, + "grad_norm": 2.6737678050994873, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8795826435089111, + "num_tokens": 349971960.0, + "step": 9014 + }, + { + "epoch": 1.1468006614934487, + "grad_norm": 2.5154855251312256, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9009206295013428, + "num_tokens": 350008326.0, + "step": 9015 + }, + { + "epoch": 1.1469278717720393, + "grad_norm": 2.660001277923584, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8967751860618591, + "num_tokens": 350044413.0, + "step": 9016 + }, + { + "epoch": 1.1470550820506298, + "grad_norm": 2.5655694007873535, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.880155086517334, + "num_tokens": 350086404.0, + "step": 9017 + }, + { + "epoch": 1.1471822923292203, + "grad_norm": 2.6606874465942383, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8873428106307983, + "num_tokens": 350123639.0, + "step": 9018 + }, + { + "epoch": 1.1473095026078106, + "grad_norm": 2.43326473236084, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8960855007171631, + "num_tokens": 350165904.0, + "step": 9019 + }, + { + "epoch": 1.1474367128864011, + "grad_norm": 2.7476272583007812, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8862855434417725, + "num_tokens": 350200817.0, + "step": 9020 + }, + { + "epoch": 1.1475639231649917, + "grad_norm": 2.6786797046661377, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8766182661056519, + "num_tokens": 350243213.0, + "step": 9021 + }, + { + "epoch": 1.1476911334435822, + "grad_norm": 2.5879204273223877, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9007511734962463, + "num_tokens": 350279036.0, + "step": 9022 + }, + { + "epoch": 1.1478183437221727, + "grad_norm": 2.606386661529541, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8987595438957214, + "num_tokens": 350319645.0, + "step": 9023 + }, + { + "epoch": 1.1479455540007633, + "grad_norm": 2.8850924968719482, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8853335976600647, + "num_tokens": 350355770.0, + "step": 9024 + }, + { + "epoch": 1.1480727642793538, + "grad_norm": 2.7740485668182373, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8832775354385376, + "num_tokens": 350394073.0, + "step": 9025 + }, + { + "epoch": 1.1481999745579443, + "grad_norm": 2.8506360054016113, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8885220885276794, + "num_tokens": 350427873.0, + "step": 9026 + }, + { + "epoch": 1.1483271848365348, + "grad_norm": 2.6780433654785156, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8852163553237915, + "num_tokens": 350469800.0, + "step": 9027 + }, + { + "epoch": 1.1484543951151254, + "grad_norm": 2.721775531768799, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8911768198013306, + "num_tokens": 350507756.0, + "step": 9028 + }, + { + "epoch": 1.148581605393716, + "grad_norm": 2.5533604621887207, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8967198729515076, + "num_tokens": 350549506.0, + "step": 9029 + }, + { + "epoch": 1.1487088156723062, + "grad_norm": 2.840198516845703, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8965275287628174, + "num_tokens": 350583838.0, + "step": 9030 + }, + { + "epoch": 1.1488360259508967, + "grad_norm": 2.5015738010406494, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9044109582901001, + "num_tokens": 350621539.0, + "step": 9031 + }, + { + "epoch": 1.1489632362294873, + "grad_norm": 2.6221346855163574, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8913065791130066, + "num_tokens": 350661712.0, + "step": 9032 + }, + { + "epoch": 1.1490904465080778, + "grad_norm": 2.795382022857666, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8753947019577026, + "num_tokens": 350703986.0, + "step": 9033 + }, + { + "epoch": 1.1492176567866683, + "grad_norm": 2.3269379138946533, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9028549194335938, + "num_tokens": 350752994.0, + "step": 9034 + }, + { + "epoch": 1.1493448670652588, + "grad_norm": 3.0485048294067383, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8747916221618652, + "num_tokens": 350789412.0, + "step": 9035 + }, + { + "epoch": 1.1494720773438494, + "grad_norm": 2.6789584159851074, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8847449421882629, + "num_tokens": 350830123.0, + "step": 9036 + }, + { + "epoch": 1.14959928762244, + "grad_norm": 2.499570369720459, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.895719051361084, + "num_tokens": 350874193.0, + "step": 9037 + }, + { + "epoch": 1.1497264979010304, + "grad_norm": 2.5875561237335205, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8822072744369507, + "num_tokens": 350919274.0, + "step": 9038 + }, + { + "epoch": 1.149853708179621, + "grad_norm": 2.5478217601776123, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8815666437149048, + "num_tokens": 350962956.0, + "step": 9039 + }, + { + "epoch": 1.1499809184582115, + "grad_norm": 2.5207154750823975, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8895881175994873, + "num_tokens": 351003664.0, + "step": 9040 + }, + { + "epoch": 1.150108128736802, + "grad_norm": 2.8330318927764893, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8834398984909058, + "num_tokens": 351044064.0, + "step": 9041 + }, + { + "epoch": 1.1502353390153925, + "grad_norm": 2.8604371547698975, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8940569162368774, + "num_tokens": 351079494.0, + "step": 9042 + }, + { + "epoch": 1.150362549293983, + "grad_norm": 2.7567152976989746, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8824091553688049, + "num_tokens": 351117608.0, + "step": 9043 + }, + { + "epoch": 1.1504897595725734, + "grad_norm": 2.8075926303863525, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8919917345046997, + "num_tokens": 351151568.0, + "step": 9044 + }, + { + "epoch": 1.150616969851164, + "grad_norm": 2.6190905570983887, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8954216241836548, + "num_tokens": 351186740.0, + "step": 9045 + }, + { + "epoch": 1.1507441801297544, + "grad_norm": 2.852705478668213, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8935761451721191, + "num_tokens": 351219694.0, + "step": 9046 + }, + { + "epoch": 1.150871390408345, + "grad_norm": 2.8546416759490967, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8919086456298828, + "num_tokens": 351254355.0, + "step": 9047 + }, + { + "epoch": 1.1509986006869355, + "grad_norm": 3.146719217300415, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8853820562362671, + "num_tokens": 351285301.0, + "step": 9048 + }, + { + "epoch": 1.151125810965526, + "grad_norm": 2.6790547370910645, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8952546119689941, + "num_tokens": 351322393.0, + "step": 9049 + }, + { + "epoch": 1.1512530212441165, + "grad_norm": 2.49760365486145, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.898534893989563, + "num_tokens": 351365316.0, + "step": 9050 + }, + { + "epoch": 1.151380231522707, + "grad_norm": 2.5972392559051514, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8975023031234741, + "num_tokens": 351406758.0, + "step": 9051 + }, + { + "epoch": 1.1515074418012976, + "grad_norm": 2.58463191986084, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8919236660003662, + "num_tokens": 351448218.0, + "step": 9052 + }, + { + "epoch": 1.1516346520798881, + "grad_norm": 2.556194305419922, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8973613977432251, + "num_tokens": 351489420.0, + "step": 9053 + }, + { + "epoch": 1.1517618623584787, + "grad_norm": 2.657198667526245, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8897613883018494, + "num_tokens": 351529501.0, + "step": 9054 + }, + { + "epoch": 1.151889072637069, + "grad_norm": 2.49050235748291, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8957257270812988, + "num_tokens": 351568586.0, + "step": 9055 + }, + { + "epoch": 1.1520162829156595, + "grad_norm": 2.597083568572998, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8951467275619507, + "num_tokens": 351610434.0, + "step": 9056 + }, + { + "epoch": 1.15214349319425, + "grad_norm": 2.672086715698242, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8834145665168762, + "num_tokens": 351652411.0, + "step": 9057 + }, + { + "epoch": 1.1522707034728406, + "grad_norm": 2.9608373641967773, + "learning_rate": 1e-06, + "loss": 0.3736, + "mean_token_accuracy": 0.8678553700447083, + "num_tokens": 351692423.0, + "step": 9058 + }, + { + "epoch": 1.152397913751431, + "grad_norm": 3.0809974670410156, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8906257152557373, + "num_tokens": 351723982.0, + "step": 9059 + }, + { + "epoch": 1.1525251240300216, + "grad_norm": 3.125983476638794, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8760948181152344, + "num_tokens": 351757058.0, + "step": 9060 + }, + { + "epoch": 1.1526523343086121, + "grad_norm": 2.7451140880584717, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8828209638595581, + "num_tokens": 351794040.0, + "step": 9061 + }, + { + "epoch": 1.1527795445872027, + "grad_norm": 2.543508529663086, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8988462090492249, + "num_tokens": 351831173.0, + "step": 9062 + }, + { + "epoch": 1.1529067548657932, + "grad_norm": 2.8743667602539062, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.879861056804657, + "num_tokens": 351867632.0, + "step": 9063 + }, + { + "epoch": 1.1530339651443837, + "grad_norm": 2.5303118228912354, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.888761579990387, + "num_tokens": 351908785.0, + "step": 9064 + }, + { + "epoch": 1.1531611754229742, + "grad_norm": 2.5668232440948486, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8909810781478882, + "num_tokens": 351949207.0, + "step": 9065 + }, + { + "epoch": 1.1532883857015648, + "grad_norm": 2.7666313648223877, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8937013149261475, + "num_tokens": 351983966.0, + "step": 9066 + }, + { + "epoch": 1.1534155959801553, + "grad_norm": 2.7083981037139893, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8933190107345581, + "num_tokens": 352024248.0, + "step": 9067 + }, + { + "epoch": 1.1535428062587456, + "grad_norm": 2.593613862991333, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8874192237854004, + "num_tokens": 352068217.0, + "step": 9068 + }, + { + "epoch": 1.1536700165373361, + "grad_norm": 2.774024724960327, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8870828151702881, + "num_tokens": 352106221.0, + "step": 9069 + }, + { + "epoch": 1.1537972268159267, + "grad_norm": 2.5263562202453613, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8949602842330933, + "num_tokens": 352149002.0, + "step": 9070 + }, + { + "epoch": 1.1539244370945172, + "grad_norm": 2.6361477375030518, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8950481414794922, + "num_tokens": 352190630.0, + "step": 9071 + }, + { + "epoch": 1.1540516473731077, + "grad_norm": 2.783001184463501, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8907293677330017, + "num_tokens": 352224725.0, + "step": 9072 + }, + { + "epoch": 1.1541788576516983, + "grad_norm": 2.6730666160583496, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8830747604370117, + "num_tokens": 352269329.0, + "step": 9073 + }, + { + "epoch": 1.1543060679302888, + "grad_norm": 2.68761944770813, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8860490322113037, + "num_tokens": 352309595.0, + "step": 9074 + }, + { + "epoch": 1.1544332782088793, + "grad_norm": 3.076730251312256, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8730301260948181, + "num_tokens": 352342946.0, + "step": 9075 + }, + { + "epoch": 1.1545604884874698, + "grad_norm": 2.421114206314087, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8855414390563965, + "num_tokens": 352386545.0, + "step": 9076 + }, + { + "epoch": 1.1546876987660604, + "grad_norm": 2.5528957843780518, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8858608603477478, + "num_tokens": 352429598.0, + "step": 9077 + }, + { + "epoch": 1.154814909044651, + "grad_norm": 2.5283560752868652, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8875249624252319, + "num_tokens": 352474123.0, + "step": 9078 + }, + { + "epoch": 1.1549421193232412, + "grad_norm": 2.6904637813568115, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8872554302215576, + "num_tokens": 352511590.0, + "step": 9079 + }, + { + "epoch": 1.1550693296018317, + "grad_norm": 3.500960350036621, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8892142176628113, + "num_tokens": 352549864.0, + "step": 9080 + }, + { + "epoch": 1.1551965398804223, + "grad_norm": 2.6725494861602783, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8910484910011292, + "num_tokens": 352590180.0, + "step": 9081 + }, + { + "epoch": 1.1553237501590128, + "grad_norm": 2.596487045288086, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8927908539772034, + "num_tokens": 352629919.0, + "step": 9082 + }, + { + "epoch": 1.1554509604376033, + "grad_norm": 2.4728102684020996, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8981153964996338, + "num_tokens": 352668882.0, + "step": 9083 + }, + { + "epoch": 1.1555781707161938, + "grad_norm": 2.5142953395843506, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8976846933364868, + "num_tokens": 352709954.0, + "step": 9084 + }, + { + "epoch": 1.1557053809947844, + "grad_norm": 2.795203685760498, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.896327018737793, + "num_tokens": 352741481.0, + "step": 9085 + }, + { + "epoch": 1.155832591273375, + "grad_norm": 2.4849724769592285, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8984293341636658, + "num_tokens": 352779998.0, + "step": 9086 + }, + { + "epoch": 1.1559598015519654, + "grad_norm": 2.7622034549713135, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8959771394729614, + "num_tokens": 352812867.0, + "step": 9087 + }, + { + "epoch": 1.156087011830556, + "grad_norm": 2.919377088546753, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8860549926757812, + "num_tokens": 352846774.0, + "step": 9088 + }, + { + "epoch": 1.1562142221091465, + "grad_norm": 2.7953600883483887, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8863857984542847, + "num_tokens": 352881889.0, + "step": 9089 + }, + { + "epoch": 1.156341432387737, + "grad_norm": 2.5597400665283203, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8959352970123291, + "num_tokens": 352921492.0, + "step": 9090 + }, + { + "epoch": 1.1564686426663275, + "grad_norm": 2.7854785919189453, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8937809467315674, + "num_tokens": 352956222.0, + "step": 9091 + }, + { + "epoch": 1.156595852944918, + "grad_norm": 2.804612398147583, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8880505561828613, + "num_tokens": 352992660.0, + "step": 9092 + }, + { + "epoch": 1.1567230632235084, + "grad_norm": 2.9885733127593994, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.898002028465271, + "num_tokens": 353026937.0, + "step": 9093 + }, + { + "epoch": 1.156850273502099, + "grad_norm": 2.4525649547576904, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8995435237884521, + "num_tokens": 353069332.0, + "step": 9094 + }, + { + "epoch": 1.1569774837806894, + "grad_norm": 2.7757585048675537, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.892606794834137, + "num_tokens": 353104869.0, + "step": 9095 + }, + { + "epoch": 1.15710469405928, + "grad_norm": 2.7145090103149414, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8912363052368164, + "num_tokens": 353145485.0, + "step": 9096 + }, + { + "epoch": 1.1572319043378705, + "grad_norm": 2.662635087966919, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.894842803478241, + "num_tokens": 353183283.0, + "step": 9097 + }, + { + "epoch": 1.157359114616461, + "grad_norm": 2.624152183532715, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8912329077720642, + "num_tokens": 353226565.0, + "step": 9098 + }, + { + "epoch": 1.1574863248950515, + "grad_norm": 3.3876419067382812, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8785772323608398, + "num_tokens": 353256089.0, + "step": 9099 + }, + { + "epoch": 1.157613535173642, + "grad_norm": 2.621425151824951, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8823436498641968, + "num_tokens": 353296573.0, + "step": 9100 + }, + { + "epoch": 1.1577407454522326, + "grad_norm": 2.420822858810425, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.8978059887886047, + "num_tokens": 353342348.0, + "step": 9101 + }, + { + "epoch": 1.1578679557308231, + "grad_norm": 2.7057254314422607, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.899308443069458, + "num_tokens": 353377697.0, + "step": 9102 + }, + { + "epoch": 1.1579951660094137, + "grad_norm": 2.620199680328369, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8930412530899048, + "num_tokens": 353419320.0, + "step": 9103 + }, + { + "epoch": 1.158122376288004, + "grad_norm": 2.825653314590454, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8887062072753906, + "num_tokens": 353454037.0, + "step": 9104 + }, + { + "epoch": 1.1582495865665945, + "grad_norm": 2.4343180656433105, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8927737474441528, + "num_tokens": 353497223.0, + "step": 9105 + }, + { + "epoch": 1.158376796845185, + "grad_norm": 2.690596342086792, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8796085119247437, + "num_tokens": 353535389.0, + "step": 9106 + }, + { + "epoch": 1.1585040071237755, + "grad_norm": 2.500401258468628, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8799213767051697, + "num_tokens": 353580295.0, + "step": 9107 + }, + { + "epoch": 1.158631217402366, + "grad_norm": 2.729572296142578, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8965747356414795, + "num_tokens": 353615398.0, + "step": 9108 + }, + { + "epoch": 1.1587584276809566, + "grad_norm": 2.9329259395599365, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8926251530647278, + "num_tokens": 353655742.0, + "step": 9109 + }, + { + "epoch": 1.1588856379595471, + "grad_norm": 2.582245349884033, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8920295238494873, + "num_tokens": 353696745.0, + "step": 9110 + }, + { + "epoch": 1.1590128482381377, + "grad_norm": 2.7847306728363037, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8919556140899658, + "num_tokens": 353731117.0, + "step": 9111 + }, + { + "epoch": 1.1591400585167282, + "grad_norm": 2.6496694087982178, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8816296458244324, + "num_tokens": 353773409.0, + "step": 9112 + }, + { + "epoch": 1.1592672687953187, + "grad_norm": 2.7358922958374023, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8776279091835022, + "num_tokens": 353812413.0, + "step": 9113 + }, + { + "epoch": 1.1593944790739092, + "grad_norm": 2.421684741973877, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8938649296760559, + "num_tokens": 353859991.0, + "step": 9114 + }, + { + "epoch": 1.1595216893524998, + "grad_norm": 3.251354694366455, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8866971135139465, + "num_tokens": 353890023.0, + "step": 9115 + }, + { + "epoch": 1.1596488996310903, + "grad_norm": 2.5094165802001953, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8954471945762634, + "num_tokens": 353931130.0, + "step": 9116 + }, + { + "epoch": 1.1597761099096806, + "grad_norm": 2.7166502475738525, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.884952962398529, + "num_tokens": 353969424.0, + "step": 9117 + }, + { + "epoch": 1.1599033201882711, + "grad_norm": 3.063459873199463, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8735402822494507, + "num_tokens": 354003552.0, + "step": 9118 + }, + { + "epoch": 1.1600305304668617, + "grad_norm": 2.6516757011413574, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.897948145866394, + "num_tokens": 354043200.0, + "step": 9119 + }, + { + "epoch": 1.1601577407454522, + "grad_norm": 2.850572109222412, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8957992792129517, + "num_tokens": 354073905.0, + "step": 9120 + }, + { + "epoch": 1.1602849510240427, + "grad_norm": 2.7285218238830566, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8913655281066895, + "num_tokens": 354112752.0, + "step": 9121 + }, + { + "epoch": 1.1604121613026332, + "grad_norm": 2.6497576236724854, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.90948486328125, + "num_tokens": 354147265.0, + "step": 9122 + }, + { + "epoch": 1.1605393715812238, + "grad_norm": 2.716904878616333, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8809442520141602, + "num_tokens": 354187070.0, + "step": 9123 + }, + { + "epoch": 1.1606665818598143, + "grad_norm": 2.780949831008911, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8940381407737732, + "num_tokens": 354220455.0, + "step": 9124 + }, + { + "epoch": 1.1607937921384048, + "grad_norm": 2.8976552486419678, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8923036456108093, + "num_tokens": 354255856.0, + "step": 9125 + }, + { + "epoch": 1.1609210024169954, + "grad_norm": 2.771149158477783, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8754209280014038, + "num_tokens": 354298309.0, + "step": 9126 + }, + { + "epoch": 1.161048212695586, + "grad_norm": 2.6356348991394043, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8880521655082703, + "num_tokens": 354336879.0, + "step": 9127 + }, + { + "epoch": 1.1611754229741762, + "grad_norm": 2.568021535873413, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8748213052749634, + "num_tokens": 354382123.0, + "step": 9128 + }, + { + "epoch": 1.1613026332527667, + "grad_norm": 2.678363084793091, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9056531190872192, + "num_tokens": 354416739.0, + "step": 9129 + }, + { + "epoch": 1.1614298435313573, + "grad_norm": 2.375014305114746, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8964775800704956, + "num_tokens": 354461201.0, + "step": 9130 + }, + { + "epoch": 1.1615570538099478, + "grad_norm": 2.5730230808258057, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.88371741771698, + "num_tokens": 354501626.0, + "step": 9131 + }, + { + "epoch": 1.1616842640885383, + "grad_norm": 2.7027101516723633, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8919512629508972, + "num_tokens": 354537595.0, + "step": 9132 + }, + { + "epoch": 1.1618114743671288, + "grad_norm": 2.7707667350769043, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8868346810340881, + "num_tokens": 354573658.0, + "step": 9133 + }, + { + "epoch": 1.1619386846457194, + "grad_norm": 2.5866832733154297, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.886622428894043, + "num_tokens": 354613038.0, + "step": 9134 + }, + { + "epoch": 1.16206589492431, + "grad_norm": 2.6818525791168213, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.8704711198806763, + "num_tokens": 354655343.0, + "step": 9135 + }, + { + "epoch": 1.1621931052029004, + "grad_norm": 2.401125192642212, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8909751176834106, + "num_tokens": 354699556.0, + "step": 9136 + }, + { + "epoch": 1.162320315481491, + "grad_norm": 2.924421548843384, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8860667943954468, + "num_tokens": 354733572.0, + "step": 9137 + }, + { + "epoch": 1.1624475257600815, + "grad_norm": 2.653928279876709, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8919159173965454, + "num_tokens": 354774243.0, + "step": 9138 + }, + { + "epoch": 1.162574736038672, + "grad_norm": 2.596682548522949, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8989797830581665, + "num_tokens": 354810573.0, + "step": 9139 + }, + { + "epoch": 1.1627019463172625, + "grad_norm": 2.560556173324585, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8946490287780762, + "num_tokens": 354852407.0, + "step": 9140 + }, + { + "epoch": 1.162829156595853, + "grad_norm": 2.4611501693725586, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.899249255657196, + "num_tokens": 354893764.0, + "step": 9141 + }, + { + "epoch": 1.1629563668744434, + "grad_norm": 2.787478446960449, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8926607966423035, + "num_tokens": 354930093.0, + "step": 9142 + }, + { + "epoch": 1.163083577153034, + "grad_norm": 2.9760067462921143, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8917050361633301, + "num_tokens": 354963754.0, + "step": 9143 + }, + { + "epoch": 1.1632107874316244, + "grad_norm": 2.7125213146209717, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8862495422363281, + "num_tokens": 355003187.0, + "step": 9144 + }, + { + "epoch": 1.163337997710215, + "grad_norm": 2.5691139698028564, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8879954218864441, + "num_tokens": 355045448.0, + "step": 9145 + }, + { + "epoch": 1.1634652079888055, + "grad_norm": 2.465862512588501, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9010285139083862, + "num_tokens": 355084174.0, + "step": 9146 + }, + { + "epoch": 1.163592418267396, + "grad_norm": 2.8314898014068604, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8959616422653198, + "num_tokens": 355115977.0, + "step": 9147 + }, + { + "epoch": 1.1637196285459865, + "grad_norm": 2.621175765991211, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8882026672363281, + "num_tokens": 355158259.0, + "step": 9148 + }, + { + "epoch": 1.163846838824577, + "grad_norm": 2.409334182739258, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9054969549179077, + "num_tokens": 355199468.0, + "step": 9149 + }, + { + "epoch": 1.1639740491031676, + "grad_norm": 2.4822332859039307, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8835675716400146, + "num_tokens": 355244592.0, + "step": 9150 + }, + { + "epoch": 1.1641012593817581, + "grad_norm": 2.5159947872161865, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.893997311592102, + "num_tokens": 355284368.0, + "step": 9151 + }, + { + "epoch": 1.1642284696603487, + "grad_norm": 2.6061251163482666, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8971171975135803, + "num_tokens": 355321162.0, + "step": 9152 + }, + { + "epoch": 1.164355679938939, + "grad_norm": 2.682061195373535, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8889170289039612, + "num_tokens": 355360146.0, + "step": 9153 + }, + { + "epoch": 1.1644828902175295, + "grad_norm": 2.9514858722686768, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8833084106445312, + "num_tokens": 355402088.0, + "step": 9154 + }, + { + "epoch": 1.16461010049612, + "grad_norm": 2.657914161682129, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8857789039611816, + "num_tokens": 355440419.0, + "step": 9155 + }, + { + "epoch": 1.1647373107747105, + "grad_norm": 2.784208297729492, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8964195251464844, + "num_tokens": 355471354.0, + "step": 9156 + }, + { + "epoch": 1.164864521053301, + "grad_norm": 3.0942018032073975, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8887619376182556, + "num_tokens": 355500868.0, + "step": 9157 + }, + { + "epoch": 1.1649917313318916, + "grad_norm": 2.592278480529785, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8981001973152161, + "num_tokens": 355537024.0, + "step": 9158 + }, + { + "epoch": 1.1651189416104821, + "grad_norm": 2.8582823276519775, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.888393759727478, + "num_tokens": 355571918.0, + "step": 9159 + }, + { + "epoch": 1.1652461518890727, + "grad_norm": 2.7377052307128906, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.892932653427124, + "num_tokens": 355605854.0, + "step": 9160 + }, + { + "epoch": 1.1653733621676632, + "grad_norm": 2.3465306758880615, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8983343243598938, + "num_tokens": 355650202.0, + "step": 9161 + }, + { + "epoch": 1.1655005724462537, + "grad_norm": 2.605156898498535, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.891944408416748, + "num_tokens": 355688188.0, + "step": 9162 + }, + { + "epoch": 1.1656277827248442, + "grad_norm": 2.4593546390533447, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8920996189117432, + "num_tokens": 355732110.0, + "step": 9163 + }, + { + "epoch": 1.1657549930034348, + "grad_norm": 2.6897788047790527, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8851807117462158, + "num_tokens": 355772706.0, + "step": 9164 + }, + { + "epoch": 1.1658822032820253, + "grad_norm": 2.672558546066284, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.897390604019165, + "num_tokens": 355807626.0, + "step": 9165 + }, + { + "epoch": 1.1660094135606156, + "grad_norm": 2.6417956352233887, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8911780714988708, + "num_tokens": 355845489.0, + "step": 9166 + }, + { + "epoch": 1.1661366238392061, + "grad_norm": 2.6153924465179443, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8905344009399414, + "num_tokens": 355885121.0, + "step": 9167 + }, + { + "epoch": 1.1662638341177967, + "grad_norm": 2.7275679111480713, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8962161540985107, + "num_tokens": 355923050.0, + "step": 9168 + }, + { + "epoch": 1.1663910443963872, + "grad_norm": 2.83644700050354, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8871768712997437, + "num_tokens": 355960718.0, + "step": 9169 + }, + { + "epoch": 1.1665182546749777, + "grad_norm": 2.968409538269043, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8853185176849365, + "num_tokens": 355993688.0, + "step": 9170 + }, + { + "epoch": 1.1666454649535682, + "grad_norm": 2.5191702842712402, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8953803181648254, + "num_tokens": 356034226.0, + "step": 9171 + }, + { + "epoch": 1.1667726752321588, + "grad_norm": 2.524197578430176, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8884422779083252, + "num_tokens": 356078312.0, + "step": 9172 + }, + { + "epoch": 1.1668998855107493, + "grad_norm": 3.100090742111206, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8823159337043762, + "num_tokens": 356111015.0, + "step": 9173 + }, + { + "epoch": 1.1670270957893398, + "grad_norm": 2.503941297531128, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8937507271766663, + "num_tokens": 356152487.0, + "step": 9174 + }, + { + "epoch": 1.1671543060679304, + "grad_norm": 2.955735445022583, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.8758963346481323, + "num_tokens": 356190225.0, + "step": 9175 + }, + { + "epoch": 1.1672815163465209, + "grad_norm": 2.889629364013672, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8863514065742493, + "num_tokens": 356226088.0, + "step": 9176 + }, + { + "epoch": 1.1674087266251112, + "grad_norm": 2.738006591796875, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8809478282928467, + "num_tokens": 356263503.0, + "step": 9177 + }, + { + "epoch": 1.1675359369037017, + "grad_norm": 2.556931972503662, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8744602203369141, + "num_tokens": 356309678.0, + "step": 9178 + }, + { + "epoch": 1.1676631471822922, + "grad_norm": 2.332157611846924, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9044789671897888, + "num_tokens": 356355413.0, + "step": 9179 + }, + { + "epoch": 1.1677903574608828, + "grad_norm": 2.4191999435424805, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8964519500732422, + "num_tokens": 356395970.0, + "step": 9180 + }, + { + "epoch": 1.1679175677394733, + "grad_norm": 2.7831289768218994, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8910501003265381, + "num_tokens": 356430909.0, + "step": 9181 + }, + { + "epoch": 1.1680447780180638, + "grad_norm": 2.558828353881836, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.899673342704773, + "num_tokens": 356467737.0, + "step": 9182 + }, + { + "epoch": 1.1681719882966544, + "grad_norm": 2.563826084136963, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8938674926757812, + "num_tokens": 356508169.0, + "step": 9183 + }, + { + "epoch": 1.168299198575245, + "grad_norm": 2.610422372817993, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8883088827133179, + "num_tokens": 356551690.0, + "step": 9184 + }, + { + "epoch": 1.1684264088538354, + "grad_norm": 2.576477527618408, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8898344039916992, + "num_tokens": 356593972.0, + "step": 9185 + }, + { + "epoch": 1.168553619132426, + "grad_norm": 2.9930527210235596, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.878588855266571, + "num_tokens": 356626354.0, + "step": 9186 + }, + { + "epoch": 1.1686808294110165, + "grad_norm": 2.8936057090759277, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8904514312744141, + "num_tokens": 356658950.0, + "step": 9187 + }, + { + "epoch": 1.168808039689607, + "grad_norm": 2.612300157546997, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8875041007995605, + "num_tokens": 356701395.0, + "step": 9188 + }, + { + "epoch": 1.1689352499681975, + "grad_norm": 2.6577770709991455, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.896783173084259, + "num_tokens": 356740242.0, + "step": 9189 + }, + { + "epoch": 1.169062460246788, + "grad_norm": 2.71803617477417, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.886893630027771, + "num_tokens": 356779023.0, + "step": 9190 + }, + { + "epoch": 1.1691896705253784, + "grad_norm": 3.0518033504486084, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.8693089485168457, + "num_tokens": 356818191.0, + "step": 9191 + }, + { + "epoch": 1.169316880803969, + "grad_norm": 2.6582651138305664, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8988623023033142, + "num_tokens": 356853576.0, + "step": 9192 + }, + { + "epoch": 1.1694440910825594, + "grad_norm": 2.607713460922241, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9013618230819702, + "num_tokens": 356891187.0, + "step": 9193 + }, + { + "epoch": 1.16957130136115, + "grad_norm": 2.4560797214508057, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8940613269805908, + "num_tokens": 356934684.0, + "step": 9194 + }, + { + "epoch": 1.1696985116397405, + "grad_norm": 2.54331636428833, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8883845806121826, + "num_tokens": 356979380.0, + "step": 9195 + }, + { + "epoch": 1.169825721918331, + "grad_norm": 2.5901033878326416, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8940616846084595, + "num_tokens": 357019694.0, + "step": 9196 + }, + { + "epoch": 1.1699529321969215, + "grad_norm": 2.690991163253784, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8883224725723267, + "num_tokens": 357058829.0, + "step": 9197 + }, + { + "epoch": 1.170080142475512, + "grad_norm": 2.6197800636291504, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8884468674659729, + "num_tokens": 357101659.0, + "step": 9198 + }, + { + "epoch": 1.1702073527541026, + "grad_norm": 2.4869120121002197, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.899156928062439, + "num_tokens": 357142490.0, + "step": 9199 + }, + { + "epoch": 1.1703345630326931, + "grad_norm": 2.921344041824341, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8885393142700195, + "num_tokens": 357174068.0, + "step": 9200 + }, + { + "epoch": 1.1704617733112836, + "grad_norm": 2.5057311058044434, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8801858425140381, + "num_tokens": 357219292.0, + "step": 9201 + }, + { + "epoch": 1.170588983589874, + "grad_norm": 2.6314921379089355, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8767642974853516, + "num_tokens": 357263615.0, + "step": 9202 + }, + { + "epoch": 1.1707161938684645, + "grad_norm": 2.646474838256836, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8903519511222839, + "num_tokens": 357300382.0, + "step": 9203 + }, + { + "epoch": 1.170843404147055, + "grad_norm": 2.7436137199401855, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8897837400436401, + "num_tokens": 357337778.0, + "step": 9204 + }, + { + "epoch": 1.1709706144256455, + "grad_norm": 2.6381494998931885, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8938630819320679, + "num_tokens": 357375331.0, + "step": 9205 + }, + { + "epoch": 1.171097824704236, + "grad_norm": 2.7297167778015137, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8956543207168579, + "num_tokens": 357414004.0, + "step": 9206 + }, + { + "epoch": 1.1712250349828266, + "grad_norm": 2.4058194160461426, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8948594331741333, + "num_tokens": 357458010.0, + "step": 9207 + }, + { + "epoch": 1.1713522452614171, + "grad_norm": 2.474355936050415, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8894163966178894, + "num_tokens": 357500924.0, + "step": 9208 + }, + { + "epoch": 1.1714794555400077, + "grad_norm": 2.6007792949676514, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9020998477935791, + "num_tokens": 357537436.0, + "step": 9209 + }, + { + "epoch": 1.1716066658185982, + "grad_norm": 3.008653163909912, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8741142749786377, + "num_tokens": 357572474.0, + "step": 9210 + }, + { + "epoch": 1.1717338760971887, + "grad_norm": 2.604900598526001, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.901275634765625, + "num_tokens": 357607385.0, + "step": 9211 + }, + { + "epoch": 1.1718610863757792, + "grad_norm": 2.584317922592163, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8939701914787292, + "num_tokens": 357648463.0, + "step": 9212 + }, + { + "epoch": 1.1719882966543698, + "grad_norm": 3.0334982872009277, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.891615092754364, + "num_tokens": 357679943.0, + "step": 9213 + }, + { + "epoch": 1.1721155069329603, + "grad_norm": 2.7690768241882324, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.903084397315979, + "num_tokens": 357712727.0, + "step": 9214 + }, + { + "epoch": 1.1722427172115506, + "grad_norm": 2.8885550498962402, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.887028157711029, + "num_tokens": 357749949.0, + "step": 9215 + }, + { + "epoch": 1.1723699274901411, + "grad_norm": 2.8235793113708496, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8908142447471619, + "num_tokens": 357785388.0, + "step": 9216 + }, + { + "epoch": 1.1724971377687317, + "grad_norm": 2.7140800952911377, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8920196294784546, + "num_tokens": 357820731.0, + "step": 9217 + }, + { + "epoch": 1.1726243480473222, + "grad_norm": 2.505553722381592, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8954627513885498, + "num_tokens": 357862602.0, + "step": 9218 + }, + { + "epoch": 1.1727515583259127, + "grad_norm": 2.720515727996826, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8937222957611084, + "num_tokens": 357899714.0, + "step": 9219 + }, + { + "epoch": 1.1728787686045032, + "grad_norm": 2.6850366592407227, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8961330652236938, + "num_tokens": 357936601.0, + "step": 9220 + }, + { + "epoch": 1.1730059788830938, + "grad_norm": 2.5873661041259766, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8950258493423462, + "num_tokens": 357978624.0, + "step": 9221 + }, + { + "epoch": 1.1731331891616843, + "grad_norm": 2.597576379776001, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8822274208068848, + "num_tokens": 358019246.0, + "step": 9222 + }, + { + "epoch": 1.1732603994402748, + "grad_norm": 2.567991256713867, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8903425931930542, + "num_tokens": 358060929.0, + "step": 9223 + }, + { + "epoch": 1.1733876097188654, + "grad_norm": 2.759390354156494, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8798723220825195, + "num_tokens": 358103677.0, + "step": 9224 + }, + { + "epoch": 1.1735148199974559, + "grad_norm": 2.6766247749328613, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8857073187828064, + "num_tokens": 358139802.0, + "step": 9225 + }, + { + "epoch": 1.1736420302760462, + "grad_norm": 2.800727605819702, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8880307078361511, + "num_tokens": 358176152.0, + "step": 9226 + }, + { + "epoch": 1.1737692405546367, + "grad_norm": 2.3804352283477783, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9022983312606812, + "num_tokens": 358219258.0, + "step": 9227 + }, + { + "epoch": 1.1738964508332272, + "grad_norm": 2.676379919052124, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8813614249229431, + "num_tokens": 358259236.0, + "step": 9228 + }, + { + "epoch": 1.1740236611118178, + "grad_norm": 2.5207021236419678, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8871297836303711, + "num_tokens": 358301505.0, + "step": 9229 + }, + { + "epoch": 1.1741508713904083, + "grad_norm": 2.519686698913574, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8869098424911499, + "num_tokens": 358348055.0, + "step": 9230 + }, + { + "epoch": 1.1742780816689988, + "grad_norm": 2.7660019397735596, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.9008247256278992, + "num_tokens": 358383511.0, + "step": 9231 + }, + { + "epoch": 1.1744052919475894, + "grad_norm": 2.865717887878418, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8785911798477173, + "num_tokens": 358421636.0, + "step": 9232 + }, + { + "epoch": 1.1745325022261799, + "grad_norm": 2.6861276626586914, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8869931101799011, + "num_tokens": 358462622.0, + "step": 9233 + }, + { + "epoch": 1.1746597125047704, + "grad_norm": 2.481008768081665, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8876256942749023, + "num_tokens": 358505732.0, + "step": 9234 + }, + { + "epoch": 1.174786922783361, + "grad_norm": 2.833423614501953, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8746042251586914, + "num_tokens": 358543928.0, + "step": 9235 + }, + { + "epoch": 1.1749141330619515, + "grad_norm": 2.5836522579193115, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8860883712768555, + "num_tokens": 358585081.0, + "step": 9236 + }, + { + "epoch": 1.175041343340542, + "grad_norm": 2.693359375, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.877137303352356, + "num_tokens": 358627192.0, + "step": 9237 + }, + { + "epoch": 1.1751685536191325, + "grad_norm": 2.640092134475708, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8922780752182007, + "num_tokens": 358665466.0, + "step": 9238 + }, + { + "epoch": 1.175295763897723, + "grad_norm": 2.5681726932525635, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8888891935348511, + "num_tokens": 358706949.0, + "step": 9239 + }, + { + "epoch": 1.1754229741763134, + "grad_norm": 2.714459180831909, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.888610303401947, + "num_tokens": 358741983.0, + "step": 9240 + }, + { + "epoch": 1.175550184454904, + "grad_norm": 2.630040407180786, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8936254382133484, + "num_tokens": 358780558.0, + "step": 9241 + }, + { + "epoch": 1.1756773947334944, + "grad_norm": 2.5843088626861572, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8905124664306641, + "num_tokens": 358823446.0, + "step": 9242 + }, + { + "epoch": 1.175804605012085, + "grad_norm": 2.646207809448242, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8879784345626831, + "num_tokens": 358862441.0, + "step": 9243 + }, + { + "epoch": 1.1759318152906755, + "grad_norm": 2.4978811740875244, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9002599716186523, + "num_tokens": 358904200.0, + "step": 9244 + }, + { + "epoch": 1.176059025569266, + "grad_norm": 2.667471408843994, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9061530232429504, + "num_tokens": 358940463.0, + "step": 9245 + }, + { + "epoch": 1.1761862358478565, + "grad_norm": 2.856020927429199, + "learning_rate": 1e-06, + "loss": 0.3549, + "mean_token_accuracy": 0.8749306201934814, + "num_tokens": 358979642.0, + "step": 9246 + }, + { + "epoch": 1.176313446126447, + "grad_norm": 3.147669792175293, + "learning_rate": 1e-06, + "loss": 0.3753, + "mean_token_accuracy": 0.873795747756958, + "num_tokens": 359013799.0, + "step": 9247 + }, + { + "epoch": 1.1764406564050376, + "grad_norm": 2.5860347747802734, + "learning_rate": 1e-06, + "loss": 0.3495, + "mean_token_accuracy": 0.8740644454956055, + "num_tokens": 359059221.0, + "step": 9248 + }, + { + "epoch": 1.1765678666836281, + "grad_norm": 2.5878496170043945, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8927686214447021, + "num_tokens": 359100704.0, + "step": 9249 + }, + { + "epoch": 1.1766950769622184, + "grad_norm": 2.754756450653076, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.906451940536499, + "num_tokens": 359134831.0, + "step": 9250 + }, + { + "epoch": 1.176822287240809, + "grad_norm": 2.609731674194336, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8974407315254211, + "num_tokens": 359172411.0, + "step": 9251 + }, + { + "epoch": 1.1769494975193995, + "grad_norm": 2.604940176010132, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9065457582473755, + "num_tokens": 359206108.0, + "step": 9252 + }, + { + "epoch": 1.17707670779799, + "grad_norm": 2.938819646835327, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8800639510154724, + "num_tokens": 359241015.0, + "step": 9253 + }, + { + "epoch": 1.1772039180765805, + "grad_norm": 2.733649730682373, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8925027847290039, + "num_tokens": 359279252.0, + "step": 9254 + }, + { + "epoch": 1.177331128355171, + "grad_norm": 2.6647138595581055, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8849284052848816, + "num_tokens": 359321736.0, + "step": 9255 + }, + { + "epoch": 1.1774583386337616, + "grad_norm": 2.667048931121826, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.8981497883796692, + "num_tokens": 359355991.0, + "step": 9256 + }, + { + "epoch": 1.1775855489123521, + "grad_norm": 2.708784580230713, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8980764746665955, + "num_tokens": 359392744.0, + "step": 9257 + }, + { + "epoch": 1.1777127591909426, + "grad_norm": 2.7202987670898438, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8950689435005188, + "num_tokens": 359432592.0, + "step": 9258 + }, + { + "epoch": 1.1778399694695332, + "grad_norm": 2.7849395275115967, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8934630751609802, + "num_tokens": 359465908.0, + "step": 9259 + }, + { + "epoch": 1.1779671797481237, + "grad_norm": 2.6635355949401855, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8968630433082581, + "num_tokens": 359501305.0, + "step": 9260 + }, + { + "epoch": 1.1780943900267142, + "grad_norm": 2.385632038116455, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8930025696754456, + "num_tokens": 359547556.0, + "step": 9261 + }, + { + "epoch": 1.1782216003053048, + "grad_norm": 2.787198781967163, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8772122859954834, + "num_tokens": 359586833.0, + "step": 9262 + }, + { + "epoch": 1.1783488105838953, + "grad_norm": 2.68588924407959, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8823599219322205, + "num_tokens": 359626470.0, + "step": 9263 + }, + { + "epoch": 1.1784760208624856, + "grad_norm": 2.638702154159546, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8870749473571777, + "num_tokens": 359668095.0, + "step": 9264 + }, + { + "epoch": 1.1786032311410761, + "grad_norm": 2.484790802001953, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8961806297302246, + "num_tokens": 359708593.0, + "step": 9265 + }, + { + "epoch": 1.1787304414196667, + "grad_norm": 2.6648459434509277, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8892956972122192, + "num_tokens": 359748198.0, + "step": 9266 + }, + { + "epoch": 1.1788576516982572, + "grad_norm": 2.6027233600616455, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.892202615737915, + "num_tokens": 359788328.0, + "step": 9267 + }, + { + "epoch": 1.1789848619768477, + "grad_norm": 2.543069362640381, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.884539008140564, + "num_tokens": 359833333.0, + "step": 9268 + }, + { + "epoch": 1.1791120722554382, + "grad_norm": 2.736032247543335, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8823138475418091, + "num_tokens": 359870778.0, + "step": 9269 + }, + { + "epoch": 1.1792392825340288, + "grad_norm": 2.7785322666168213, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8961681127548218, + "num_tokens": 359904826.0, + "step": 9270 + }, + { + "epoch": 1.1793664928126193, + "grad_norm": 2.657644510269165, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.899669349193573, + "num_tokens": 359946108.0, + "step": 9271 + }, + { + "epoch": 1.1794937030912098, + "grad_norm": 2.61014461517334, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8915095329284668, + "num_tokens": 359984092.0, + "step": 9272 + }, + { + "epoch": 1.1796209133698004, + "grad_norm": 2.891512632369995, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8816484212875366, + "num_tokens": 360022587.0, + "step": 9273 + }, + { + "epoch": 1.1797481236483909, + "grad_norm": 2.422715187072754, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9051616787910461, + "num_tokens": 360062716.0, + "step": 9274 + }, + { + "epoch": 1.1798753339269812, + "grad_norm": 2.5206103324890137, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8972508907318115, + "num_tokens": 360106020.0, + "step": 9275 + }, + { + "epoch": 1.1800025442055717, + "grad_norm": 3.0072481632232666, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8795076012611389, + "num_tokens": 360142605.0, + "step": 9276 + }, + { + "epoch": 1.1801297544841622, + "grad_norm": 2.4272379875183105, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8991562724113464, + "num_tokens": 360181357.0, + "step": 9277 + }, + { + "epoch": 1.1802569647627528, + "grad_norm": 2.5518932342529297, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.894655704498291, + "num_tokens": 360220571.0, + "step": 9278 + }, + { + "epoch": 1.1803841750413433, + "grad_norm": 2.539120674133301, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8932982087135315, + "num_tokens": 360259538.0, + "step": 9279 + }, + { + "epoch": 1.1805113853199338, + "grad_norm": 2.51235032081604, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8994573950767517, + "num_tokens": 360296772.0, + "step": 9280 + }, + { + "epoch": 1.1806385955985244, + "grad_norm": 2.949252128601074, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8818103075027466, + "num_tokens": 360334956.0, + "step": 9281 + }, + { + "epoch": 1.1807658058771149, + "grad_norm": 2.6952505111694336, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8916463851928711, + "num_tokens": 360375059.0, + "step": 9282 + }, + { + "epoch": 1.1808930161557054, + "grad_norm": 2.5235745906829834, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8960400819778442, + "num_tokens": 360415446.0, + "step": 9283 + }, + { + "epoch": 1.181020226434296, + "grad_norm": 2.7799289226531982, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8740754127502441, + "num_tokens": 360453302.0, + "step": 9284 + }, + { + "epoch": 1.1811474367128865, + "grad_norm": 2.731062412261963, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.883319616317749, + "num_tokens": 360494030.0, + "step": 9285 + }, + { + "epoch": 1.181274646991477, + "grad_norm": 2.7562899589538574, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8908661603927612, + "num_tokens": 360533082.0, + "step": 9286 + }, + { + "epoch": 1.1814018572700675, + "grad_norm": 2.2054247856140137, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9031649827957153, + "num_tokens": 360579228.0, + "step": 9287 + }, + { + "epoch": 1.181529067548658, + "grad_norm": 2.814324378967285, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8845596313476562, + "num_tokens": 360615515.0, + "step": 9288 + }, + { + "epoch": 1.1816562778272484, + "grad_norm": 2.7684144973754883, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8924092650413513, + "num_tokens": 360651162.0, + "step": 9289 + }, + { + "epoch": 1.1817834881058389, + "grad_norm": 2.6015846729278564, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8842591643333435, + "num_tokens": 360694910.0, + "step": 9290 + }, + { + "epoch": 1.1819106983844294, + "grad_norm": 2.547098159790039, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9028090238571167, + "num_tokens": 360734991.0, + "step": 9291 + }, + { + "epoch": 1.18203790866302, + "grad_norm": 2.857137680053711, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.885117769241333, + "num_tokens": 360773979.0, + "step": 9292 + }, + { + "epoch": 1.1821651189416105, + "grad_norm": 2.9296419620513916, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8808155655860901, + "num_tokens": 360812179.0, + "step": 9293 + }, + { + "epoch": 1.182292329220201, + "grad_norm": 3.2384843826293945, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8702582120895386, + "num_tokens": 360851036.0, + "step": 9294 + }, + { + "epoch": 1.1824195394987915, + "grad_norm": 2.7670252323150635, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8949055671691895, + "num_tokens": 360883489.0, + "step": 9295 + }, + { + "epoch": 1.182546749777382, + "grad_norm": 2.869072437286377, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8889429569244385, + "num_tokens": 360919393.0, + "step": 9296 + }, + { + "epoch": 1.1826739600559726, + "grad_norm": 2.609781265258789, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9035077095031738, + "num_tokens": 360955488.0, + "step": 9297 + }, + { + "epoch": 1.1828011703345631, + "grad_norm": 2.737002372741699, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8893877267837524, + "num_tokens": 360993332.0, + "step": 9298 + }, + { + "epoch": 1.1829283806131534, + "grad_norm": 2.4896090030670166, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9055581092834473, + "num_tokens": 361030713.0, + "step": 9299 + }, + { + "epoch": 1.183055590891744, + "grad_norm": 2.6006081104278564, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8934755325317383, + "num_tokens": 361068720.0, + "step": 9300 + }, + { + "epoch": 1.1831828011703345, + "grad_norm": 2.6638875007629395, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8812378644943237, + "num_tokens": 361110166.0, + "step": 9301 + }, + { + "epoch": 1.183310011448925, + "grad_norm": 2.86141300201416, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8865618109703064, + "num_tokens": 361146569.0, + "step": 9302 + }, + { + "epoch": 1.1834372217275155, + "grad_norm": 2.82291841506958, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8765125274658203, + "num_tokens": 361187007.0, + "step": 9303 + }, + { + "epoch": 1.183564432006106, + "grad_norm": 2.585327625274658, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.8956263065338135, + "num_tokens": 361224842.0, + "step": 9304 + }, + { + "epoch": 1.1836916422846966, + "grad_norm": 2.8277652263641357, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8795112371444702, + "num_tokens": 361260611.0, + "step": 9305 + }, + { + "epoch": 1.1838188525632871, + "grad_norm": 2.500452995300293, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8893733024597168, + "num_tokens": 361304205.0, + "step": 9306 + }, + { + "epoch": 1.1839460628418776, + "grad_norm": 2.858988046646118, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8853533864021301, + "num_tokens": 361339728.0, + "step": 9307 + }, + { + "epoch": 1.1840732731204682, + "grad_norm": 2.655384063720703, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.89726722240448, + "num_tokens": 361376398.0, + "step": 9308 + }, + { + "epoch": 1.1842004833990587, + "grad_norm": 3.1318838596343994, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8928154706954956, + "num_tokens": 361406282.0, + "step": 9309 + }, + { + "epoch": 1.1843276936776492, + "grad_norm": 2.825038433074951, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.880344033241272, + "num_tokens": 361443955.0, + "step": 9310 + }, + { + "epoch": 1.1844549039562398, + "grad_norm": 2.9901444911956787, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8835564851760864, + "num_tokens": 361477889.0, + "step": 9311 + }, + { + "epoch": 1.1845821142348303, + "grad_norm": 2.65788197517395, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8872417211532593, + "num_tokens": 361517320.0, + "step": 9312 + }, + { + "epoch": 1.1847093245134206, + "grad_norm": 2.4273529052734375, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8912967443466187, + "num_tokens": 361560641.0, + "step": 9313 + }, + { + "epoch": 1.1848365347920111, + "grad_norm": 2.639822244644165, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9048411250114441, + "num_tokens": 361595231.0, + "step": 9314 + }, + { + "epoch": 1.1849637450706016, + "grad_norm": 2.7093188762664795, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8848285675048828, + "num_tokens": 361637304.0, + "step": 9315 + }, + { + "epoch": 1.1850909553491922, + "grad_norm": 2.6274313926696777, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8856599926948547, + "num_tokens": 361678077.0, + "step": 9316 + }, + { + "epoch": 1.1852181656277827, + "grad_norm": 2.7413485050201416, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.889232873916626, + "num_tokens": 361717011.0, + "step": 9317 + }, + { + "epoch": 1.1853453759063732, + "grad_norm": 2.9197304248809814, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8835499286651611, + "num_tokens": 361753957.0, + "step": 9318 + }, + { + "epoch": 1.1854725861849638, + "grad_norm": 2.6314585208892822, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.885051429271698, + "num_tokens": 361796225.0, + "step": 9319 + }, + { + "epoch": 1.1855997964635543, + "grad_norm": 2.761141777038574, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.892715573310852, + "num_tokens": 361833056.0, + "step": 9320 + }, + { + "epoch": 1.1857270067421448, + "grad_norm": 2.5980165004730225, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8809000849723816, + "num_tokens": 361875912.0, + "step": 9321 + }, + { + "epoch": 1.1858542170207353, + "grad_norm": 2.5029807090759277, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8988121747970581, + "num_tokens": 361916551.0, + "step": 9322 + }, + { + "epoch": 1.1859814272993259, + "grad_norm": 2.594203472137451, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8842540979385376, + "num_tokens": 361962634.0, + "step": 9323 + }, + { + "epoch": 1.1861086375779162, + "grad_norm": 2.859941244125366, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8944230079650879, + "num_tokens": 361998267.0, + "step": 9324 + }, + { + "epoch": 1.1862358478565067, + "grad_norm": 2.6522276401519775, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.90213942527771, + "num_tokens": 362035188.0, + "step": 9325 + }, + { + "epoch": 1.1863630581350972, + "grad_norm": 2.5625104904174805, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8838953375816345, + "num_tokens": 362076266.0, + "step": 9326 + }, + { + "epoch": 1.1864902684136878, + "grad_norm": 2.772183656692505, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8924878835678101, + "num_tokens": 362112218.0, + "step": 9327 + }, + { + "epoch": 1.1866174786922783, + "grad_norm": 2.532715082168579, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8983380794525146, + "num_tokens": 362155310.0, + "step": 9328 + }, + { + "epoch": 1.1867446889708688, + "grad_norm": 2.497232437133789, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8844457864761353, + "num_tokens": 362198456.0, + "step": 9329 + }, + { + "epoch": 1.1868718992494594, + "grad_norm": 2.5197596549987793, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8991879820823669, + "num_tokens": 362237516.0, + "step": 9330 + }, + { + "epoch": 1.1869991095280499, + "grad_norm": 2.74106764793396, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9089507460594177, + "num_tokens": 362271202.0, + "step": 9331 + }, + { + "epoch": 1.1871263198066404, + "grad_norm": 2.5127737522125244, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9001846313476562, + "num_tokens": 362310680.0, + "step": 9332 + }, + { + "epoch": 1.187253530085231, + "grad_norm": 2.6283118724823, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8927384614944458, + "num_tokens": 362351602.0, + "step": 9333 + }, + { + "epoch": 1.1873807403638215, + "grad_norm": 2.908900499343872, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8979679942131042, + "num_tokens": 362385132.0, + "step": 9334 + }, + { + "epoch": 1.187507950642412, + "grad_norm": 2.6964006423950195, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8915873765945435, + "num_tokens": 362424053.0, + "step": 9335 + }, + { + "epoch": 1.1876351609210025, + "grad_norm": 2.5326285362243652, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8964704275131226, + "num_tokens": 362464996.0, + "step": 9336 + }, + { + "epoch": 1.187762371199593, + "grad_norm": 2.88435697555542, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.888323187828064, + "num_tokens": 362502837.0, + "step": 9337 + }, + { + "epoch": 1.1878895814781834, + "grad_norm": 2.902507781982422, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8763778209686279, + "num_tokens": 362543267.0, + "step": 9338 + }, + { + "epoch": 1.1880167917567739, + "grad_norm": 2.4311270713806152, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8957692384719849, + "num_tokens": 362585794.0, + "step": 9339 + }, + { + "epoch": 1.1881440020353644, + "grad_norm": 2.6738624572753906, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8913313150405884, + "num_tokens": 362625349.0, + "step": 9340 + }, + { + "epoch": 1.188271212313955, + "grad_norm": 2.85428786277771, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8842897415161133, + "num_tokens": 362660866.0, + "step": 9341 + }, + { + "epoch": 1.1883984225925455, + "grad_norm": 3.125990152359009, + "learning_rate": 1e-06, + "loss": 0.3528, + "mean_token_accuracy": 0.8805654048919678, + "num_tokens": 362696665.0, + "step": 9342 + }, + { + "epoch": 1.188525632871136, + "grad_norm": 2.516998767852783, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8837240934371948, + "num_tokens": 362740069.0, + "step": 9343 + }, + { + "epoch": 1.1886528431497265, + "grad_norm": 2.742690086364746, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8929510116577148, + "num_tokens": 362776481.0, + "step": 9344 + }, + { + "epoch": 1.188780053428317, + "grad_norm": 2.9906439781188965, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8835461735725403, + "num_tokens": 362810446.0, + "step": 9345 + }, + { + "epoch": 1.1889072637069076, + "grad_norm": 2.7801461219787598, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8909103870391846, + "num_tokens": 362846439.0, + "step": 9346 + }, + { + "epoch": 1.189034473985498, + "grad_norm": 2.5157573223114014, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8909618854522705, + "num_tokens": 362889997.0, + "step": 9347 + }, + { + "epoch": 1.1891616842640884, + "grad_norm": 2.729292392730713, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8909639120101929, + "num_tokens": 362926554.0, + "step": 9348 + }, + { + "epoch": 1.189288894542679, + "grad_norm": 2.539735794067383, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8958112001419067, + "num_tokens": 362968958.0, + "step": 9349 + }, + { + "epoch": 1.1894161048212695, + "grad_norm": 2.6527352333068848, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.894099235534668, + "num_tokens": 363004542.0, + "step": 9350 + }, + { + "epoch": 1.18954331509986, + "grad_norm": 2.5051350593566895, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8908838033676147, + "num_tokens": 363049170.0, + "step": 9351 + }, + { + "epoch": 1.1896705253784505, + "grad_norm": 2.6068525314331055, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8966541290283203, + "num_tokens": 363090282.0, + "step": 9352 + }, + { + "epoch": 1.189797735657041, + "grad_norm": 3.1457765102386475, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8926407098770142, + "num_tokens": 363126155.0, + "step": 9353 + }, + { + "epoch": 1.1899249459356316, + "grad_norm": 2.511545419692993, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8850377202033997, + "num_tokens": 363166883.0, + "step": 9354 + }, + { + "epoch": 1.1900521562142221, + "grad_norm": 2.8200812339782715, + "learning_rate": 1e-06, + "loss": 0.3669, + "mean_token_accuracy": 0.8736985921859741, + "num_tokens": 363204961.0, + "step": 9355 + }, + { + "epoch": 1.1901793664928126, + "grad_norm": 2.5063858032226562, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8883826732635498, + "num_tokens": 363248877.0, + "step": 9356 + }, + { + "epoch": 1.1903065767714032, + "grad_norm": 2.7535274028778076, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8817638158798218, + "num_tokens": 363286876.0, + "step": 9357 + }, + { + "epoch": 1.1904337870499937, + "grad_norm": 2.547902822494507, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8922176361083984, + "num_tokens": 363327338.0, + "step": 9358 + }, + { + "epoch": 1.1905609973285842, + "grad_norm": 2.7987306118011475, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8852970600128174, + "num_tokens": 363361446.0, + "step": 9359 + }, + { + "epoch": 1.1906882076071748, + "grad_norm": 2.803617000579834, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8822169303894043, + "num_tokens": 363399290.0, + "step": 9360 + }, + { + "epoch": 1.1908154178857653, + "grad_norm": 2.628344774246216, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8882440328598022, + "num_tokens": 363438263.0, + "step": 9361 + }, + { + "epoch": 1.1909426281643556, + "grad_norm": 2.342493772506714, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8966708183288574, + "num_tokens": 363483539.0, + "step": 9362 + }, + { + "epoch": 1.1910698384429461, + "grad_norm": 2.4268064498901367, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8986517190933228, + "num_tokens": 363527036.0, + "step": 9363 + }, + { + "epoch": 1.1911970487215366, + "grad_norm": 2.9074511528015137, + "learning_rate": 1e-06, + "loss": 0.3481, + "mean_token_accuracy": 0.8764824271202087, + "num_tokens": 363565149.0, + "step": 9364 + }, + { + "epoch": 1.1913242590001272, + "grad_norm": 2.970550298690796, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.892516016960144, + "num_tokens": 363597711.0, + "step": 9365 + }, + { + "epoch": 1.1914514692787177, + "grad_norm": 2.7000255584716797, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8957289457321167, + "num_tokens": 363636388.0, + "step": 9366 + }, + { + "epoch": 1.1915786795573082, + "grad_norm": 2.6433000564575195, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8934646844863892, + "num_tokens": 363677389.0, + "step": 9367 + }, + { + "epoch": 1.1917058898358988, + "grad_norm": 2.920891284942627, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8743688464164734, + "num_tokens": 363713292.0, + "step": 9368 + }, + { + "epoch": 1.1918331001144893, + "grad_norm": 2.7239861488342285, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8890769481658936, + "num_tokens": 363750483.0, + "step": 9369 + }, + { + "epoch": 1.1919603103930798, + "grad_norm": 2.852144479751587, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8827338218688965, + "num_tokens": 363786110.0, + "step": 9370 + }, + { + "epoch": 1.1920875206716703, + "grad_norm": 2.747952938079834, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8751755952835083, + "num_tokens": 363829490.0, + "step": 9371 + }, + { + "epoch": 1.1922147309502609, + "grad_norm": 2.861387014389038, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8798738718032837, + "num_tokens": 363864482.0, + "step": 9372 + }, + { + "epoch": 1.1923419412288512, + "grad_norm": 2.575648069381714, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9049668908119202, + "num_tokens": 363901126.0, + "step": 9373 + }, + { + "epoch": 1.1924691515074417, + "grad_norm": 2.799945116043091, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8933318853378296, + "num_tokens": 363935307.0, + "step": 9374 + }, + { + "epoch": 1.1925963617860322, + "grad_norm": 2.656412363052368, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8936364650726318, + "num_tokens": 363973582.0, + "step": 9375 + }, + { + "epoch": 1.1927235720646228, + "grad_norm": 2.685987710952759, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8939435482025146, + "num_tokens": 364010768.0, + "step": 9376 + }, + { + "epoch": 1.1928507823432133, + "grad_norm": 2.4557178020477295, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8976931571960449, + "num_tokens": 364054049.0, + "step": 9377 + }, + { + "epoch": 1.1929779926218038, + "grad_norm": 2.8210105895996094, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8849837779998779, + "num_tokens": 364093402.0, + "step": 9378 + }, + { + "epoch": 1.1931052029003943, + "grad_norm": 2.624692440032959, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8894649744033813, + "num_tokens": 364131654.0, + "step": 9379 + }, + { + "epoch": 1.1932324131789849, + "grad_norm": 2.4690139293670654, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.8998442888259888, + "num_tokens": 364171718.0, + "step": 9380 + }, + { + "epoch": 1.1933596234575754, + "grad_norm": 2.898895740509033, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8860224485397339, + "num_tokens": 364206622.0, + "step": 9381 + }, + { + "epoch": 1.193486833736166, + "grad_norm": 2.5861012935638428, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9077615737915039, + "num_tokens": 364240393.0, + "step": 9382 + }, + { + "epoch": 1.1936140440147565, + "grad_norm": 2.940661907196045, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8766152262687683, + "num_tokens": 364278390.0, + "step": 9383 + }, + { + "epoch": 1.193741254293347, + "grad_norm": 2.5923655033111572, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8925737142562866, + "num_tokens": 364317053.0, + "step": 9384 + }, + { + "epoch": 1.1938684645719375, + "grad_norm": 3.083315849304199, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8805025815963745, + "num_tokens": 364351002.0, + "step": 9385 + }, + { + "epoch": 1.193995674850528, + "grad_norm": 2.603518009185791, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8792071342468262, + "num_tokens": 364396139.0, + "step": 9386 + }, + { + "epoch": 1.1941228851291183, + "grad_norm": 2.702303171157837, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8942017555236816, + "num_tokens": 364432980.0, + "step": 9387 + }, + { + "epoch": 1.1942500954077089, + "grad_norm": 2.7405645847320557, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8933296203613281, + "num_tokens": 364469372.0, + "step": 9388 + }, + { + "epoch": 1.1943773056862994, + "grad_norm": 2.7024781703948975, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8877273201942444, + "num_tokens": 364507609.0, + "step": 9389 + }, + { + "epoch": 1.19450451596489, + "grad_norm": 2.6323604583740234, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8883647918701172, + "num_tokens": 364547263.0, + "step": 9390 + }, + { + "epoch": 1.1946317262434805, + "grad_norm": 2.898263454437256, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.8625115156173706, + "num_tokens": 364588158.0, + "step": 9391 + }, + { + "epoch": 1.194758936522071, + "grad_norm": 2.940650224685669, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.885752260684967, + "num_tokens": 364622527.0, + "step": 9392 + }, + { + "epoch": 1.1948861468006615, + "grad_norm": 2.5814003944396973, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.900016188621521, + "num_tokens": 364660259.0, + "step": 9393 + }, + { + "epoch": 1.195013357079252, + "grad_norm": 2.7634401321411133, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8784003853797913, + "num_tokens": 364700542.0, + "step": 9394 + }, + { + "epoch": 1.1951405673578426, + "grad_norm": 2.711705207824707, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8978691697120667, + "num_tokens": 364733518.0, + "step": 9395 + }, + { + "epoch": 1.195267777636433, + "grad_norm": 2.78926944732666, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8926815390586853, + "num_tokens": 364767414.0, + "step": 9396 + }, + { + "epoch": 1.1953949879150234, + "grad_norm": 2.705484390258789, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.883622407913208, + "num_tokens": 364804404.0, + "step": 9397 + }, + { + "epoch": 1.195522198193614, + "grad_norm": 2.6989147663116455, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8937639594078064, + "num_tokens": 364845727.0, + "step": 9398 + }, + { + "epoch": 1.1956494084722045, + "grad_norm": 2.5257091522216797, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8911488056182861, + "num_tokens": 364888432.0, + "step": 9399 + }, + { + "epoch": 1.195776618750795, + "grad_norm": 2.531881332397461, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.8984291553497314, + "num_tokens": 364926719.0, + "step": 9400 + }, + { + "epoch": 1.1959038290293855, + "grad_norm": 2.8899223804473877, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8810877799987793, + "num_tokens": 364963593.0, + "step": 9401 + }, + { + "epoch": 1.196031039307976, + "grad_norm": 2.7791810035705566, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.881420910358429, + "num_tokens": 365000323.0, + "step": 9402 + }, + { + "epoch": 1.1961582495865666, + "grad_norm": 2.5925047397613525, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8847379684448242, + "num_tokens": 365045998.0, + "step": 9403 + }, + { + "epoch": 1.196285459865157, + "grad_norm": 3.0611748695373535, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8720701932907104, + "num_tokens": 365077990.0, + "step": 9404 + }, + { + "epoch": 1.1964126701437476, + "grad_norm": 2.5051157474517822, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8940510749816895, + "num_tokens": 365120718.0, + "step": 9405 + }, + { + "epoch": 1.1965398804223382, + "grad_norm": 2.8663973808288574, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8877587914466858, + "num_tokens": 365161304.0, + "step": 9406 + }, + { + "epoch": 1.1966670907009287, + "grad_norm": 3.0886330604553223, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8899385929107666, + "num_tokens": 365197487.0, + "step": 9407 + }, + { + "epoch": 1.1967943009795192, + "grad_norm": 2.4885926246643066, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8905003070831299, + "num_tokens": 365239379.0, + "step": 9408 + }, + { + "epoch": 1.1969215112581097, + "grad_norm": 3.0470097064971924, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8857337832450867, + "num_tokens": 365270742.0, + "step": 9409 + }, + { + "epoch": 1.1970487215367003, + "grad_norm": 2.5756287574768066, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8897584676742554, + "num_tokens": 365313960.0, + "step": 9410 + }, + { + "epoch": 1.1971759318152906, + "grad_norm": 2.471700668334961, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8930730819702148, + "num_tokens": 365357591.0, + "step": 9411 + }, + { + "epoch": 1.1973031420938811, + "grad_norm": 2.7920055389404297, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8857484459877014, + "num_tokens": 365392796.0, + "step": 9412 + }, + { + "epoch": 1.1974303523724716, + "grad_norm": 2.4931137561798096, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8958835601806641, + "num_tokens": 365432516.0, + "step": 9413 + }, + { + "epoch": 1.1975575626510622, + "grad_norm": 2.492537498474121, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8949819207191467, + "num_tokens": 365474606.0, + "step": 9414 + }, + { + "epoch": 1.1976847729296527, + "grad_norm": 2.7239506244659424, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8956650495529175, + "num_tokens": 365510890.0, + "step": 9415 + }, + { + "epoch": 1.1978119832082432, + "grad_norm": 2.6798081398010254, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8959300518035889, + "num_tokens": 365547919.0, + "step": 9416 + }, + { + "epoch": 1.1979391934868338, + "grad_norm": 2.909221887588501, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8852280378341675, + "num_tokens": 365584943.0, + "step": 9417 + }, + { + "epoch": 1.1980664037654243, + "grad_norm": 3.117887020111084, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8823106288909912, + "num_tokens": 365615991.0, + "step": 9418 + }, + { + "epoch": 1.1981936140440148, + "grad_norm": 2.7280917167663574, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8818312883377075, + "num_tokens": 365655385.0, + "step": 9419 + }, + { + "epoch": 1.1983208243226053, + "grad_norm": 2.9541354179382324, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8833910226821899, + "num_tokens": 365690228.0, + "step": 9420 + }, + { + "epoch": 1.1984480346011959, + "grad_norm": 2.6441075801849365, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8935233354568481, + "num_tokens": 365728496.0, + "step": 9421 + }, + { + "epoch": 1.1985752448797862, + "grad_norm": 2.484081745147705, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8891910910606384, + "num_tokens": 365771867.0, + "step": 9422 + }, + { + "epoch": 1.1987024551583767, + "grad_norm": 2.8249149322509766, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8802857398986816, + "num_tokens": 365809213.0, + "step": 9423 + }, + { + "epoch": 1.1988296654369672, + "grad_norm": 2.4794130325317383, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9061132073402405, + "num_tokens": 365849372.0, + "step": 9424 + }, + { + "epoch": 1.1989568757155578, + "grad_norm": 2.689178228378296, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.8972358107566833, + "num_tokens": 365881532.0, + "step": 9425 + }, + { + "epoch": 1.1990840859941483, + "grad_norm": 2.311771869659424, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8971594572067261, + "num_tokens": 365927690.0, + "step": 9426 + }, + { + "epoch": 1.1992112962727388, + "grad_norm": 2.596784830093384, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9004344940185547, + "num_tokens": 365964031.0, + "step": 9427 + }, + { + "epoch": 1.1993385065513293, + "grad_norm": 2.7819926738739014, + "learning_rate": 1e-06, + "loss": 0.3693, + "mean_token_accuracy": 0.8733123540878296, + "num_tokens": 366006003.0, + "step": 9428 + }, + { + "epoch": 1.1994657168299199, + "grad_norm": 2.691911220550537, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.892930269241333, + "num_tokens": 366040444.0, + "step": 9429 + }, + { + "epoch": 1.1995929271085104, + "grad_norm": 2.588682174682617, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8835002779960632, + "num_tokens": 366083725.0, + "step": 9430 + }, + { + "epoch": 1.199720137387101, + "grad_norm": 2.7151095867156982, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8787910342216492, + "num_tokens": 366121519.0, + "step": 9431 + }, + { + "epoch": 1.1998473476656915, + "grad_norm": 2.3765504360198975, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9044758081436157, + "num_tokens": 366163467.0, + "step": 9432 + }, + { + "epoch": 1.199974557944282, + "grad_norm": 2.565000534057617, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.8994707465171814, + "num_tokens": 366200077.0, + "step": 9433 + }, + { + "epoch": 1.2001017682228725, + "grad_norm": 2.4932804107666016, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9133339524269104, + "num_tokens": 366234939.0, + "step": 9434 + }, + { + "epoch": 1.200228978501463, + "grad_norm": 2.5435378551483154, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.897769570350647, + "num_tokens": 366274420.0, + "step": 9435 + }, + { + "epoch": 1.2003561887800533, + "grad_norm": 2.9402058124542236, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8814705014228821, + "num_tokens": 366311511.0, + "step": 9436 + }, + { + "epoch": 1.2004833990586439, + "grad_norm": 2.8147919178009033, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8954108357429504, + "num_tokens": 366346817.0, + "step": 9437 + }, + { + "epoch": 1.2006106093372344, + "grad_norm": 2.6889798641204834, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8773303031921387, + "num_tokens": 366387181.0, + "step": 9438 + }, + { + "epoch": 1.200737819615825, + "grad_norm": 2.4000165462493896, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.897132933139801, + "num_tokens": 366431550.0, + "step": 9439 + }, + { + "epoch": 1.2008650298944155, + "grad_norm": 2.785944938659668, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8820347785949707, + "num_tokens": 366469705.0, + "step": 9440 + }, + { + "epoch": 1.200992240173006, + "grad_norm": 2.3117690086364746, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9001981616020203, + "num_tokens": 366515105.0, + "step": 9441 + }, + { + "epoch": 1.2011194504515965, + "grad_norm": 2.800813913345337, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.889220118522644, + "num_tokens": 366547486.0, + "step": 9442 + }, + { + "epoch": 1.201246660730187, + "grad_norm": 2.7145912647247314, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8809231519699097, + "num_tokens": 366588831.0, + "step": 9443 + }, + { + "epoch": 1.2013738710087776, + "grad_norm": 2.5461690425872803, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8858180046081543, + "num_tokens": 366633036.0, + "step": 9444 + }, + { + "epoch": 1.201501081287368, + "grad_norm": 2.5135035514831543, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8870460987091064, + "num_tokens": 366677000.0, + "step": 9445 + }, + { + "epoch": 1.2016282915659584, + "grad_norm": 2.5059051513671875, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8908861875534058, + "num_tokens": 366717735.0, + "step": 9446 + }, + { + "epoch": 1.201755501844549, + "grad_norm": 2.648557662963867, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8947615027427673, + "num_tokens": 366757557.0, + "step": 9447 + }, + { + "epoch": 1.2018827121231395, + "grad_norm": 2.5058703422546387, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8941078186035156, + "num_tokens": 366799848.0, + "step": 9448 + }, + { + "epoch": 1.20200992240173, + "grad_norm": 2.523653984069824, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9018256068229675, + "num_tokens": 366836324.0, + "step": 9449 + }, + { + "epoch": 1.2021371326803205, + "grad_norm": 2.6797544956207275, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9020200967788696, + "num_tokens": 366873950.0, + "step": 9450 + }, + { + "epoch": 1.202264342958911, + "grad_norm": 2.729560136795044, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8964552283287048, + "num_tokens": 366913589.0, + "step": 9451 + }, + { + "epoch": 1.2023915532375016, + "grad_norm": 2.4520461559295654, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8951091766357422, + "num_tokens": 366957171.0, + "step": 9452 + }, + { + "epoch": 1.202518763516092, + "grad_norm": 2.658388376235962, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8930490016937256, + "num_tokens": 366996495.0, + "step": 9453 + }, + { + "epoch": 1.2026459737946826, + "grad_norm": 2.935936450958252, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8804498314857483, + "num_tokens": 367033584.0, + "step": 9454 + }, + { + "epoch": 1.2027731840732732, + "grad_norm": 2.980435371398926, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8922563791275024, + "num_tokens": 367066931.0, + "step": 9455 + }, + { + "epoch": 1.2029003943518637, + "grad_norm": 2.570316791534424, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8986548185348511, + "num_tokens": 367108399.0, + "step": 9456 + }, + { + "epoch": 1.2030276046304542, + "grad_norm": 2.6028506755828857, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8940613269805908, + "num_tokens": 367147783.0, + "step": 9457 + }, + { + "epoch": 1.2031548149090447, + "grad_norm": 2.6694464683532715, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8918098211288452, + "num_tokens": 367186215.0, + "step": 9458 + }, + { + "epoch": 1.2032820251876353, + "grad_norm": 2.4626972675323486, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8996447920799255, + "num_tokens": 367226706.0, + "step": 9459 + }, + { + "epoch": 1.2034092354662256, + "grad_norm": 2.7617287635803223, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9022331833839417, + "num_tokens": 367266022.0, + "step": 9460 + }, + { + "epoch": 1.203536445744816, + "grad_norm": 2.7558066844940186, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8997288942337036, + "num_tokens": 367301730.0, + "step": 9461 + }, + { + "epoch": 1.2036636560234066, + "grad_norm": 2.6737818717956543, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.895777702331543, + "num_tokens": 367342677.0, + "step": 9462 + }, + { + "epoch": 1.2037908663019972, + "grad_norm": 2.589625120162964, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8932467699050903, + "num_tokens": 367384442.0, + "step": 9463 + }, + { + "epoch": 1.2039180765805877, + "grad_norm": 2.6605632305145264, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8884730339050293, + "num_tokens": 367421825.0, + "step": 9464 + }, + { + "epoch": 1.2040452868591782, + "grad_norm": 2.5588772296905518, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8934430480003357, + "num_tokens": 367463324.0, + "step": 9465 + }, + { + "epoch": 1.2041724971377687, + "grad_norm": 2.7494096755981445, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8869564533233643, + "num_tokens": 367501071.0, + "step": 9466 + }, + { + "epoch": 1.2042997074163593, + "grad_norm": 2.6297738552093506, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.893749475479126, + "num_tokens": 367539960.0, + "step": 9467 + }, + { + "epoch": 1.2044269176949498, + "grad_norm": 2.6267802715301514, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8925483822822571, + "num_tokens": 367580623.0, + "step": 9468 + }, + { + "epoch": 1.2045541279735403, + "grad_norm": 2.507639169692993, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8948767185211182, + "num_tokens": 367621747.0, + "step": 9469 + }, + { + "epoch": 1.2046813382521309, + "grad_norm": 2.6701598167419434, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8910505771636963, + "num_tokens": 367660547.0, + "step": 9470 + }, + { + "epoch": 1.2048085485307212, + "grad_norm": 2.915863513946533, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8892205953598022, + "num_tokens": 367694187.0, + "step": 9471 + }, + { + "epoch": 1.2049357588093117, + "grad_norm": 3.149783134460449, + "learning_rate": 1e-06, + "loss": 0.375, + "mean_token_accuracy": 0.8676941990852356, + "num_tokens": 367730817.0, + "step": 9472 + }, + { + "epoch": 1.2050629690879022, + "grad_norm": 2.7355082035064697, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8923092484474182, + "num_tokens": 367766477.0, + "step": 9473 + }, + { + "epoch": 1.2051901793664928, + "grad_norm": 2.8080930709838867, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8936234712600708, + "num_tokens": 367801353.0, + "step": 9474 + }, + { + "epoch": 1.2053173896450833, + "grad_norm": 2.5811221599578857, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8933297395706177, + "num_tokens": 367840536.0, + "step": 9475 + }, + { + "epoch": 1.2054445999236738, + "grad_norm": 2.9238579273223877, + "learning_rate": 1e-06, + "loss": 0.3911, + "mean_token_accuracy": 0.8666245341300964, + "num_tokens": 367878949.0, + "step": 9476 + }, + { + "epoch": 1.2055718102022643, + "grad_norm": 2.800973653793335, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8998170495033264, + "num_tokens": 367913833.0, + "step": 9477 + }, + { + "epoch": 1.2056990204808549, + "grad_norm": 2.730381488800049, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9006118774414062, + "num_tokens": 367949146.0, + "step": 9478 + }, + { + "epoch": 1.2058262307594454, + "grad_norm": 2.53930926322937, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8946737051010132, + "num_tokens": 367990675.0, + "step": 9479 + }, + { + "epoch": 1.205953441038036, + "grad_norm": 2.631983995437622, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.896598219871521, + "num_tokens": 368029317.0, + "step": 9480 + }, + { + "epoch": 1.2060806513166265, + "grad_norm": 2.9209649562835693, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8846332430839539, + "num_tokens": 368066063.0, + "step": 9481 + }, + { + "epoch": 1.206207861595217, + "grad_norm": 2.5997676849365234, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8953084349632263, + "num_tokens": 368104930.0, + "step": 9482 + }, + { + "epoch": 1.2063350718738075, + "grad_norm": 2.6743929386138916, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8956020474433899, + "num_tokens": 368142120.0, + "step": 9483 + }, + { + "epoch": 1.206462282152398, + "grad_norm": 2.9194929599761963, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8840900659561157, + "num_tokens": 368174472.0, + "step": 9484 + }, + { + "epoch": 1.2065894924309883, + "grad_norm": 2.8524248600006104, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8960776925086975, + "num_tokens": 368207508.0, + "step": 9485 + }, + { + "epoch": 1.2067167027095789, + "grad_norm": 2.846773624420166, + "learning_rate": 1e-06, + "loss": 0.3689, + "mean_token_accuracy": 0.8694977760314941, + "num_tokens": 368245480.0, + "step": 9486 + }, + { + "epoch": 1.2068439129881694, + "grad_norm": 2.886174440383911, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8902726173400879, + "num_tokens": 368279394.0, + "step": 9487 + }, + { + "epoch": 1.20697112326676, + "grad_norm": 2.462064266204834, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8909592628479004, + "num_tokens": 368320450.0, + "step": 9488 + }, + { + "epoch": 1.2070983335453505, + "grad_norm": 2.5204660892486572, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8983462452888489, + "num_tokens": 368361021.0, + "step": 9489 + }, + { + "epoch": 1.207225543823941, + "grad_norm": 2.4570841789245605, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8913180828094482, + "num_tokens": 368403249.0, + "step": 9490 + }, + { + "epoch": 1.2073527541025315, + "grad_norm": 2.839240789413452, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8805642127990723, + "num_tokens": 368439008.0, + "step": 9491 + }, + { + "epoch": 1.207479964381122, + "grad_norm": 2.9477529525756836, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.889097273349762, + "num_tokens": 368469267.0, + "step": 9492 + }, + { + "epoch": 1.2076071746597126, + "grad_norm": 2.6311845779418945, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8885155916213989, + "num_tokens": 368509775.0, + "step": 9493 + }, + { + "epoch": 1.207734384938303, + "grad_norm": 2.4685606956481934, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.888755738735199, + "num_tokens": 368550104.0, + "step": 9494 + }, + { + "epoch": 1.2078615952168934, + "grad_norm": 2.750823497772217, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8878129124641418, + "num_tokens": 368588403.0, + "step": 9495 + }, + { + "epoch": 1.207988805495484, + "grad_norm": 2.433976411819458, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9013900756835938, + "num_tokens": 368630007.0, + "step": 9496 + }, + { + "epoch": 1.2081160157740745, + "grad_norm": 2.483682632446289, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.897510290145874, + "num_tokens": 368674965.0, + "step": 9497 + }, + { + "epoch": 1.208243226052665, + "grad_norm": 2.801266670227051, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8873195648193359, + "num_tokens": 368713188.0, + "step": 9498 + }, + { + "epoch": 1.2083704363312555, + "grad_norm": 2.5327861309051514, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8986791372299194, + "num_tokens": 368750244.0, + "step": 9499 + }, + { + "epoch": 1.208497646609846, + "grad_norm": 2.6632986068725586, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8855122923851013, + "num_tokens": 368791795.0, + "step": 9500 + }, + { + "epoch": 1.2086248568884366, + "grad_norm": 2.8484652042388916, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8926125168800354, + "num_tokens": 368828636.0, + "step": 9501 + }, + { + "epoch": 1.208752067167027, + "grad_norm": 2.920822858810425, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8849812746047974, + "num_tokens": 368864447.0, + "step": 9502 + }, + { + "epoch": 1.2088792774456176, + "grad_norm": 2.736715793609619, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8958179950714111, + "num_tokens": 368899276.0, + "step": 9503 + }, + { + "epoch": 1.2090064877242082, + "grad_norm": 2.769226551055908, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.876885175704956, + "num_tokens": 368939707.0, + "step": 9504 + }, + { + "epoch": 1.2091336980027987, + "grad_norm": 2.4096834659576416, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8940068483352661, + "num_tokens": 368981626.0, + "step": 9505 + }, + { + "epoch": 1.2092609082813892, + "grad_norm": 2.5745506286621094, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8844611644744873, + "num_tokens": 369025903.0, + "step": 9506 + }, + { + "epoch": 1.2093881185599797, + "grad_norm": 2.7561752796173096, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.888957142829895, + "num_tokens": 369066385.0, + "step": 9507 + }, + { + "epoch": 1.2095153288385703, + "grad_norm": 2.5850930213928223, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9021714925765991, + "num_tokens": 369106384.0, + "step": 9508 + }, + { + "epoch": 1.2096425391171606, + "grad_norm": 2.761514902114868, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8857051134109497, + "num_tokens": 369144640.0, + "step": 9509 + }, + { + "epoch": 1.209769749395751, + "grad_norm": 2.63340425491333, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8933148980140686, + "num_tokens": 369183155.0, + "step": 9510 + }, + { + "epoch": 1.2098969596743416, + "grad_norm": 2.7504374980926514, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8854978084564209, + "num_tokens": 369221050.0, + "step": 9511 + }, + { + "epoch": 1.2100241699529322, + "grad_norm": 2.8539812564849854, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8814877867698669, + "num_tokens": 369258022.0, + "step": 9512 + }, + { + "epoch": 1.2101513802315227, + "grad_norm": 2.6537768840789795, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8907535076141357, + "num_tokens": 369294147.0, + "step": 9513 + }, + { + "epoch": 1.2102785905101132, + "grad_norm": 2.861882448196411, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8846938610076904, + "num_tokens": 369330249.0, + "step": 9514 + }, + { + "epoch": 1.2104058007887037, + "grad_norm": 2.771136999130249, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.8976304531097412, + "num_tokens": 369364016.0, + "step": 9515 + }, + { + "epoch": 1.2105330110672943, + "grad_norm": 2.4705843925476074, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8908865451812744, + "num_tokens": 369405972.0, + "step": 9516 + }, + { + "epoch": 1.2106602213458848, + "grad_norm": 2.739933490753174, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8846436738967896, + "num_tokens": 369440583.0, + "step": 9517 + }, + { + "epoch": 1.2107874316244753, + "grad_norm": 2.930356502532959, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8932638168334961, + "num_tokens": 369484407.0, + "step": 9518 + }, + { + "epoch": 1.2109146419030659, + "grad_norm": 2.758350372314453, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8954119086265564, + "num_tokens": 369519311.0, + "step": 9519 + }, + { + "epoch": 1.2110418521816562, + "grad_norm": 2.9290409088134766, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8883066177368164, + "num_tokens": 369552854.0, + "step": 9520 + }, + { + "epoch": 1.2111690624602467, + "grad_norm": 2.800163984298706, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8699449896812439, + "num_tokens": 369595120.0, + "step": 9521 + }, + { + "epoch": 1.2112962727388372, + "grad_norm": 2.5382637977600098, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9015869498252869, + "num_tokens": 369632249.0, + "step": 9522 + }, + { + "epoch": 1.2114234830174277, + "grad_norm": 2.6390793323516846, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8937139511108398, + "num_tokens": 369670630.0, + "step": 9523 + }, + { + "epoch": 1.2115506932960183, + "grad_norm": 2.757328748703003, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8937228918075562, + "num_tokens": 369706717.0, + "step": 9524 + }, + { + "epoch": 1.2116779035746088, + "grad_norm": 2.61777400970459, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.899293065071106, + "num_tokens": 369742909.0, + "step": 9525 + }, + { + "epoch": 1.2118051138531993, + "grad_norm": 2.755845546722412, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8887847661972046, + "num_tokens": 369782153.0, + "step": 9526 + }, + { + "epoch": 1.2119323241317899, + "grad_norm": 2.5748672485351562, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8976396322250366, + "num_tokens": 369822022.0, + "step": 9527 + }, + { + "epoch": 1.2120595344103804, + "grad_norm": 2.8153975009918213, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8738358020782471, + "num_tokens": 369864678.0, + "step": 9528 + }, + { + "epoch": 1.212186744688971, + "grad_norm": 2.576231002807617, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8882713317871094, + "num_tokens": 369903978.0, + "step": 9529 + }, + { + "epoch": 1.2123139549675614, + "grad_norm": 2.423736333847046, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9068799614906311, + "num_tokens": 369942919.0, + "step": 9530 + }, + { + "epoch": 1.212441165246152, + "grad_norm": 2.699648380279541, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8920767307281494, + "num_tokens": 369979595.0, + "step": 9531 + }, + { + "epoch": 1.2125683755247425, + "grad_norm": 2.632351875305176, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8856956958770752, + "num_tokens": 370021471.0, + "step": 9532 + }, + { + "epoch": 1.212695585803333, + "grad_norm": 2.785867214202881, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8891772031784058, + "num_tokens": 370053436.0, + "step": 9533 + }, + { + "epoch": 1.2128227960819233, + "grad_norm": 2.60304594039917, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9012609720230103, + "num_tokens": 370089470.0, + "step": 9534 + }, + { + "epoch": 1.2129500063605139, + "grad_norm": 2.6727728843688965, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8959672451019287, + "num_tokens": 370125402.0, + "step": 9535 + }, + { + "epoch": 1.2130772166391044, + "grad_norm": 2.725792169570923, + "learning_rate": 1e-06, + "loss": 0.3484, + "mean_token_accuracy": 0.8754575848579407, + "num_tokens": 370166905.0, + "step": 9536 + }, + { + "epoch": 1.213204426917695, + "grad_norm": 2.75539231300354, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8906970024108887, + "num_tokens": 370204919.0, + "step": 9537 + }, + { + "epoch": 1.2133316371962855, + "grad_norm": 2.7039132118225098, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9002474546432495, + "num_tokens": 370239226.0, + "step": 9538 + }, + { + "epoch": 1.213458847474876, + "grad_norm": 2.57767915725708, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8944627046585083, + "num_tokens": 370280010.0, + "step": 9539 + }, + { + "epoch": 1.2135860577534665, + "grad_norm": 2.5556013584136963, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.886376142501831, + "num_tokens": 370326212.0, + "step": 9540 + }, + { + "epoch": 1.213713268032057, + "grad_norm": 2.6315736770629883, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8994172811508179, + "num_tokens": 370360944.0, + "step": 9541 + }, + { + "epoch": 1.2138404783106476, + "grad_norm": 2.953275442123413, + "learning_rate": 1e-06, + "loss": 0.3445, + "mean_token_accuracy": 0.8774840235710144, + "num_tokens": 370398746.0, + "step": 9542 + }, + { + "epoch": 1.213967688589238, + "grad_norm": 2.6280105113983154, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8900341987609863, + "num_tokens": 370439898.0, + "step": 9543 + }, + { + "epoch": 1.2140948988678284, + "grad_norm": 3.0417158603668213, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8885028958320618, + "num_tokens": 370473836.0, + "step": 9544 + }, + { + "epoch": 1.214222109146419, + "grad_norm": 2.758864164352417, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8910021781921387, + "num_tokens": 370511295.0, + "step": 9545 + }, + { + "epoch": 1.2143493194250095, + "grad_norm": 2.7176055908203125, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8871151208877563, + "num_tokens": 370552739.0, + "step": 9546 + }, + { + "epoch": 1.2144765297036, + "grad_norm": 3.083364248275757, + "learning_rate": 1e-06, + "loss": 0.3822, + "mean_token_accuracy": 0.8642770051956177, + "num_tokens": 370589338.0, + "step": 9547 + }, + { + "epoch": 1.2146037399821905, + "grad_norm": 2.5231523513793945, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.894332766532898, + "num_tokens": 370629850.0, + "step": 9548 + }, + { + "epoch": 1.214730950260781, + "grad_norm": 3.0759260654449463, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8721302151679993, + "num_tokens": 370668020.0, + "step": 9549 + }, + { + "epoch": 1.2148581605393716, + "grad_norm": 2.5634536743164062, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9053643941879272, + "num_tokens": 370705679.0, + "step": 9550 + }, + { + "epoch": 1.214985370817962, + "grad_norm": 3.0373036861419678, + "learning_rate": 1e-06, + "loss": 0.3546, + "mean_token_accuracy": 0.8764113187789917, + "num_tokens": 370740008.0, + "step": 9551 + }, + { + "epoch": 1.2151125810965526, + "grad_norm": 2.7377431392669678, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8951703310012817, + "num_tokens": 370774822.0, + "step": 9552 + }, + { + "epoch": 1.2152397913751432, + "grad_norm": 2.733966112136841, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8822658061981201, + "num_tokens": 370813389.0, + "step": 9553 + }, + { + "epoch": 1.2153670016537337, + "grad_norm": 2.708422899246216, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8896913528442383, + "num_tokens": 370852858.0, + "step": 9554 + }, + { + "epoch": 1.2154942119323242, + "grad_norm": 2.638768196105957, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8940391540527344, + "num_tokens": 370891014.0, + "step": 9555 + }, + { + "epoch": 1.2156214222109147, + "grad_norm": 2.5922131538391113, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8917313814163208, + "num_tokens": 370933371.0, + "step": 9556 + }, + { + "epoch": 1.2157486324895053, + "grad_norm": 2.548598289489746, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8869956731796265, + "num_tokens": 370977885.0, + "step": 9557 + }, + { + "epoch": 1.2158758427680956, + "grad_norm": 2.6623311042785645, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8894808292388916, + "num_tokens": 371015133.0, + "step": 9558 + }, + { + "epoch": 1.216003053046686, + "grad_norm": 2.404243230819702, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.897180438041687, + "num_tokens": 371061142.0, + "step": 9559 + }, + { + "epoch": 1.2161302633252766, + "grad_norm": 2.6465225219726562, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8907471895217896, + "num_tokens": 371101416.0, + "step": 9560 + }, + { + "epoch": 1.2162574736038672, + "grad_norm": 3.006667137145996, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8786998391151428, + "num_tokens": 371133007.0, + "step": 9561 + }, + { + "epoch": 1.2163846838824577, + "grad_norm": 2.882648229598999, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8985311985015869, + "num_tokens": 371165528.0, + "step": 9562 + }, + { + "epoch": 1.2165118941610482, + "grad_norm": 2.7510993480682373, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8909682035446167, + "num_tokens": 371203967.0, + "step": 9563 + }, + { + "epoch": 1.2166391044396387, + "grad_norm": 2.7768988609313965, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8923095464706421, + "num_tokens": 371239359.0, + "step": 9564 + }, + { + "epoch": 1.2167663147182293, + "grad_norm": 3.1090657711029053, + "learning_rate": 1e-06, + "loss": 0.386, + "mean_token_accuracy": 0.8719574809074402, + "num_tokens": 371278070.0, + "step": 9565 + }, + { + "epoch": 1.2168935249968198, + "grad_norm": 2.744530439376831, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8922317028045654, + "num_tokens": 371312975.0, + "step": 9566 + }, + { + "epoch": 1.2170207352754103, + "grad_norm": 2.681992769241333, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8916511535644531, + "num_tokens": 371353069.0, + "step": 9567 + }, + { + "epoch": 1.2171479455540009, + "grad_norm": 2.88139009475708, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8870636820793152, + "num_tokens": 371386713.0, + "step": 9568 + }, + { + "epoch": 1.2172751558325912, + "grad_norm": 2.757460832595825, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8947765827178955, + "num_tokens": 371421513.0, + "step": 9569 + }, + { + "epoch": 1.2174023661111817, + "grad_norm": 2.7287561893463135, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8901938199996948, + "num_tokens": 371458979.0, + "step": 9570 + }, + { + "epoch": 1.2175295763897722, + "grad_norm": 2.8016035556793213, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.881540060043335, + "num_tokens": 371497695.0, + "step": 9571 + }, + { + "epoch": 1.2176567866683627, + "grad_norm": 2.59567928314209, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8957589864730835, + "num_tokens": 371538250.0, + "step": 9572 + }, + { + "epoch": 1.2177839969469533, + "grad_norm": 2.6317057609558105, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.890251100063324, + "num_tokens": 371579115.0, + "step": 9573 + }, + { + "epoch": 1.2179112072255438, + "grad_norm": 2.434368371963501, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8896837830543518, + "num_tokens": 371625514.0, + "step": 9574 + }, + { + "epoch": 1.2180384175041343, + "grad_norm": 2.6602389812469482, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8943063020706177, + "num_tokens": 371662673.0, + "step": 9575 + }, + { + "epoch": 1.2181656277827249, + "grad_norm": 2.8127048015594482, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8989740014076233, + "num_tokens": 371696254.0, + "step": 9576 + }, + { + "epoch": 1.2182928380613154, + "grad_norm": 2.5823891162872314, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8831182718276978, + "num_tokens": 371738285.0, + "step": 9577 + }, + { + "epoch": 1.218420048339906, + "grad_norm": 2.6384730339050293, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8893587589263916, + "num_tokens": 371777661.0, + "step": 9578 + }, + { + "epoch": 1.2185472586184964, + "grad_norm": 2.767716884613037, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8817775249481201, + "num_tokens": 371815277.0, + "step": 9579 + }, + { + "epoch": 1.218674468897087, + "grad_norm": 2.5847582817077637, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8873744606971741, + "num_tokens": 371854855.0, + "step": 9580 + }, + { + "epoch": 1.2188016791756775, + "grad_norm": 2.3013341426849365, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8881883025169373, + "num_tokens": 371899402.0, + "step": 9581 + }, + { + "epoch": 1.218928889454268, + "grad_norm": 2.8022820949554443, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8983011245727539, + "num_tokens": 371931202.0, + "step": 9582 + }, + { + "epoch": 1.2190560997328583, + "grad_norm": 2.641116142272949, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8875797390937805, + "num_tokens": 371969899.0, + "step": 9583 + }, + { + "epoch": 1.2191833100114489, + "grad_norm": 2.615086555480957, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8826985955238342, + "num_tokens": 372012580.0, + "step": 9584 + }, + { + "epoch": 1.2193105202900394, + "grad_norm": 2.4764301776885986, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8999977707862854, + "num_tokens": 372051391.0, + "step": 9585 + }, + { + "epoch": 1.21943773056863, + "grad_norm": 2.790586471557617, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8790578842163086, + "num_tokens": 372089564.0, + "step": 9586 + }, + { + "epoch": 1.2195649408472204, + "grad_norm": 2.4657554626464844, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9008793830871582, + "num_tokens": 372132233.0, + "step": 9587 + }, + { + "epoch": 1.219692151125811, + "grad_norm": 2.6136629581451416, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8803344368934631, + "num_tokens": 372177558.0, + "step": 9588 + }, + { + "epoch": 1.2198193614044015, + "grad_norm": 2.755218267440796, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8952760696411133, + "num_tokens": 372210539.0, + "step": 9589 + }, + { + "epoch": 1.219946571682992, + "grad_norm": 2.779686450958252, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8848839998245239, + "num_tokens": 372248335.0, + "step": 9590 + }, + { + "epoch": 1.2200737819615826, + "grad_norm": 3.430197238922119, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8981621861457825, + "num_tokens": 372284063.0, + "step": 9591 + }, + { + "epoch": 1.220200992240173, + "grad_norm": 2.6266584396362305, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8978586792945862, + "num_tokens": 372324301.0, + "step": 9592 + }, + { + "epoch": 1.2203282025187634, + "grad_norm": 2.840231418609619, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.887471616268158, + "num_tokens": 372359349.0, + "step": 9593 + }, + { + "epoch": 1.220455412797354, + "grad_norm": 2.56378436088562, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.8993027210235596, + "num_tokens": 372396708.0, + "step": 9594 + }, + { + "epoch": 1.2205826230759445, + "grad_norm": 2.769253969192505, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8879806995391846, + "num_tokens": 372433336.0, + "step": 9595 + }, + { + "epoch": 1.220709833354535, + "grad_norm": 2.549283981323242, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8894120454788208, + "num_tokens": 372477871.0, + "step": 9596 + }, + { + "epoch": 1.2208370436331255, + "grad_norm": 2.67573618888855, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8830863237380981, + "num_tokens": 372519476.0, + "step": 9597 + }, + { + "epoch": 1.220964253911716, + "grad_norm": 2.7955424785614014, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8857616186141968, + "num_tokens": 372556358.0, + "step": 9598 + }, + { + "epoch": 1.2210914641903066, + "grad_norm": 2.346234083175659, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8990391492843628, + "num_tokens": 372600069.0, + "step": 9599 + }, + { + "epoch": 1.221218674468897, + "grad_norm": 2.5732438564300537, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8941433429718018, + "num_tokens": 372639573.0, + "step": 9600 + }, + { + "epoch": 1.2213458847474876, + "grad_norm": 2.7228188514709473, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8800066709518433, + "num_tokens": 372677664.0, + "step": 9601 + }, + { + "epoch": 1.2214730950260781, + "grad_norm": 2.5152766704559326, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.894607663154602, + "num_tokens": 372716837.0, + "step": 9602 + }, + { + "epoch": 1.2216003053046687, + "grad_norm": 2.435206174850464, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9077287912368774, + "num_tokens": 372757804.0, + "step": 9603 + }, + { + "epoch": 1.2217275155832592, + "grad_norm": 2.5195178985595703, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8894941806793213, + "num_tokens": 372801510.0, + "step": 9604 + }, + { + "epoch": 1.2218547258618497, + "grad_norm": 2.6095714569091797, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8813048005104065, + "num_tokens": 372843346.0, + "step": 9605 + }, + { + "epoch": 1.2219819361404403, + "grad_norm": 2.618912696838379, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9041388034820557, + "num_tokens": 372880147.0, + "step": 9606 + }, + { + "epoch": 1.2221091464190306, + "grad_norm": 2.3691108226776123, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8988379240036011, + "num_tokens": 372924772.0, + "step": 9607 + }, + { + "epoch": 1.222236356697621, + "grad_norm": 2.407207727432251, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8966022729873657, + "num_tokens": 372966558.0, + "step": 9608 + }, + { + "epoch": 1.2223635669762116, + "grad_norm": 2.6452724933624268, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.869637131690979, + "num_tokens": 373012122.0, + "step": 9609 + }, + { + "epoch": 1.2224907772548022, + "grad_norm": 2.566377878189087, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8909898400306702, + "num_tokens": 373052175.0, + "step": 9610 + }, + { + "epoch": 1.2226179875333927, + "grad_norm": 2.582871675491333, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8835633397102356, + "num_tokens": 373094995.0, + "step": 9611 + }, + { + "epoch": 1.2227451978119832, + "grad_norm": 2.522191047668457, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.893745481967926, + "num_tokens": 373135499.0, + "step": 9612 + }, + { + "epoch": 1.2228724080905737, + "grad_norm": 2.623504161834717, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8888638019561768, + "num_tokens": 373173767.0, + "step": 9613 + }, + { + "epoch": 1.2229996183691643, + "grad_norm": 2.678988218307495, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8886045217514038, + "num_tokens": 373213127.0, + "step": 9614 + }, + { + "epoch": 1.2231268286477548, + "grad_norm": 2.7115225791931152, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8938254117965698, + "num_tokens": 373252386.0, + "step": 9615 + }, + { + "epoch": 1.2232540389263453, + "grad_norm": 2.5963199138641357, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8807169198989868, + "num_tokens": 373295568.0, + "step": 9616 + }, + { + "epoch": 1.2233812492049359, + "grad_norm": 2.6931161880493164, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8923633694648743, + "num_tokens": 373336427.0, + "step": 9617 + }, + { + "epoch": 1.2235084594835262, + "grad_norm": 2.6027896404266357, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8956933617591858, + "num_tokens": 373375602.0, + "step": 9618 + }, + { + "epoch": 1.2236356697621167, + "grad_norm": 2.468836545944214, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.8985085487365723, + "num_tokens": 373418234.0, + "step": 9619 + }, + { + "epoch": 1.2237628800407072, + "grad_norm": 2.6102750301361084, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8908296823501587, + "num_tokens": 373458592.0, + "step": 9620 + }, + { + "epoch": 1.2238900903192977, + "grad_norm": 2.8889498710632324, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8907324075698853, + "num_tokens": 373491191.0, + "step": 9621 + }, + { + "epoch": 1.2240173005978883, + "grad_norm": 2.8363513946533203, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8840755820274353, + "num_tokens": 373527367.0, + "step": 9622 + }, + { + "epoch": 1.2241445108764788, + "grad_norm": 2.5709736347198486, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8993717432022095, + "num_tokens": 373562833.0, + "step": 9623 + }, + { + "epoch": 1.2242717211550693, + "grad_norm": 2.506002902984619, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8919262886047363, + "num_tokens": 373605854.0, + "step": 9624 + }, + { + "epoch": 1.2243989314336599, + "grad_norm": 2.842406749725342, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8916741609573364, + "num_tokens": 373638807.0, + "step": 9625 + }, + { + "epoch": 1.2245261417122504, + "grad_norm": 2.7234864234924316, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8938391208648682, + "num_tokens": 373676178.0, + "step": 9626 + }, + { + "epoch": 1.224653351990841, + "grad_norm": 2.595590591430664, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8886696696281433, + "num_tokens": 373719045.0, + "step": 9627 + }, + { + "epoch": 1.2247805622694314, + "grad_norm": 2.5845744609832764, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8972848653793335, + "num_tokens": 373758640.0, + "step": 9628 + }, + { + "epoch": 1.224907772548022, + "grad_norm": 2.8706772327423096, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8869928121566772, + "num_tokens": 373793829.0, + "step": 9629 + }, + { + "epoch": 1.2250349828266125, + "grad_norm": 2.7767481803894043, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8818726539611816, + "num_tokens": 373831874.0, + "step": 9630 + }, + { + "epoch": 1.225162193105203, + "grad_norm": 2.5263888835906982, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8966552019119263, + "num_tokens": 373874030.0, + "step": 9631 + }, + { + "epoch": 1.2252894033837933, + "grad_norm": 2.9235005378723145, + "learning_rate": 1e-06, + "loss": 0.3809, + "mean_token_accuracy": 0.8673884868621826, + "num_tokens": 373912380.0, + "step": 9632 + }, + { + "epoch": 1.2254166136623839, + "grad_norm": 2.9828526973724365, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.868320643901825, + "num_tokens": 373948627.0, + "step": 9633 + }, + { + "epoch": 1.2255438239409744, + "grad_norm": 2.784048318862915, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8883453607559204, + "num_tokens": 373987583.0, + "step": 9634 + }, + { + "epoch": 1.225671034219565, + "grad_norm": 2.947389602661133, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8837472200393677, + "num_tokens": 374026347.0, + "step": 9635 + }, + { + "epoch": 1.2257982444981554, + "grad_norm": 2.4766743183135986, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8879091739654541, + "num_tokens": 374072462.0, + "step": 9636 + }, + { + "epoch": 1.225925454776746, + "grad_norm": 2.5887060165405273, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8901600241661072, + "num_tokens": 374111023.0, + "step": 9637 + }, + { + "epoch": 1.2260526650553365, + "grad_norm": 3.0169992446899414, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8865176439285278, + "num_tokens": 374142222.0, + "step": 9638 + }, + { + "epoch": 1.226179875333927, + "grad_norm": 2.728879451751709, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8897974491119385, + "num_tokens": 374181138.0, + "step": 9639 + }, + { + "epoch": 1.2263070856125176, + "grad_norm": 2.5159459114074707, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9004181623458862, + "num_tokens": 374217859.0, + "step": 9640 + }, + { + "epoch": 1.226434295891108, + "grad_norm": 2.5222549438476562, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.893332839012146, + "num_tokens": 374259434.0, + "step": 9641 + }, + { + "epoch": 1.2265615061696984, + "grad_norm": 2.467445135116577, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8969420194625854, + "num_tokens": 374301972.0, + "step": 9642 + }, + { + "epoch": 1.226688716448289, + "grad_norm": 2.4959053993225098, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9061832427978516, + "num_tokens": 374337648.0, + "step": 9643 + }, + { + "epoch": 1.2268159267268794, + "grad_norm": 2.65519118309021, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.8985316753387451, + "num_tokens": 374372854.0, + "step": 9644 + }, + { + "epoch": 1.22694313700547, + "grad_norm": 2.574077606201172, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8859999775886536, + "num_tokens": 374414947.0, + "step": 9645 + }, + { + "epoch": 1.2270703472840605, + "grad_norm": 2.740133047103882, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8938683867454529, + "num_tokens": 374448110.0, + "step": 9646 + }, + { + "epoch": 1.227197557562651, + "grad_norm": 2.9219589233398438, + "learning_rate": 1e-06, + "loss": 0.3491, + "mean_token_accuracy": 0.8824871778488159, + "num_tokens": 374489659.0, + "step": 9647 + }, + { + "epoch": 1.2273247678412416, + "grad_norm": 2.688734531402588, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8917407989501953, + "num_tokens": 374526602.0, + "step": 9648 + }, + { + "epoch": 1.227451978119832, + "grad_norm": 3.129065990447998, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8859894275665283, + "num_tokens": 374557265.0, + "step": 9649 + }, + { + "epoch": 1.2275791883984226, + "grad_norm": 2.793280839920044, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8721696138381958, + "num_tokens": 374596865.0, + "step": 9650 + }, + { + "epoch": 1.2277063986770131, + "grad_norm": 2.525280714035034, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8915685415267944, + "num_tokens": 374637667.0, + "step": 9651 + }, + { + "epoch": 1.2278336089556037, + "grad_norm": 2.633759021759033, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8850367069244385, + "num_tokens": 374678217.0, + "step": 9652 + }, + { + "epoch": 1.2279608192341942, + "grad_norm": 2.9107136726379395, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8871685266494751, + "num_tokens": 374715402.0, + "step": 9653 + }, + { + "epoch": 1.2280880295127847, + "grad_norm": 2.4367125034332275, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9028788805007935, + "num_tokens": 374754076.0, + "step": 9654 + }, + { + "epoch": 1.2282152397913753, + "grad_norm": 2.725557565689087, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8953174352645874, + "num_tokens": 374791517.0, + "step": 9655 + }, + { + "epoch": 1.2283424500699656, + "grad_norm": 2.527479648590088, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.880559504032135, + "num_tokens": 374837953.0, + "step": 9656 + }, + { + "epoch": 1.228469660348556, + "grad_norm": 2.3541066646575928, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8985695838928223, + "num_tokens": 374880329.0, + "step": 9657 + }, + { + "epoch": 1.2285968706271466, + "grad_norm": 2.475137233734131, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8941694498062134, + "num_tokens": 374921045.0, + "step": 9658 + }, + { + "epoch": 1.2287240809057371, + "grad_norm": 2.857257604598999, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8816840648651123, + "num_tokens": 374955657.0, + "step": 9659 + }, + { + "epoch": 1.2288512911843277, + "grad_norm": 2.741166591644287, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8893918991088867, + "num_tokens": 374992977.0, + "step": 9660 + }, + { + "epoch": 1.2289785014629182, + "grad_norm": 2.89219331741333, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8780760169029236, + "num_tokens": 375031958.0, + "step": 9661 + }, + { + "epoch": 1.2291057117415087, + "grad_norm": 2.842325448989868, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8989828824996948, + "num_tokens": 375065664.0, + "step": 9662 + }, + { + "epoch": 1.2292329220200993, + "grad_norm": 2.7278971672058105, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8775026798248291, + "num_tokens": 375106454.0, + "step": 9663 + }, + { + "epoch": 1.2293601322986898, + "grad_norm": 2.823424816131592, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8828820586204529, + "num_tokens": 375145398.0, + "step": 9664 + }, + { + "epoch": 1.2294873425772803, + "grad_norm": 2.6769626140594482, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9034989476203918, + "num_tokens": 375181398.0, + "step": 9665 + }, + { + "epoch": 1.2296145528558708, + "grad_norm": 2.6458024978637695, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8879653215408325, + "num_tokens": 375223702.0, + "step": 9666 + }, + { + "epoch": 1.2297417631344612, + "grad_norm": 2.926215887069702, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8862674832344055, + "num_tokens": 375257200.0, + "step": 9667 + }, + { + "epoch": 1.2298689734130517, + "grad_norm": 2.7280259132385254, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8847091197967529, + "num_tokens": 375299397.0, + "step": 9668 + }, + { + "epoch": 1.2299961836916422, + "grad_norm": 2.587449550628662, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.893606424331665, + "num_tokens": 375339625.0, + "step": 9669 + }, + { + "epoch": 1.2301233939702327, + "grad_norm": 2.8966336250305176, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8974598050117493, + "num_tokens": 375374480.0, + "step": 9670 + }, + { + "epoch": 1.2302506042488233, + "grad_norm": 2.7054193019866943, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.8754562139511108, + "num_tokens": 375415097.0, + "step": 9671 + }, + { + "epoch": 1.2303778145274138, + "grad_norm": 2.5372869968414307, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9014430642127991, + "num_tokens": 375451415.0, + "step": 9672 + }, + { + "epoch": 1.2305050248060043, + "grad_norm": 2.6916866302490234, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.893338680267334, + "num_tokens": 375492915.0, + "step": 9673 + }, + { + "epoch": 1.2306322350845948, + "grad_norm": 2.447599411010742, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.892679750919342, + "num_tokens": 375535482.0, + "step": 9674 + }, + { + "epoch": 1.2307594453631854, + "grad_norm": 2.8666629791259766, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.8996482491493225, + "num_tokens": 375569018.0, + "step": 9675 + }, + { + "epoch": 1.230886655641776, + "grad_norm": 2.568106174468994, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8955038189888, + "num_tokens": 375607575.0, + "step": 9676 + }, + { + "epoch": 1.2310138659203664, + "grad_norm": 2.712101936340332, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8954636454582214, + "num_tokens": 375642569.0, + "step": 9677 + }, + { + "epoch": 1.231141076198957, + "grad_norm": 2.6127877235412598, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8898168206214905, + "num_tokens": 375683293.0, + "step": 9678 + }, + { + "epoch": 1.2312682864775475, + "grad_norm": 2.625495672225952, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8860824704170227, + "num_tokens": 375724112.0, + "step": 9679 + }, + { + "epoch": 1.231395496756138, + "grad_norm": 2.5512919425964355, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8947056531906128, + "num_tokens": 375762743.0, + "step": 9680 + }, + { + "epoch": 1.2315227070347283, + "grad_norm": 2.9894399642944336, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8841272592544556, + "num_tokens": 375800894.0, + "step": 9681 + }, + { + "epoch": 1.2316499173133189, + "grad_norm": 2.628265619277954, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8943626284599304, + "num_tokens": 375837420.0, + "step": 9682 + }, + { + "epoch": 1.2317771275919094, + "grad_norm": 2.9317798614501953, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8983451128005981, + "num_tokens": 375869984.0, + "step": 9683 + }, + { + "epoch": 1.2319043378705, + "grad_norm": 2.6752398014068604, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8865101933479309, + "num_tokens": 375907188.0, + "step": 9684 + }, + { + "epoch": 1.2320315481490904, + "grad_norm": 2.6250693798065186, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.893351137638092, + "num_tokens": 375946182.0, + "step": 9685 + }, + { + "epoch": 1.232158758427681, + "grad_norm": 2.7003583908081055, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8868327140808105, + "num_tokens": 375984797.0, + "step": 9686 + }, + { + "epoch": 1.2322859687062715, + "grad_norm": 2.6936216354370117, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8847885131835938, + "num_tokens": 376021937.0, + "step": 9687 + }, + { + "epoch": 1.232413178984862, + "grad_norm": 2.503571033477783, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.899149477481842, + "num_tokens": 376063067.0, + "step": 9688 + }, + { + "epoch": 1.2325403892634526, + "grad_norm": 2.6891140937805176, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8873498439788818, + "num_tokens": 376106212.0, + "step": 9689 + }, + { + "epoch": 1.232667599542043, + "grad_norm": 2.482365369796753, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.893950343132019, + "num_tokens": 376152237.0, + "step": 9690 + }, + { + "epoch": 1.2327948098206334, + "grad_norm": 2.8572304248809814, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.884372353553772, + "num_tokens": 376187689.0, + "step": 9691 + }, + { + "epoch": 1.232922020099224, + "grad_norm": 2.707878589630127, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.893280565738678, + "num_tokens": 376222203.0, + "step": 9692 + }, + { + "epoch": 1.2330492303778144, + "grad_norm": 2.6119518280029297, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8900994062423706, + "num_tokens": 376260039.0, + "step": 9693 + }, + { + "epoch": 1.233176440656405, + "grad_norm": 2.7083327770233154, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8919610381126404, + "num_tokens": 376297169.0, + "step": 9694 + }, + { + "epoch": 1.2333036509349955, + "grad_norm": 2.920393705368042, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8791757225990295, + "num_tokens": 376331209.0, + "step": 9695 + }, + { + "epoch": 1.233430861213586, + "grad_norm": 2.743751049041748, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8881821036338806, + "num_tokens": 376367939.0, + "step": 9696 + }, + { + "epoch": 1.2335580714921766, + "grad_norm": 2.6749212741851807, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8760918378829956, + "num_tokens": 376409945.0, + "step": 9697 + }, + { + "epoch": 1.233685281770767, + "grad_norm": 2.4204928874969482, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9005627036094666, + "num_tokens": 376456232.0, + "step": 9698 + }, + { + "epoch": 1.2338124920493576, + "grad_norm": 2.7838480472564697, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8856339454650879, + "num_tokens": 376493547.0, + "step": 9699 + }, + { + "epoch": 1.2339397023279481, + "grad_norm": 2.787221670150757, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8772846460342407, + "num_tokens": 376535079.0, + "step": 9700 + }, + { + "epoch": 1.2340669126065387, + "grad_norm": 2.6328213214874268, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.881606936454773, + "num_tokens": 376577716.0, + "step": 9701 + }, + { + "epoch": 1.2341941228851292, + "grad_norm": 2.582122802734375, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8820382952690125, + "num_tokens": 376623323.0, + "step": 9702 + }, + { + "epoch": 1.2343213331637197, + "grad_norm": 2.707610607147217, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8864610195159912, + "num_tokens": 376661168.0, + "step": 9703 + }, + { + "epoch": 1.2344485434423103, + "grad_norm": 2.753328800201416, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8950960636138916, + "num_tokens": 376696912.0, + "step": 9704 + }, + { + "epoch": 1.2345757537209006, + "grad_norm": 2.5629613399505615, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8786299228668213, + "num_tokens": 376739887.0, + "step": 9705 + }, + { + "epoch": 1.234702963999491, + "grad_norm": 2.59028959274292, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8841128349304199, + "num_tokens": 376781652.0, + "step": 9706 + }, + { + "epoch": 1.2348301742780816, + "grad_norm": 2.444011688232422, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8961160778999329, + "num_tokens": 376823125.0, + "step": 9707 + }, + { + "epoch": 1.2349573845566721, + "grad_norm": 2.7932558059692383, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8893017768859863, + "num_tokens": 376857009.0, + "step": 9708 + }, + { + "epoch": 1.2350845948352627, + "grad_norm": 2.483560800552368, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8977950215339661, + "num_tokens": 376898361.0, + "step": 9709 + }, + { + "epoch": 1.2352118051138532, + "grad_norm": 2.650947332382202, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.89396071434021, + "num_tokens": 376933621.0, + "step": 9710 + }, + { + "epoch": 1.2353390153924437, + "grad_norm": 2.527667999267578, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.894718587398529, + "num_tokens": 376975908.0, + "step": 9711 + }, + { + "epoch": 1.2354662256710343, + "grad_norm": 2.694039821624756, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8880094289779663, + "num_tokens": 377017488.0, + "step": 9712 + }, + { + "epoch": 1.2355934359496248, + "grad_norm": 2.609835624694824, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8845299482345581, + "num_tokens": 377062027.0, + "step": 9713 + }, + { + "epoch": 1.2357206462282153, + "grad_norm": 2.726924419403076, + "learning_rate": 1e-06, + "loss": 0.3418, + "mean_token_accuracy": 0.8785678148269653, + "num_tokens": 377104479.0, + "step": 9714 + }, + { + "epoch": 1.2358478565068058, + "grad_norm": 2.7255215644836426, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8921076059341431, + "num_tokens": 377141794.0, + "step": 9715 + }, + { + "epoch": 1.2359750667853961, + "grad_norm": 2.5700767040252686, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8865169286727905, + "num_tokens": 377183724.0, + "step": 9716 + }, + { + "epoch": 1.2361022770639867, + "grad_norm": 2.621457099914551, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8950801491737366, + "num_tokens": 377224245.0, + "step": 9717 + }, + { + "epoch": 1.2362294873425772, + "grad_norm": 2.6630313396453857, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8891282081604004, + "num_tokens": 377262843.0, + "step": 9718 + }, + { + "epoch": 1.2363566976211677, + "grad_norm": 2.617908000946045, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8928476572036743, + "num_tokens": 377304012.0, + "step": 9719 + }, + { + "epoch": 1.2364839078997583, + "grad_norm": 2.8073372840881348, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8883869051933289, + "num_tokens": 377339902.0, + "step": 9720 + }, + { + "epoch": 1.2366111181783488, + "grad_norm": 2.824212074279785, + "learning_rate": 1e-06, + "loss": 0.3446, + "mean_token_accuracy": 0.8804473280906677, + "num_tokens": 377378092.0, + "step": 9721 + }, + { + "epoch": 1.2367383284569393, + "grad_norm": 3.046018600463867, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8838000893592834, + "num_tokens": 377414298.0, + "step": 9722 + }, + { + "epoch": 1.2368655387355298, + "grad_norm": 2.7764179706573486, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8834511637687683, + "num_tokens": 377453077.0, + "step": 9723 + }, + { + "epoch": 1.2369927490141204, + "grad_norm": 2.5591487884521484, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.891960859298706, + "num_tokens": 377493114.0, + "step": 9724 + }, + { + "epoch": 1.237119959292711, + "grad_norm": 2.6549417972564697, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8866028785705566, + "num_tokens": 377533519.0, + "step": 9725 + }, + { + "epoch": 1.2372471695713014, + "grad_norm": 2.464672327041626, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8887530565261841, + "num_tokens": 377573831.0, + "step": 9726 + }, + { + "epoch": 1.237374379849892, + "grad_norm": 2.6645894050598145, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8917130827903748, + "num_tokens": 377613022.0, + "step": 9727 + }, + { + "epoch": 1.2375015901284825, + "grad_norm": 2.883509635925293, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8855921626091003, + "num_tokens": 377647348.0, + "step": 9728 + }, + { + "epoch": 1.237628800407073, + "grad_norm": 3.0401759147644043, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9002705812454224, + "num_tokens": 377686921.0, + "step": 9729 + }, + { + "epoch": 1.2377560106856633, + "grad_norm": 2.746464967727661, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8784937858581543, + "num_tokens": 377727462.0, + "step": 9730 + }, + { + "epoch": 1.2378832209642538, + "grad_norm": 2.7195653915405273, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8968042135238647, + "num_tokens": 377760821.0, + "step": 9731 + }, + { + "epoch": 1.2380104312428444, + "grad_norm": 2.499912738800049, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.887779712677002, + "num_tokens": 377806935.0, + "step": 9732 + }, + { + "epoch": 1.238137641521435, + "grad_norm": 2.6836440563201904, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8993146419525146, + "num_tokens": 377842790.0, + "step": 9733 + }, + { + "epoch": 1.2382648518000254, + "grad_norm": 2.7380802631378174, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8847715854644775, + "num_tokens": 377882781.0, + "step": 9734 + }, + { + "epoch": 1.238392062078616, + "grad_norm": 2.5907225608825684, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8920220136642456, + "num_tokens": 377922291.0, + "step": 9735 + }, + { + "epoch": 1.2385192723572065, + "grad_norm": 2.4641716480255127, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8906962871551514, + "num_tokens": 377967811.0, + "step": 9736 + }, + { + "epoch": 1.238646482635797, + "grad_norm": 2.5752882957458496, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8885796070098877, + "num_tokens": 378006230.0, + "step": 9737 + }, + { + "epoch": 1.2387736929143875, + "grad_norm": 2.620537281036377, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8851765394210815, + "num_tokens": 378047892.0, + "step": 9738 + }, + { + "epoch": 1.238900903192978, + "grad_norm": 2.916957139968872, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8825766444206238, + "num_tokens": 378084435.0, + "step": 9739 + }, + { + "epoch": 1.2390281134715684, + "grad_norm": 2.600886106491089, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8931649327278137, + "num_tokens": 378122774.0, + "step": 9740 + }, + { + "epoch": 1.239155323750159, + "grad_norm": 2.618589162826538, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.900587797164917, + "num_tokens": 378161128.0, + "step": 9741 + }, + { + "epoch": 1.2392825340287494, + "grad_norm": 2.4252431392669678, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8980939388275146, + "num_tokens": 378203978.0, + "step": 9742 + }, + { + "epoch": 1.23940974430734, + "grad_norm": 2.611541509628296, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8909484148025513, + "num_tokens": 378247618.0, + "step": 9743 + }, + { + "epoch": 1.2395369545859305, + "grad_norm": 2.7340354919433594, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8863167762756348, + "num_tokens": 378289826.0, + "step": 9744 + }, + { + "epoch": 1.239664164864521, + "grad_norm": 2.6258294582366943, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8855578899383545, + "num_tokens": 378327716.0, + "step": 9745 + }, + { + "epoch": 1.2397913751431116, + "grad_norm": 2.66152024269104, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9000583291053772, + "num_tokens": 378364214.0, + "step": 9746 + }, + { + "epoch": 1.239918585421702, + "grad_norm": 2.613584518432617, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.892774224281311, + "num_tokens": 378403776.0, + "step": 9747 + }, + { + "epoch": 1.2400457957002926, + "grad_norm": 2.518441677093506, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8926987051963806, + "num_tokens": 378446801.0, + "step": 9748 + }, + { + "epoch": 1.2401730059788831, + "grad_norm": 2.6744589805603027, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8875380754470825, + "num_tokens": 378487976.0, + "step": 9749 + }, + { + "epoch": 1.2403002162574737, + "grad_norm": 2.7676479816436768, + "learning_rate": 1e-06, + "loss": 0.3482, + "mean_token_accuracy": 0.8743752241134644, + "num_tokens": 378526410.0, + "step": 9750 + }, + { + "epoch": 1.2404274265360642, + "grad_norm": 3.0223755836486816, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8698013424873352, + "num_tokens": 378563100.0, + "step": 9751 + }, + { + "epoch": 1.2405546368146547, + "grad_norm": 2.685992479324341, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8995599746704102, + "num_tokens": 378598876.0, + "step": 9752 + }, + { + "epoch": 1.2406818470932452, + "grad_norm": 2.657061815261841, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9025189876556396, + "num_tokens": 378633111.0, + "step": 9753 + }, + { + "epoch": 1.2408090573718356, + "grad_norm": 2.488738536834717, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8892347812652588, + "num_tokens": 378675954.0, + "step": 9754 + }, + { + "epoch": 1.240936267650426, + "grad_norm": 2.7552874088287354, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8784627914428711, + "num_tokens": 378715049.0, + "step": 9755 + }, + { + "epoch": 1.2410634779290166, + "grad_norm": 2.8043384552001953, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8824386596679688, + "num_tokens": 378751198.0, + "step": 9756 + }, + { + "epoch": 1.2411906882076071, + "grad_norm": 2.5995917320251465, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8905060291290283, + "num_tokens": 378792096.0, + "step": 9757 + }, + { + "epoch": 1.2413178984861977, + "grad_norm": 2.6048896312713623, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8918383717536926, + "num_tokens": 378831743.0, + "step": 9758 + }, + { + "epoch": 1.2414451087647882, + "grad_norm": 2.4868321418762207, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8974404335021973, + "num_tokens": 378872636.0, + "step": 9759 + }, + { + "epoch": 1.2415723190433787, + "grad_norm": 2.809037923812866, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8803525567054749, + "num_tokens": 378911449.0, + "step": 9760 + }, + { + "epoch": 1.2416995293219693, + "grad_norm": 2.765500783920288, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8882669806480408, + "num_tokens": 378946114.0, + "step": 9761 + }, + { + "epoch": 1.2418267396005598, + "grad_norm": 2.5116517543792725, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.900504469871521, + "num_tokens": 378984025.0, + "step": 9762 + }, + { + "epoch": 1.2419539498791503, + "grad_norm": 2.4816038608551025, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.903313398361206, + "num_tokens": 379020721.0, + "step": 9763 + }, + { + "epoch": 1.2420811601577408, + "grad_norm": 2.6059043407440186, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8834632039070129, + "num_tokens": 379061418.0, + "step": 9764 + }, + { + "epoch": 1.2422083704363311, + "grad_norm": 2.7900404930114746, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8848389983177185, + "num_tokens": 379095572.0, + "step": 9765 + }, + { + "epoch": 1.2423355807149217, + "grad_norm": 2.764285087585449, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8919435739517212, + "num_tokens": 379130986.0, + "step": 9766 + }, + { + "epoch": 1.2424627909935122, + "grad_norm": 2.667470932006836, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8879295587539673, + "num_tokens": 379174042.0, + "step": 9767 + }, + { + "epoch": 1.2425900012721027, + "grad_norm": 2.3798093795776367, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8955994844436646, + "num_tokens": 379218323.0, + "step": 9768 + }, + { + "epoch": 1.2427172115506933, + "grad_norm": 2.749722957611084, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8793882131576538, + "num_tokens": 379258480.0, + "step": 9769 + }, + { + "epoch": 1.2428444218292838, + "grad_norm": 2.814558267593384, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8881922960281372, + "num_tokens": 379297457.0, + "step": 9770 + }, + { + "epoch": 1.2429716321078743, + "grad_norm": 2.331397771835327, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8943824768066406, + "num_tokens": 379342213.0, + "step": 9771 + }, + { + "epoch": 1.2430988423864648, + "grad_norm": 2.647266387939453, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8982117772102356, + "num_tokens": 379378282.0, + "step": 9772 + }, + { + "epoch": 1.2432260526650554, + "grad_norm": 2.737785816192627, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8932861089706421, + "num_tokens": 379415105.0, + "step": 9773 + }, + { + "epoch": 1.243353262943646, + "grad_norm": 2.6850762367248535, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9075905680656433, + "num_tokens": 379447926.0, + "step": 9774 + }, + { + "epoch": 1.2434804732222364, + "grad_norm": 2.8139278888702393, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.880058765411377, + "num_tokens": 379483893.0, + "step": 9775 + }, + { + "epoch": 1.243607683500827, + "grad_norm": 2.600292921066284, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8856790065765381, + "num_tokens": 379530037.0, + "step": 9776 + }, + { + "epoch": 1.2437348937794175, + "grad_norm": 2.9590904712677, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8939785957336426, + "num_tokens": 379561096.0, + "step": 9777 + }, + { + "epoch": 1.243862104058008, + "grad_norm": 2.5363426208496094, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8962967991828918, + "num_tokens": 379602907.0, + "step": 9778 + }, + { + "epoch": 1.2439893143365983, + "grad_norm": 2.573157548904419, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8906663656234741, + "num_tokens": 379643476.0, + "step": 9779 + }, + { + "epoch": 1.2441165246151888, + "grad_norm": 2.471442222595215, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8915215730667114, + "num_tokens": 379688637.0, + "step": 9780 + }, + { + "epoch": 1.2442437348937794, + "grad_norm": 2.621293306350708, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8892059326171875, + "num_tokens": 379731940.0, + "step": 9781 + }, + { + "epoch": 1.24437094517237, + "grad_norm": 2.4925222396850586, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8862643241882324, + "num_tokens": 379776576.0, + "step": 9782 + }, + { + "epoch": 1.2444981554509604, + "grad_norm": 2.6123063564300537, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9017624258995056, + "num_tokens": 379813816.0, + "step": 9783 + }, + { + "epoch": 1.244625365729551, + "grad_norm": 2.591811418533325, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8975222110748291, + "num_tokens": 379849820.0, + "step": 9784 + }, + { + "epoch": 1.2447525760081415, + "grad_norm": 2.6311864852905273, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8786906599998474, + "num_tokens": 379894013.0, + "step": 9785 + }, + { + "epoch": 1.244879786286732, + "grad_norm": 2.8372490406036377, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8871821165084839, + "num_tokens": 379928510.0, + "step": 9786 + }, + { + "epoch": 1.2450069965653225, + "grad_norm": 2.4659554958343506, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.891165018081665, + "num_tokens": 379971678.0, + "step": 9787 + }, + { + "epoch": 1.245134206843913, + "grad_norm": 2.3720171451568604, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9031152725219727, + "num_tokens": 380014578.0, + "step": 9788 + }, + { + "epoch": 1.2452614171225034, + "grad_norm": 2.8231875896453857, + "learning_rate": 1e-06, + "loss": 0.3477, + "mean_token_accuracy": 0.8786908388137817, + "num_tokens": 380054023.0, + "step": 9789 + }, + { + "epoch": 1.245388627401094, + "grad_norm": 2.661127805709839, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8958534002304077, + "num_tokens": 380090364.0, + "step": 9790 + }, + { + "epoch": 1.2455158376796844, + "grad_norm": 2.5437989234924316, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9086720943450928, + "num_tokens": 380127497.0, + "step": 9791 + }, + { + "epoch": 1.245643047958275, + "grad_norm": 2.8873629570007324, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8851888179779053, + "num_tokens": 380160660.0, + "step": 9792 + }, + { + "epoch": 1.2457702582368655, + "grad_norm": 2.72538685798645, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.885125994682312, + "num_tokens": 380200316.0, + "step": 9793 + }, + { + "epoch": 1.245897468515456, + "grad_norm": 2.746605157852173, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8946784138679504, + "num_tokens": 380238798.0, + "step": 9794 + }, + { + "epoch": 1.2460246787940465, + "grad_norm": 2.3714828491210938, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8968357443809509, + "num_tokens": 380281132.0, + "step": 9795 + }, + { + "epoch": 1.246151889072637, + "grad_norm": 2.942026138305664, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8846005797386169, + "num_tokens": 380314471.0, + "step": 9796 + }, + { + "epoch": 1.2462790993512276, + "grad_norm": 2.5733847618103027, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8913201689720154, + "num_tokens": 380357504.0, + "step": 9797 + }, + { + "epoch": 1.2464063096298181, + "grad_norm": 2.5076770782470703, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8840565085411072, + "num_tokens": 380397887.0, + "step": 9798 + }, + { + "epoch": 1.2465335199084087, + "grad_norm": 2.722450017929077, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8820112943649292, + "num_tokens": 380437558.0, + "step": 9799 + }, + { + "epoch": 1.2466607301869992, + "grad_norm": 2.56771183013916, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8945255875587463, + "num_tokens": 380480063.0, + "step": 9800 + }, + { + "epoch": 1.2467879404655897, + "grad_norm": 2.558868646621704, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8860467076301575, + "num_tokens": 380523467.0, + "step": 9801 + }, + { + "epoch": 1.2469151507441802, + "grad_norm": 3.233567714691162, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8876381516456604, + "num_tokens": 380556774.0, + "step": 9802 + }, + { + "epoch": 1.2470423610227706, + "grad_norm": 2.48274302482605, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9000616669654846, + "num_tokens": 380596834.0, + "step": 9803 + }, + { + "epoch": 1.247169571301361, + "grad_norm": 2.7872660160064697, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8755742311477661, + "num_tokens": 380642520.0, + "step": 9804 + }, + { + "epoch": 1.2472967815799516, + "grad_norm": 2.6117007732391357, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8835899829864502, + "num_tokens": 380687524.0, + "step": 9805 + }, + { + "epoch": 1.2474239918585421, + "grad_norm": 2.4169106483459473, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8901224136352539, + "num_tokens": 380731027.0, + "step": 9806 + }, + { + "epoch": 1.2475512021371327, + "grad_norm": 2.8421459197998047, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8799732327461243, + "num_tokens": 380768164.0, + "step": 9807 + }, + { + "epoch": 1.2476784124157232, + "grad_norm": 2.4655776023864746, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8953001499176025, + "num_tokens": 380812236.0, + "step": 9808 + }, + { + "epoch": 1.2478056226943137, + "grad_norm": 2.6778810024261475, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8869153261184692, + "num_tokens": 380851586.0, + "step": 9809 + }, + { + "epoch": 1.2479328329729042, + "grad_norm": 3.0018205642700195, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8838342428207397, + "num_tokens": 380884830.0, + "step": 9810 + }, + { + "epoch": 1.2480600432514948, + "grad_norm": 2.765343427658081, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8782424926757812, + "num_tokens": 380925312.0, + "step": 9811 + }, + { + "epoch": 1.2481872535300853, + "grad_norm": 2.7424848079681396, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8887397646903992, + "num_tokens": 380963048.0, + "step": 9812 + }, + { + "epoch": 1.2483144638086758, + "grad_norm": 2.9004480838775635, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8786103129386902, + "num_tokens": 381000911.0, + "step": 9813 + }, + { + "epoch": 1.2484416740872661, + "grad_norm": 2.647090435028076, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.895963191986084, + "num_tokens": 381038572.0, + "step": 9814 + }, + { + "epoch": 1.2485688843658567, + "grad_norm": 2.4943151473999023, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8950560092926025, + "num_tokens": 381079677.0, + "step": 9815 + }, + { + "epoch": 1.2486960946444472, + "grad_norm": 2.8681459426879883, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8818951845169067, + "num_tokens": 381115066.0, + "step": 9816 + }, + { + "epoch": 1.2488233049230377, + "grad_norm": 2.695221424102783, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.883431613445282, + "num_tokens": 381156202.0, + "step": 9817 + }, + { + "epoch": 1.2489505152016283, + "grad_norm": 2.5491626262664795, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8942927122116089, + "num_tokens": 381196165.0, + "step": 9818 + }, + { + "epoch": 1.2490777254802188, + "grad_norm": 2.6825931072235107, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9007124900817871, + "num_tokens": 381231659.0, + "step": 9819 + }, + { + "epoch": 1.2492049357588093, + "grad_norm": 2.8257358074188232, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8896319270133972, + "num_tokens": 381267131.0, + "step": 9820 + }, + { + "epoch": 1.2493321460373998, + "grad_norm": 2.7836804389953613, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8958793878555298, + "num_tokens": 381304309.0, + "step": 9821 + }, + { + "epoch": 1.2494593563159904, + "grad_norm": 2.75053334236145, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8895038366317749, + "num_tokens": 381343048.0, + "step": 9822 + }, + { + "epoch": 1.249586566594581, + "grad_norm": 2.6500022411346436, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8922889232635498, + "num_tokens": 381380503.0, + "step": 9823 + }, + { + "epoch": 1.2497137768731714, + "grad_norm": 2.5780131816864014, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8921840786933899, + "num_tokens": 381418930.0, + "step": 9824 + }, + { + "epoch": 1.249840987151762, + "grad_norm": 2.895707368850708, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8772543668746948, + "num_tokens": 381455061.0, + "step": 9825 + }, + { + "epoch": 1.2499681974303525, + "grad_norm": 3.0756731033325195, + "learning_rate": 1e-06, + "loss": 0.3695, + "mean_token_accuracy": 0.8724875450134277, + "num_tokens": 381490899.0, + "step": 9826 + }, + { + "epoch": 1.250095407708943, + "grad_norm": 2.687458038330078, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8977435231208801, + "num_tokens": 381531647.0, + "step": 9827 + }, + { + "epoch": 1.2502226179875333, + "grad_norm": 3.066042184829712, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8791990876197815, + "num_tokens": 381566762.0, + "step": 9828 + }, + { + "epoch": 1.2503498282661238, + "grad_norm": 2.722357988357544, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8866454362869263, + "num_tokens": 381606421.0, + "step": 9829 + }, + { + "epoch": 1.2504770385447144, + "grad_norm": 2.812626600265503, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.895424485206604, + "num_tokens": 381642632.0, + "step": 9830 + }, + { + "epoch": 1.250604248823305, + "grad_norm": 2.668245553970337, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8994906544685364, + "num_tokens": 381678726.0, + "step": 9831 + }, + { + "epoch": 1.2507314591018954, + "grad_norm": 2.8614630699157715, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8842169046401978, + "num_tokens": 381715332.0, + "step": 9832 + }, + { + "epoch": 1.250858669380486, + "grad_norm": 2.5385828018188477, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8890769481658936, + "num_tokens": 381756453.0, + "step": 9833 + }, + { + "epoch": 1.2509858796590765, + "grad_norm": 2.6927483081817627, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8820959329605103, + "num_tokens": 381794663.0, + "step": 9834 + }, + { + "epoch": 1.251113089937667, + "grad_norm": 2.967625856399536, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.878991961479187, + "num_tokens": 381831210.0, + "step": 9835 + }, + { + "epoch": 1.2512403002162575, + "grad_norm": 2.7785797119140625, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8798953890800476, + "num_tokens": 381870492.0, + "step": 9836 + }, + { + "epoch": 1.2513675104948478, + "grad_norm": 2.57978892326355, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8922251462936401, + "num_tokens": 381911889.0, + "step": 9837 + }, + { + "epoch": 1.2514947207734384, + "grad_norm": 2.7098922729492188, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8724067807197571, + "num_tokens": 381950990.0, + "step": 9838 + }, + { + "epoch": 1.251621931052029, + "grad_norm": 2.8254106044769287, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8777209520339966, + "num_tokens": 381988655.0, + "step": 9839 + }, + { + "epoch": 1.2517491413306194, + "grad_norm": 2.7935445308685303, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.9004180431365967, + "num_tokens": 382021349.0, + "step": 9840 + }, + { + "epoch": 1.25187635160921, + "grad_norm": 2.5608153343200684, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8940645456314087, + "num_tokens": 382061569.0, + "step": 9841 + }, + { + "epoch": 1.2520035618878005, + "grad_norm": 2.684112071990967, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.884813666343689, + "num_tokens": 382105034.0, + "step": 9842 + }, + { + "epoch": 1.252130772166391, + "grad_norm": 2.7240421772003174, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8867255449295044, + "num_tokens": 382142904.0, + "step": 9843 + }, + { + "epoch": 1.2522579824449815, + "grad_norm": 2.453040599822998, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8927668333053589, + "num_tokens": 382184406.0, + "step": 9844 + }, + { + "epoch": 1.252385192723572, + "grad_norm": 2.629781723022461, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8917772769927979, + "num_tokens": 382222858.0, + "step": 9845 + }, + { + "epoch": 1.2525124030021626, + "grad_norm": 2.5537102222442627, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8827823400497437, + "num_tokens": 382266777.0, + "step": 9846 + }, + { + "epoch": 1.2526396132807531, + "grad_norm": 2.841813325881958, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8927935361862183, + "num_tokens": 382304350.0, + "step": 9847 + }, + { + "epoch": 1.2527668235593437, + "grad_norm": 2.8147757053375244, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8848047852516174, + "num_tokens": 382340161.0, + "step": 9848 + }, + { + "epoch": 1.2528940338379342, + "grad_norm": 2.834845781326294, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8842135071754456, + "num_tokens": 382377498.0, + "step": 9849 + }, + { + "epoch": 1.2530212441165247, + "grad_norm": 2.8592000007629395, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8981156945228577, + "num_tokens": 382410483.0, + "step": 9850 + }, + { + "epoch": 1.2531484543951152, + "grad_norm": 2.720587730407715, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8863554000854492, + "num_tokens": 382449310.0, + "step": 9851 + }, + { + "epoch": 1.2532756646737058, + "grad_norm": 2.7479965686798096, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8834719061851501, + "num_tokens": 382494167.0, + "step": 9852 + }, + { + "epoch": 1.253402874952296, + "grad_norm": 2.5545451641082764, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8951404094696045, + "num_tokens": 382531444.0, + "step": 9853 + }, + { + "epoch": 1.2535300852308866, + "grad_norm": 2.785475254058838, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8810926079750061, + "num_tokens": 382569723.0, + "step": 9854 + }, + { + "epoch": 1.2536572955094771, + "grad_norm": 2.5699093341827393, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8915429711341858, + "num_tokens": 382612566.0, + "step": 9855 + }, + { + "epoch": 1.2537845057880677, + "grad_norm": 2.5802392959594727, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8961559534072876, + "num_tokens": 382653355.0, + "step": 9856 + }, + { + "epoch": 1.2539117160666582, + "grad_norm": 2.5436854362487793, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8956618905067444, + "num_tokens": 382690161.0, + "step": 9857 + }, + { + "epoch": 1.2540389263452487, + "grad_norm": 2.7789559364318848, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9023944139480591, + "num_tokens": 382722122.0, + "step": 9858 + }, + { + "epoch": 1.2541661366238392, + "grad_norm": 2.753735065460205, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8857735395431519, + "num_tokens": 382759887.0, + "step": 9859 + }, + { + "epoch": 1.2542933469024298, + "grad_norm": 2.765235185623169, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8939890265464783, + "num_tokens": 382795535.0, + "step": 9860 + }, + { + "epoch": 1.2544205571810203, + "grad_norm": 2.588820457458496, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8859364986419678, + "num_tokens": 382839872.0, + "step": 9861 + }, + { + "epoch": 1.2545477674596106, + "grad_norm": 2.539734125137329, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8860994577407837, + "num_tokens": 382887883.0, + "step": 9862 + }, + { + "epoch": 1.2546749777382011, + "grad_norm": 2.6631717681884766, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.895521879196167, + "num_tokens": 382923367.0, + "step": 9863 + }, + { + "epoch": 1.2548021880167917, + "grad_norm": 2.630375623703003, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9014126062393188, + "num_tokens": 382959087.0, + "step": 9864 + }, + { + "epoch": 1.2549293982953822, + "grad_norm": 2.4481537342071533, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9050799608230591, + "num_tokens": 383005239.0, + "step": 9865 + }, + { + "epoch": 1.2550566085739727, + "grad_norm": 2.719102144241333, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.895591139793396, + "num_tokens": 383044855.0, + "step": 9866 + }, + { + "epoch": 1.2551838188525632, + "grad_norm": 2.7638981342315674, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8853724002838135, + "num_tokens": 383084060.0, + "step": 9867 + }, + { + "epoch": 1.2553110291311538, + "grad_norm": 2.7660388946533203, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8847702741622925, + "num_tokens": 383123016.0, + "step": 9868 + }, + { + "epoch": 1.2554382394097443, + "grad_norm": 2.8518919944763184, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.879051685333252, + "num_tokens": 383161699.0, + "step": 9869 + }, + { + "epoch": 1.2555654496883348, + "grad_norm": 2.538761615753174, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8937407732009888, + "num_tokens": 383204197.0, + "step": 9870 + }, + { + "epoch": 1.2556926599669254, + "grad_norm": 2.8582096099853516, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8907003998756409, + "num_tokens": 383240907.0, + "step": 9871 + }, + { + "epoch": 1.255819870245516, + "grad_norm": 2.671370506286621, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8871569633483887, + "num_tokens": 383276384.0, + "step": 9872 + }, + { + "epoch": 1.2559470805241064, + "grad_norm": 2.7727813720703125, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9063866138458252, + "num_tokens": 383309589.0, + "step": 9873 + }, + { + "epoch": 1.256074290802697, + "grad_norm": 2.7968955039978027, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8960804343223572, + "num_tokens": 383346061.0, + "step": 9874 + }, + { + "epoch": 1.2562015010812875, + "grad_norm": 2.6109564304351807, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8860257863998413, + "num_tokens": 383385876.0, + "step": 9875 + }, + { + "epoch": 1.256328711359878, + "grad_norm": 2.5149452686309814, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8948100805282593, + "num_tokens": 383426567.0, + "step": 9876 + }, + { + "epoch": 1.2564559216384683, + "grad_norm": 2.860992193222046, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8897563219070435, + "num_tokens": 383465691.0, + "step": 9877 + }, + { + "epoch": 1.2565831319170588, + "grad_norm": 2.8127028942108154, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.899997889995575, + "num_tokens": 383499538.0, + "step": 9878 + }, + { + "epoch": 1.2567103421956494, + "grad_norm": 2.6226606369018555, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8927414417266846, + "num_tokens": 383536350.0, + "step": 9879 + }, + { + "epoch": 1.25683755247424, + "grad_norm": 2.8633058071136475, + "learning_rate": 1e-06, + "loss": 0.3585, + "mean_token_accuracy": 0.8696528673171997, + "num_tokens": 383578865.0, + "step": 9880 + }, + { + "epoch": 1.2569647627528304, + "grad_norm": 2.7545948028564453, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8898221850395203, + "num_tokens": 383621187.0, + "step": 9881 + }, + { + "epoch": 1.257091973031421, + "grad_norm": 2.706057548522949, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8955928087234497, + "num_tokens": 383658192.0, + "step": 9882 + }, + { + "epoch": 1.2572191833100115, + "grad_norm": 2.5550668239593506, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8921164274215698, + "num_tokens": 383696999.0, + "step": 9883 + }, + { + "epoch": 1.257346393588602, + "grad_norm": 2.427258014678955, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.894756555557251, + "num_tokens": 383743454.0, + "step": 9884 + }, + { + "epoch": 1.2574736038671925, + "grad_norm": 2.6909091472625732, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8977547883987427, + "num_tokens": 383779044.0, + "step": 9885 + }, + { + "epoch": 1.2576008141457828, + "grad_norm": 2.7828896045684814, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.895986020565033, + "num_tokens": 383817512.0, + "step": 9886 + }, + { + "epoch": 1.2577280244243734, + "grad_norm": 2.5700740814208984, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8911021947860718, + "num_tokens": 383858005.0, + "step": 9887 + }, + { + "epoch": 1.257855234702964, + "grad_norm": 2.6980855464935303, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8844778537750244, + "num_tokens": 383900993.0, + "step": 9888 + }, + { + "epoch": 1.2579824449815544, + "grad_norm": 2.516049861907959, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8884085416793823, + "num_tokens": 383942390.0, + "step": 9889 + }, + { + "epoch": 1.258109655260145, + "grad_norm": 2.4821577072143555, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8939053416252136, + "num_tokens": 383986807.0, + "step": 9890 + }, + { + "epoch": 1.2582368655387355, + "grad_norm": 2.3667867183685303, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8901059627532959, + "num_tokens": 384034210.0, + "step": 9891 + }, + { + "epoch": 1.258364075817326, + "grad_norm": 2.5361080169677734, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8948812484741211, + "num_tokens": 384074060.0, + "step": 9892 + }, + { + "epoch": 1.2584912860959165, + "grad_norm": 2.7376749515533447, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9004802703857422, + "num_tokens": 384107184.0, + "step": 9893 + }, + { + "epoch": 1.258618496374507, + "grad_norm": 2.385660171508789, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8849313259124756, + "num_tokens": 384153456.0, + "step": 9894 + }, + { + "epoch": 1.2587457066530976, + "grad_norm": 2.7008860111236572, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8853768706321716, + "num_tokens": 384192347.0, + "step": 9895 + }, + { + "epoch": 1.2588729169316881, + "grad_norm": 2.458930730819702, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8964266777038574, + "num_tokens": 384236671.0, + "step": 9896 + }, + { + "epoch": 1.2590001272102787, + "grad_norm": 2.679898738861084, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.889498233795166, + "num_tokens": 384276350.0, + "step": 9897 + }, + { + "epoch": 1.2591273374888692, + "grad_norm": 2.5942270755767822, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8822442293167114, + "num_tokens": 384318288.0, + "step": 9898 + }, + { + "epoch": 1.2592545477674597, + "grad_norm": 2.768521308898926, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.899084210395813, + "num_tokens": 384355347.0, + "step": 9899 + }, + { + "epoch": 1.2593817580460502, + "grad_norm": 2.7965662479400635, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8716951012611389, + "num_tokens": 384395538.0, + "step": 9900 + }, + { + "epoch": 1.2595089683246408, + "grad_norm": 2.715463399887085, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8856318593025208, + "num_tokens": 384435692.0, + "step": 9901 + }, + { + "epoch": 1.259636178603231, + "grad_norm": 2.518409013748169, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8916306495666504, + "num_tokens": 384480260.0, + "step": 9902 + }, + { + "epoch": 1.2597633888818216, + "grad_norm": 2.733574151992798, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8982235789299011, + "num_tokens": 384514494.0, + "step": 9903 + }, + { + "epoch": 1.2598905991604121, + "grad_norm": 3.3203723430633545, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.8990703821182251, + "num_tokens": 384548280.0, + "step": 9904 + }, + { + "epoch": 1.2600178094390027, + "grad_norm": 2.489860773086548, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8956747055053711, + "num_tokens": 384586953.0, + "step": 9905 + }, + { + "epoch": 1.2601450197175932, + "grad_norm": 2.629645347595215, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8928035497665405, + "num_tokens": 384631289.0, + "step": 9906 + }, + { + "epoch": 1.2602722299961837, + "grad_norm": 2.716815710067749, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8907612562179565, + "num_tokens": 384669979.0, + "step": 9907 + }, + { + "epoch": 1.2603994402747742, + "grad_norm": 2.7823572158813477, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8832824230194092, + "num_tokens": 384709287.0, + "step": 9908 + }, + { + "epoch": 1.2605266505533648, + "grad_norm": 2.717585325241089, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8895844221115112, + "num_tokens": 384750180.0, + "step": 9909 + }, + { + "epoch": 1.2606538608319553, + "grad_norm": 2.6940720081329346, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8900793790817261, + "num_tokens": 384786568.0, + "step": 9910 + }, + { + "epoch": 1.2607810711105456, + "grad_norm": 2.752685308456421, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8926974534988403, + "num_tokens": 384820890.0, + "step": 9911 + }, + { + "epoch": 1.2609082813891361, + "grad_norm": 2.8279941082000732, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8782505989074707, + "num_tokens": 384856510.0, + "step": 9912 + }, + { + "epoch": 1.2610354916677267, + "grad_norm": 2.9392616748809814, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8856967687606812, + "num_tokens": 384890631.0, + "step": 9913 + }, + { + "epoch": 1.2611627019463172, + "grad_norm": 2.7337334156036377, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8932220935821533, + "num_tokens": 384926540.0, + "step": 9914 + }, + { + "epoch": 1.2612899122249077, + "grad_norm": 2.638516902923584, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.892475962638855, + "num_tokens": 384965608.0, + "step": 9915 + }, + { + "epoch": 1.2614171225034982, + "grad_norm": 2.5666344165802, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9048277139663696, + "num_tokens": 385004233.0, + "step": 9916 + }, + { + "epoch": 1.2615443327820888, + "grad_norm": 2.793205738067627, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.884257972240448, + "num_tokens": 385042101.0, + "step": 9917 + }, + { + "epoch": 1.2616715430606793, + "grad_norm": 2.5706300735473633, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8988087177276611, + "num_tokens": 385084204.0, + "step": 9918 + }, + { + "epoch": 1.2617987533392698, + "grad_norm": 2.5373919010162354, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8982320427894592, + "num_tokens": 385125934.0, + "step": 9919 + }, + { + "epoch": 1.2619259636178604, + "grad_norm": 2.6243338584899902, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8934626579284668, + "num_tokens": 385165160.0, + "step": 9920 + }, + { + "epoch": 1.2620531738964509, + "grad_norm": 2.5855870246887207, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8856070637702942, + "num_tokens": 385207090.0, + "step": 9921 + }, + { + "epoch": 1.2621803841750414, + "grad_norm": 2.722498893737793, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.8986026644706726, + "num_tokens": 385240420.0, + "step": 9922 + }, + { + "epoch": 1.262307594453632, + "grad_norm": 2.7321958541870117, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8825833201408386, + "num_tokens": 385282439.0, + "step": 9923 + }, + { + "epoch": 1.2624348047322225, + "grad_norm": 2.662184953689575, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8901945352554321, + "num_tokens": 385321733.0, + "step": 9924 + }, + { + "epoch": 1.262562015010813, + "grad_norm": 2.5570478439331055, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8810380697250366, + "num_tokens": 385367475.0, + "step": 9925 + }, + { + "epoch": 1.2626892252894033, + "grad_norm": 2.604830265045166, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8915848731994629, + "num_tokens": 385407340.0, + "step": 9926 + }, + { + "epoch": 1.2628164355679938, + "grad_norm": 2.522770643234253, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8890831470489502, + "num_tokens": 385448391.0, + "step": 9927 + }, + { + "epoch": 1.2629436458465844, + "grad_norm": 2.624495029449463, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9011163711547852, + "num_tokens": 385485038.0, + "step": 9928 + }, + { + "epoch": 1.263070856125175, + "grad_norm": 2.986037492752075, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8895187377929688, + "num_tokens": 385517763.0, + "step": 9929 + }, + { + "epoch": 1.2631980664037654, + "grad_norm": 2.6492178440093994, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8952877521514893, + "num_tokens": 385558723.0, + "step": 9930 + }, + { + "epoch": 1.263325276682356, + "grad_norm": 2.964993953704834, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8845432996749878, + "num_tokens": 385595104.0, + "step": 9931 + }, + { + "epoch": 1.2634524869609465, + "grad_norm": 2.7553956508636475, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8816075921058655, + "num_tokens": 385634480.0, + "step": 9932 + }, + { + "epoch": 1.263579697239537, + "grad_norm": 2.593393087387085, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9007338285446167, + "num_tokens": 385672377.0, + "step": 9933 + }, + { + "epoch": 1.2637069075181275, + "grad_norm": 2.5958259105682373, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8984417915344238, + "num_tokens": 385711108.0, + "step": 9934 + }, + { + "epoch": 1.2638341177967178, + "grad_norm": 2.627365827560425, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8832786083221436, + "num_tokens": 385754755.0, + "step": 9935 + }, + { + "epoch": 1.2639613280753084, + "grad_norm": 2.654327869415283, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8994897603988647, + "num_tokens": 385794164.0, + "step": 9936 + }, + { + "epoch": 1.264088538353899, + "grad_norm": 2.756005048751831, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8924355506896973, + "num_tokens": 385830129.0, + "step": 9937 + }, + { + "epoch": 1.2642157486324894, + "grad_norm": 3.102553367614746, + "learning_rate": 1e-06, + "loss": 0.3728, + "mean_token_accuracy": 0.8658991456031799, + "num_tokens": 385866972.0, + "step": 9938 + }, + { + "epoch": 1.26434295891108, + "grad_norm": 2.9061882495880127, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8800337910652161, + "num_tokens": 385905060.0, + "step": 9939 + }, + { + "epoch": 1.2644701691896705, + "grad_norm": 2.6354405879974365, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8880483508110046, + "num_tokens": 385945257.0, + "step": 9940 + }, + { + "epoch": 1.264597379468261, + "grad_norm": 2.7295026779174805, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.907430112361908, + "num_tokens": 385978231.0, + "step": 9941 + }, + { + "epoch": 1.2647245897468515, + "grad_norm": 2.664479970932007, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8887242078781128, + "num_tokens": 386014437.0, + "step": 9942 + }, + { + "epoch": 1.264851800025442, + "grad_norm": 2.7485368251800537, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8937917947769165, + "num_tokens": 386051332.0, + "step": 9943 + }, + { + "epoch": 1.2649790103040326, + "grad_norm": 2.9261064529418945, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8890416622161865, + "num_tokens": 386087475.0, + "step": 9944 + }, + { + "epoch": 1.2651062205826231, + "grad_norm": 2.772225856781006, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8879929780960083, + "num_tokens": 386125214.0, + "step": 9945 + }, + { + "epoch": 1.2652334308612136, + "grad_norm": 2.56673264503479, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8918102979660034, + "num_tokens": 386163343.0, + "step": 9946 + }, + { + "epoch": 1.2653606411398042, + "grad_norm": 2.9685373306274414, + "learning_rate": 1e-06, + "loss": 0.3592, + "mean_token_accuracy": 0.8729547262191772, + "num_tokens": 386198955.0, + "step": 9947 + }, + { + "epoch": 1.2654878514183947, + "grad_norm": 2.6504464149475098, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8873742818832397, + "num_tokens": 386238532.0, + "step": 9948 + }, + { + "epoch": 1.2656150616969852, + "grad_norm": 2.7882308959960938, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8941472768783569, + "num_tokens": 386273193.0, + "step": 9949 + }, + { + "epoch": 1.2657422719755758, + "grad_norm": 2.5869383811950684, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8885558843612671, + "num_tokens": 386314466.0, + "step": 9950 + }, + { + "epoch": 1.265869482254166, + "grad_norm": 2.7380049228668213, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8991479873657227, + "num_tokens": 386349517.0, + "step": 9951 + }, + { + "epoch": 1.2659966925327566, + "grad_norm": 2.76818585395813, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8741551637649536, + "num_tokens": 386391209.0, + "step": 9952 + }, + { + "epoch": 1.2661239028113471, + "grad_norm": 2.873029947280884, + "learning_rate": 1e-06, + "loss": 0.3406, + "mean_token_accuracy": 0.8786717057228088, + "num_tokens": 386426553.0, + "step": 9953 + }, + { + "epoch": 1.2662511130899377, + "grad_norm": 2.7698981761932373, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.886212944984436, + "num_tokens": 386463358.0, + "step": 9954 + }, + { + "epoch": 1.2663783233685282, + "grad_norm": 3.04215931892395, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8780704736709595, + "num_tokens": 386497191.0, + "step": 9955 + }, + { + "epoch": 1.2665055336471187, + "grad_norm": 2.8011624813079834, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.887526273727417, + "num_tokens": 386535166.0, + "step": 9956 + }, + { + "epoch": 1.2666327439257092, + "grad_norm": 2.773636817932129, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8943592309951782, + "num_tokens": 386568394.0, + "step": 9957 + }, + { + "epoch": 1.2667599542042998, + "grad_norm": 2.6328463554382324, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9025305509567261, + "num_tokens": 386604577.0, + "step": 9958 + }, + { + "epoch": 1.2668871644828903, + "grad_norm": 2.4290363788604736, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9000405073165894, + "num_tokens": 386647580.0, + "step": 9959 + }, + { + "epoch": 1.2670143747614806, + "grad_norm": 2.5872738361358643, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8900058269500732, + "num_tokens": 386687502.0, + "step": 9960 + }, + { + "epoch": 1.2671415850400711, + "grad_norm": 2.5277507305145264, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8745023608207703, + "num_tokens": 386731060.0, + "step": 9961 + }, + { + "epoch": 1.2672687953186617, + "grad_norm": 2.99230694770813, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.886020839214325, + "num_tokens": 386762534.0, + "step": 9962 + }, + { + "epoch": 1.2673960055972522, + "grad_norm": 2.6774184703826904, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9002958536148071, + "num_tokens": 386798002.0, + "step": 9963 + }, + { + "epoch": 1.2675232158758427, + "grad_norm": 2.7321243286132812, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8941110968589783, + "num_tokens": 386832573.0, + "step": 9964 + }, + { + "epoch": 1.2676504261544332, + "grad_norm": 2.505336046218872, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.902480959892273, + "num_tokens": 386873681.0, + "step": 9965 + }, + { + "epoch": 1.2677776364330238, + "grad_norm": 2.9665169715881348, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8795821070671082, + "num_tokens": 386906993.0, + "step": 9966 + }, + { + "epoch": 1.2679048467116143, + "grad_norm": 2.746197462081909, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8915421962738037, + "num_tokens": 386943209.0, + "step": 9967 + }, + { + "epoch": 1.2680320569902048, + "grad_norm": 2.7875425815582275, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8888806700706482, + "num_tokens": 386982811.0, + "step": 9968 + }, + { + "epoch": 1.2681592672687954, + "grad_norm": 2.62565541267395, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8789329528808594, + "num_tokens": 387028024.0, + "step": 9969 + }, + { + "epoch": 1.2682864775473859, + "grad_norm": 2.514479398727417, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8983148336410522, + "num_tokens": 387067210.0, + "step": 9970 + }, + { + "epoch": 1.2684136878259764, + "grad_norm": 2.866107702255249, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8856199383735657, + "num_tokens": 387101883.0, + "step": 9971 + }, + { + "epoch": 1.268540898104567, + "grad_norm": 2.777395486831665, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8848487734794617, + "num_tokens": 387135741.0, + "step": 9972 + }, + { + "epoch": 1.2686681083831575, + "grad_norm": 2.90244460105896, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.887862503528595, + "num_tokens": 387170097.0, + "step": 9973 + }, + { + "epoch": 1.268795318661748, + "grad_norm": 2.5330395698547363, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9088504314422607, + "num_tokens": 387205755.0, + "step": 9974 + }, + { + "epoch": 1.2689225289403383, + "grad_norm": 2.7770071029663086, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.889106810092926, + "num_tokens": 387243986.0, + "step": 9975 + }, + { + "epoch": 1.2690497392189288, + "grad_norm": 2.848259210586548, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8899723291397095, + "num_tokens": 387279096.0, + "step": 9976 + }, + { + "epoch": 1.2691769494975194, + "grad_norm": 3.0421342849731445, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8907160758972168, + "num_tokens": 387310331.0, + "step": 9977 + }, + { + "epoch": 1.2693041597761099, + "grad_norm": 2.8603856563568115, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.8734625577926636, + "num_tokens": 387347172.0, + "step": 9978 + }, + { + "epoch": 1.2694313700547004, + "grad_norm": 2.6139867305755615, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8849868178367615, + "num_tokens": 387388137.0, + "step": 9979 + }, + { + "epoch": 1.269558580333291, + "grad_norm": 2.6255300045013428, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.883430540561676, + "num_tokens": 387429527.0, + "step": 9980 + }, + { + "epoch": 1.2696857906118815, + "grad_norm": 2.5620293617248535, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8885266184806824, + "num_tokens": 387466532.0, + "step": 9981 + }, + { + "epoch": 1.269813000890472, + "grad_norm": 2.6302599906921387, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8821494579315186, + "num_tokens": 387505245.0, + "step": 9982 + }, + { + "epoch": 1.2699402111690625, + "grad_norm": 2.485943555831909, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8890252113342285, + "num_tokens": 387548389.0, + "step": 9983 + }, + { + "epoch": 1.2700674214476528, + "grad_norm": 2.482316732406616, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9007935523986816, + "num_tokens": 387587570.0, + "step": 9984 + }, + { + "epoch": 1.2701946317262434, + "grad_norm": 2.4596147537231445, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9027391672134399, + "num_tokens": 387624595.0, + "step": 9985 + }, + { + "epoch": 1.270321842004834, + "grad_norm": 2.697165012359619, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8920937180519104, + "num_tokens": 387661219.0, + "step": 9986 + }, + { + "epoch": 1.2704490522834244, + "grad_norm": 2.6009082794189453, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8855003714561462, + "num_tokens": 387702883.0, + "step": 9987 + }, + { + "epoch": 1.270576262562015, + "grad_norm": 2.6090824604034424, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8887806534767151, + "num_tokens": 387740840.0, + "step": 9988 + }, + { + "epoch": 1.2707034728406055, + "grad_norm": 2.6715376377105713, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8828113079071045, + "num_tokens": 387781167.0, + "step": 9989 + }, + { + "epoch": 1.270830683119196, + "grad_norm": 2.640535593032837, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8884993195533752, + "num_tokens": 387818178.0, + "step": 9990 + }, + { + "epoch": 1.2709578933977865, + "grad_norm": 2.777090072631836, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8893371224403381, + "num_tokens": 387854509.0, + "step": 9991 + }, + { + "epoch": 1.271085103676377, + "grad_norm": 2.7312517166137695, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8878172636032104, + "num_tokens": 387892684.0, + "step": 9992 + }, + { + "epoch": 1.2712123139549676, + "grad_norm": 2.9036941528320312, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.871527373790741, + "num_tokens": 387932601.0, + "step": 9993 + }, + { + "epoch": 1.2713395242335581, + "grad_norm": 2.7179486751556396, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8922386765480042, + "num_tokens": 387968264.0, + "step": 9994 + }, + { + "epoch": 1.2714667345121486, + "grad_norm": 2.7659757137298584, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8944779634475708, + "num_tokens": 388003034.0, + "step": 9995 + }, + { + "epoch": 1.2715939447907392, + "grad_norm": 2.683577299118042, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8883894085884094, + "num_tokens": 388039972.0, + "step": 9996 + }, + { + "epoch": 1.2717211550693297, + "grad_norm": 2.3828916549682617, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8855766654014587, + "num_tokens": 388088543.0, + "step": 9997 + }, + { + "epoch": 1.2718483653479202, + "grad_norm": 2.8435416221618652, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8843599557876587, + "num_tokens": 388127544.0, + "step": 9998 + }, + { + "epoch": 1.2719755756265108, + "grad_norm": 2.660390615463257, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.901099681854248, + "num_tokens": 388165423.0, + "step": 9999 + }, + { + "epoch": 1.272102785905101, + "grad_norm": 2.5949172973632812, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8978308439254761, + "num_tokens": 388207713.0, + "step": 10000 + }, + { + "epoch": 1.2722299961836916, + "grad_norm": 2.9498045444488525, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.885301947593689, + "num_tokens": 388241228.0, + "step": 10001 + }, + { + "epoch": 1.2723572064622821, + "grad_norm": 2.6082286834716797, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.9024819135665894, + "num_tokens": 388279910.0, + "step": 10002 + }, + { + "epoch": 1.2724844167408726, + "grad_norm": 2.826150417327881, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.904079794883728, + "num_tokens": 388311148.0, + "step": 10003 + }, + { + "epoch": 1.2726116270194632, + "grad_norm": 2.560410261154175, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8914520740509033, + "num_tokens": 388352894.0, + "step": 10004 + }, + { + "epoch": 1.2727388372980537, + "grad_norm": 2.7659363746643066, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8822649717330933, + "num_tokens": 388391942.0, + "step": 10005 + }, + { + "epoch": 1.2728660475766442, + "grad_norm": 2.553746223449707, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.889977216720581, + "num_tokens": 388434930.0, + "step": 10006 + }, + { + "epoch": 1.2729932578552348, + "grad_norm": 2.615922451019287, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8878484964370728, + "num_tokens": 388473329.0, + "step": 10007 + }, + { + "epoch": 1.2731204681338253, + "grad_norm": 2.5101518630981445, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8980952501296997, + "num_tokens": 388511698.0, + "step": 10008 + }, + { + "epoch": 1.2732476784124156, + "grad_norm": 2.7943387031555176, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8859607577323914, + "num_tokens": 388551057.0, + "step": 10009 + }, + { + "epoch": 1.2733748886910061, + "grad_norm": 2.5030128955841064, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8940083980560303, + "num_tokens": 388592190.0, + "step": 10010 + }, + { + "epoch": 1.2735020989695967, + "grad_norm": 2.541940689086914, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8911051750183105, + "num_tokens": 388635621.0, + "step": 10011 + }, + { + "epoch": 1.2736293092481872, + "grad_norm": 2.6007983684539795, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8943774700164795, + "num_tokens": 388674845.0, + "step": 10012 + }, + { + "epoch": 1.2737565195267777, + "grad_norm": 2.7627835273742676, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8904404640197754, + "num_tokens": 388710801.0, + "step": 10013 + }, + { + "epoch": 1.2738837298053682, + "grad_norm": 2.5518798828125, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.893613338470459, + "num_tokens": 388748540.0, + "step": 10014 + }, + { + "epoch": 1.2740109400839588, + "grad_norm": 2.6092677116394043, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8868072032928467, + "num_tokens": 388787253.0, + "step": 10015 + }, + { + "epoch": 1.2741381503625493, + "grad_norm": 2.4198856353759766, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8988317251205444, + "num_tokens": 388828566.0, + "step": 10016 + }, + { + "epoch": 1.2742653606411398, + "grad_norm": 2.7543790340423584, + "learning_rate": 1e-06, + "loss": 0.3602, + "mean_token_accuracy": 0.8726732730865479, + "num_tokens": 388871534.0, + "step": 10017 + }, + { + "epoch": 1.2743925709197303, + "grad_norm": 2.7837419509887695, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8910102248191833, + "num_tokens": 388910096.0, + "step": 10018 + }, + { + "epoch": 1.2745197811983209, + "grad_norm": 2.680535078048706, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8789747953414917, + "num_tokens": 388953049.0, + "step": 10019 + }, + { + "epoch": 1.2746469914769114, + "grad_norm": 2.68476939201355, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9000972509384155, + "num_tokens": 388988360.0, + "step": 10020 + }, + { + "epoch": 1.274774201755502, + "grad_norm": 2.498218059539795, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8863765597343445, + "num_tokens": 389032773.0, + "step": 10021 + }, + { + "epoch": 1.2749014120340925, + "grad_norm": 2.6677134037017822, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8892186880111694, + "num_tokens": 389073314.0, + "step": 10022 + }, + { + "epoch": 1.275028622312683, + "grad_norm": 2.854612350463867, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8863292932510376, + "num_tokens": 389111866.0, + "step": 10023 + }, + { + "epoch": 1.2751558325912733, + "grad_norm": 2.866119384765625, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8781865835189819, + "num_tokens": 389154097.0, + "step": 10024 + }, + { + "epoch": 1.2752830428698638, + "grad_norm": 2.6403656005859375, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9027721881866455, + "num_tokens": 389188894.0, + "step": 10025 + }, + { + "epoch": 1.2754102531484544, + "grad_norm": 3.1439297199249268, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8861296772956848, + "num_tokens": 389218753.0, + "step": 10026 + }, + { + "epoch": 1.2755374634270449, + "grad_norm": 2.4583537578582764, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8948094844818115, + "num_tokens": 389261403.0, + "step": 10027 + }, + { + "epoch": 1.2756646737056354, + "grad_norm": 2.658634662628174, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8906776309013367, + "num_tokens": 389302276.0, + "step": 10028 + }, + { + "epoch": 1.275791883984226, + "grad_norm": 2.7132647037506104, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8958596587181091, + "num_tokens": 389339460.0, + "step": 10029 + }, + { + "epoch": 1.2759190942628165, + "grad_norm": 2.5047082901000977, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8816871047019958, + "num_tokens": 389386111.0, + "step": 10030 + }, + { + "epoch": 1.276046304541407, + "grad_norm": 2.354879856109619, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9021886587142944, + "num_tokens": 389427862.0, + "step": 10031 + }, + { + "epoch": 1.2761735148199975, + "grad_norm": 2.662888288497925, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8874489068984985, + "num_tokens": 389467948.0, + "step": 10032 + }, + { + "epoch": 1.2763007250985878, + "grad_norm": 2.7076659202575684, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8836367726325989, + "num_tokens": 389507415.0, + "step": 10033 + }, + { + "epoch": 1.2764279353771784, + "grad_norm": 2.5467751026153564, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8887491226196289, + "num_tokens": 389549876.0, + "step": 10034 + }, + { + "epoch": 1.2765551456557689, + "grad_norm": 2.294450521469116, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9020513296127319, + "num_tokens": 389598698.0, + "step": 10035 + }, + { + "epoch": 1.2766823559343594, + "grad_norm": 2.4802627563476562, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8863745927810669, + "num_tokens": 389641730.0, + "step": 10036 + }, + { + "epoch": 1.27680956621295, + "grad_norm": 2.474686622619629, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8847352266311646, + "num_tokens": 389682594.0, + "step": 10037 + }, + { + "epoch": 1.2769367764915405, + "grad_norm": 2.8076486587524414, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8889554142951965, + "num_tokens": 389717001.0, + "step": 10038 + }, + { + "epoch": 1.277063986770131, + "grad_norm": 2.33970308303833, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8993247151374817, + "num_tokens": 389763058.0, + "step": 10039 + }, + { + "epoch": 1.2771911970487215, + "grad_norm": 2.6675000190734863, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9034252166748047, + "num_tokens": 389794006.0, + "step": 10040 + }, + { + "epoch": 1.277318407327312, + "grad_norm": 2.787872314453125, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8761849403381348, + "num_tokens": 389834430.0, + "step": 10041 + }, + { + "epoch": 1.2774456176059026, + "grad_norm": 2.5264546871185303, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9038755893707275, + "num_tokens": 389876038.0, + "step": 10042 + }, + { + "epoch": 1.2775728278844931, + "grad_norm": 2.618128538131714, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8951414823532104, + "num_tokens": 389913843.0, + "step": 10043 + }, + { + "epoch": 1.2777000381630836, + "grad_norm": 2.8116307258605957, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8875676393508911, + "num_tokens": 389950458.0, + "step": 10044 + }, + { + "epoch": 1.2778272484416742, + "grad_norm": 2.7159271240234375, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.892760694026947, + "num_tokens": 389987539.0, + "step": 10045 + }, + { + "epoch": 1.2779544587202647, + "grad_norm": 3.1040890216827393, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8749436140060425, + "num_tokens": 390022295.0, + "step": 10046 + }, + { + "epoch": 1.2780816689988552, + "grad_norm": 2.8676881790161133, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8841345906257629, + "num_tokens": 390059750.0, + "step": 10047 + }, + { + "epoch": 1.2782088792774458, + "grad_norm": 2.84101939201355, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8795751929283142, + "num_tokens": 390099622.0, + "step": 10048 + }, + { + "epoch": 1.278336089556036, + "grad_norm": 2.5669820308685303, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8794007897377014, + "num_tokens": 390143911.0, + "step": 10049 + }, + { + "epoch": 1.2784632998346266, + "grad_norm": 2.536479949951172, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8989874124526978, + "num_tokens": 390182386.0, + "step": 10050 + }, + { + "epoch": 1.2785905101132171, + "grad_norm": 2.59273624420166, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8905043005943298, + "num_tokens": 390224223.0, + "step": 10051 + }, + { + "epoch": 1.2787177203918076, + "grad_norm": 2.605278491973877, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8800495862960815, + "num_tokens": 390264827.0, + "step": 10052 + }, + { + "epoch": 1.2788449306703982, + "grad_norm": 2.7130374908447266, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8902653455734253, + "num_tokens": 390305228.0, + "step": 10053 + }, + { + "epoch": 1.2789721409489887, + "grad_norm": 2.423449993133545, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8961457014083862, + "num_tokens": 390349344.0, + "step": 10054 + }, + { + "epoch": 1.2790993512275792, + "grad_norm": 2.731196165084839, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8828916549682617, + "num_tokens": 390388852.0, + "step": 10055 + }, + { + "epoch": 1.2792265615061698, + "grad_norm": 2.613264560699463, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.891838550567627, + "num_tokens": 390428413.0, + "step": 10056 + }, + { + "epoch": 1.2793537717847603, + "grad_norm": 2.991844654083252, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8882439732551575, + "num_tokens": 390458364.0, + "step": 10057 + }, + { + "epoch": 1.2794809820633506, + "grad_norm": 2.7189364433288574, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8780502080917358, + "num_tokens": 390497809.0, + "step": 10058 + }, + { + "epoch": 1.2796081923419411, + "grad_norm": 2.9301364421844482, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8891369104385376, + "num_tokens": 390530552.0, + "step": 10059 + }, + { + "epoch": 1.2797354026205316, + "grad_norm": 2.4983222484588623, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8916969299316406, + "num_tokens": 390570143.0, + "step": 10060 + }, + { + "epoch": 1.2798626128991222, + "grad_norm": 2.704625129699707, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8820368051528931, + "num_tokens": 390610888.0, + "step": 10061 + }, + { + "epoch": 1.2799898231777127, + "grad_norm": 2.6778345108032227, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8981363773345947, + "num_tokens": 390646392.0, + "step": 10062 + }, + { + "epoch": 1.2801170334563032, + "grad_norm": 2.9217753410339355, + "learning_rate": 1e-06, + "loss": 0.3702, + "mean_token_accuracy": 0.870966911315918, + "num_tokens": 390687323.0, + "step": 10063 + }, + { + "epoch": 1.2802442437348938, + "grad_norm": 2.8427932262420654, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8781222701072693, + "num_tokens": 390723863.0, + "step": 10064 + }, + { + "epoch": 1.2803714540134843, + "grad_norm": 2.7224884033203125, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8804061412811279, + "num_tokens": 390761117.0, + "step": 10065 + }, + { + "epoch": 1.2804986642920748, + "grad_norm": 2.708548069000244, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8829836845397949, + "num_tokens": 390802556.0, + "step": 10066 + }, + { + "epoch": 1.2806258745706653, + "grad_norm": 2.6297528743743896, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8932909369468689, + "num_tokens": 390844193.0, + "step": 10067 + }, + { + "epoch": 1.2807530848492559, + "grad_norm": 2.6187853813171387, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8841527104377747, + "num_tokens": 390886197.0, + "step": 10068 + }, + { + "epoch": 1.2808802951278464, + "grad_norm": 2.43281626701355, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8948765993118286, + "num_tokens": 390927924.0, + "step": 10069 + }, + { + "epoch": 1.281007505406437, + "grad_norm": 2.679699420928955, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8815641403198242, + "num_tokens": 390965582.0, + "step": 10070 + }, + { + "epoch": 1.2811347156850275, + "grad_norm": 2.561114549636841, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8979621529579163, + "num_tokens": 391001135.0, + "step": 10071 + }, + { + "epoch": 1.281261925963618, + "grad_norm": 2.560469627380371, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8760859370231628, + "num_tokens": 391045139.0, + "step": 10072 + }, + { + "epoch": 1.2813891362422083, + "grad_norm": 2.7240657806396484, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8866053223609924, + "num_tokens": 391083710.0, + "step": 10073 + }, + { + "epoch": 1.2815163465207988, + "grad_norm": 2.737553834915161, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8939293622970581, + "num_tokens": 391118719.0, + "step": 10074 + }, + { + "epoch": 1.2816435567993893, + "grad_norm": 2.7499048709869385, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8953890800476074, + "num_tokens": 391157783.0, + "step": 10075 + }, + { + "epoch": 1.2817707670779799, + "grad_norm": 2.513197422027588, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9020705223083496, + "num_tokens": 391195398.0, + "step": 10076 + }, + { + "epoch": 1.2818979773565704, + "grad_norm": 2.369126319885254, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8966954350471497, + "num_tokens": 391241117.0, + "step": 10077 + }, + { + "epoch": 1.282025187635161, + "grad_norm": 2.6803135871887207, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.880852222442627, + "num_tokens": 391279793.0, + "step": 10078 + }, + { + "epoch": 1.2821523979137515, + "grad_norm": 2.6403021812438965, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8885737657546997, + "num_tokens": 391319108.0, + "step": 10079 + }, + { + "epoch": 1.282279608192342, + "grad_norm": 2.760509729385376, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8870100975036621, + "num_tokens": 391357069.0, + "step": 10080 + }, + { + "epoch": 1.2824068184709325, + "grad_norm": 2.8404345512390137, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8895399570465088, + "num_tokens": 391389993.0, + "step": 10081 + }, + { + "epoch": 1.2825340287495228, + "grad_norm": 2.7401645183563232, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8876559138298035, + "num_tokens": 391430206.0, + "step": 10082 + }, + { + "epoch": 1.2826612390281134, + "grad_norm": 2.668856620788574, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8866500854492188, + "num_tokens": 391471612.0, + "step": 10083 + }, + { + "epoch": 1.2827884493067039, + "grad_norm": 2.4512436389923096, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8986256122589111, + "num_tokens": 391515700.0, + "step": 10084 + }, + { + "epoch": 1.2829156595852944, + "grad_norm": 2.6852550506591797, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8913651704788208, + "num_tokens": 391554608.0, + "step": 10085 + }, + { + "epoch": 1.283042869863885, + "grad_norm": 2.8292016983032227, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.88719642162323, + "num_tokens": 391591217.0, + "step": 10086 + }, + { + "epoch": 1.2831700801424755, + "grad_norm": 2.4254043102264404, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8902449011802673, + "num_tokens": 391636603.0, + "step": 10087 + }, + { + "epoch": 1.283297290421066, + "grad_norm": 2.4887471199035645, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9014495015144348, + "num_tokens": 391677238.0, + "step": 10088 + }, + { + "epoch": 1.2834245006996565, + "grad_norm": 2.585204839706421, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8903772830963135, + "num_tokens": 391718357.0, + "step": 10089 + }, + { + "epoch": 1.283551710978247, + "grad_norm": 2.9488959312438965, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8857846260070801, + "num_tokens": 391753663.0, + "step": 10090 + }, + { + "epoch": 1.2836789212568376, + "grad_norm": 2.6974453926086426, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8862916231155396, + "num_tokens": 391793034.0, + "step": 10091 + }, + { + "epoch": 1.283806131535428, + "grad_norm": 2.925870180130005, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8892138004302979, + "num_tokens": 391824096.0, + "step": 10092 + }, + { + "epoch": 1.2839333418140186, + "grad_norm": 2.71358585357666, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8754727840423584, + "num_tokens": 391862757.0, + "step": 10093 + }, + { + "epoch": 1.2840605520926092, + "grad_norm": 2.658428430557251, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9055958986282349, + "num_tokens": 391898679.0, + "step": 10094 + }, + { + "epoch": 1.2841877623711997, + "grad_norm": 2.8720619678497314, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8837260007858276, + "num_tokens": 391933521.0, + "step": 10095 + }, + { + "epoch": 1.2843149726497902, + "grad_norm": 2.8115832805633545, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8910547494888306, + "num_tokens": 391967839.0, + "step": 10096 + }, + { + "epoch": 1.2844421829283807, + "grad_norm": 2.957716226577759, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8843786120414734, + "num_tokens": 392005487.0, + "step": 10097 + }, + { + "epoch": 1.284569393206971, + "grad_norm": 2.537419080734253, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8969792127609253, + "num_tokens": 392045570.0, + "step": 10098 + }, + { + "epoch": 1.2846966034855616, + "grad_norm": 2.5953915119171143, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8874481916427612, + "num_tokens": 392086734.0, + "step": 10099 + }, + { + "epoch": 1.2848238137641521, + "grad_norm": 2.9873056411743164, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8848295211791992, + "num_tokens": 392119672.0, + "step": 10100 + }, + { + "epoch": 1.2849510240427426, + "grad_norm": 2.774271249771118, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8981152772903442, + "num_tokens": 392156623.0, + "step": 10101 + }, + { + "epoch": 1.2850782343213332, + "grad_norm": 2.677356004714966, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.885323166847229, + "num_tokens": 392196478.0, + "step": 10102 + }, + { + "epoch": 1.2852054445999237, + "grad_norm": 2.7097949981689453, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8871133327484131, + "num_tokens": 392237307.0, + "step": 10103 + }, + { + "epoch": 1.2853326548785142, + "grad_norm": 2.540565013885498, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.881294846534729, + "num_tokens": 392280735.0, + "step": 10104 + }, + { + "epoch": 1.2854598651571048, + "grad_norm": 2.4549200534820557, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8920497298240662, + "num_tokens": 392322730.0, + "step": 10105 + }, + { + "epoch": 1.2855870754356953, + "grad_norm": 2.661468267440796, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9000515341758728, + "num_tokens": 392360596.0, + "step": 10106 + }, + { + "epoch": 1.2857142857142856, + "grad_norm": 2.9106993675231934, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9009629487991333, + "num_tokens": 392391268.0, + "step": 10107 + }, + { + "epoch": 1.2858414959928761, + "grad_norm": 2.598362684249878, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8920915722846985, + "num_tokens": 392429860.0, + "step": 10108 + }, + { + "epoch": 1.2859687062714666, + "grad_norm": 2.65554141998291, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8868957757949829, + "num_tokens": 392467248.0, + "step": 10109 + }, + { + "epoch": 1.2860959165500572, + "grad_norm": 2.399489402770996, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8984193801879883, + "num_tokens": 392512766.0, + "step": 10110 + }, + { + "epoch": 1.2862231268286477, + "grad_norm": 2.918863534927368, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8808009028434753, + "num_tokens": 392546059.0, + "step": 10111 + }, + { + "epoch": 1.2863503371072382, + "grad_norm": 2.674903154373169, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8906222581863403, + "num_tokens": 392587197.0, + "step": 10112 + }, + { + "epoch": 1.2864775473858288, + "grad_norm": 2.5717551708221436, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8919815421104431, + "num_tokens": 392627758.0, + "step": 10113 + }, + { + "epoch": 1.2866047576644193, + "grad_norm": 2.5250847339630127, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9005084037780762, + "num_tokens": 392667165.0, + "step": 10114 + }, + { + "epoch": 1.2867319679430098, + "grad_norm": 2.563605785369873, + "learning_rate": 1e-06, + "loss": 0.369, + "mean_token_accuracy": 0.8667728900909424, + "num_tokens": 392713825.0, + "step": 10115 + }, + { + "epoch": 1.2868591782216003, + "grad_norm": 2.599425792694092, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8898818492889404, + "num_tokens": 392751402.0, + "step": 10116 + }, + { + "epoch": 1.2869863885001909, + "grad_norm": 2.8157849311828613, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.8824766874313354, + "num_tokens": 392787342.0, + "step": 10117 + }, + { + "epoch": 1.2871135987787814, + "grad_norm": 2.6316754817962646, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8961794376373291, + "num_tokens": 392827055.0, + "step": 10118 + }, + { + "epoch": 1.287240809057372, + "grad_norm": 2.6026010513305664, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8999917507171631, + "num_tokens": 392865582.0, + "step": 10119 + }, + { + "epoch": 1.2873680193359625, + "grad_norm": 2.5975987911224365, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8936407566070557, + "num_tokens": 392903020.0, + "step": 10120 + }, + { + "epoch": 1.287495229614553, + "grad_norm": 2.6031227111816406, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8985027074813843, + "num_tokens": 392942757.0, + "step": 10121 + }, + { + "epoch": 1.2876224398931433, + "grad_norm": 2.499664068222046, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9015823602676392, + "num_tokens": 392980458.0, + "step": 10122 + }, + { + "epoch": 1.2877496501717338, + "grad_norm": 2.749432325363159, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8874968886375427, + "num_tokens": 393017844.0, + "step": 10123 + }, + { + "epoch": 1.2878768604503243, + "grad_norm": 2.8673648834228516, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8891941905021667, + "num_tokens": 393051065.0, + "step": 10124 + }, + { + "epoch": 1.2880040707289149, + "grad_norm": 2.591723918914795, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8887133002281189, + "num_tokens": 393090149.0, + "step": 10125 + }, + { + "epoch": 1.2881312810075054, + "grad_norm": 2.5117557048797607, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8916999101638794, + "num_tokens": 393133996.0, + "step": 10126 + }, + { + "epoch": 1.288258491286096, + "grad_norm": 2.6110808849334717, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8799814581871033, + "num_tokens": 393177267.0, + "step": 10127 + }, + { + "epoch": 1.2883857015646865, + "grad_norm": 2.6076908111572266, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8850266337394714, + "num_tokens": 393217373.0, + "step": 10128 + }, + { + "epoch": 1.288512911843277, + "grad_norm": 3.0906529426574707, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8803579211235046, + "num_tokens": 393254800.0, + "step": 10129 + }, + { + "epoch": 1.2886401221218675, + "grad_norm": 2.498307943344116, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8907830119132996, + "num_tokens": 393298400.0, + "step": 10130 + }, + { + "epoch": 1.2887673324004578, + "grad_norm": 2.6672513484954834, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8903293609619141, + "num_tokens": 393340369.0, + "step": 10131 + }, + { + "epoch": 1.2888945426790483, + "grad_norm": 2.704461097717285, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8901584148406982, + "num_tokens": 393379174.0, + "step": 10132 + }, + { + "epoch": 1.2890217529576389, + "grad_norm": 2.725144386291504, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8920786380767822, + "num_tokens": 393414554.0, + "step": 10133 + }, + { + "epoch": 1.2891489632362294, + "grad_norm": 2.4339499473571777, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9057843685150146, + "num_tokens": 393453808.0, + "step": 10134 + }, + { + "epoch": 1.28927617351482, + "grad_norm": 2.668008327484131, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8889440894126892, + "num_tokens": 393494345.0, + "step": 10135 + }, + { + "epoch": 1.2894033837934105, + "grad_norm": 2.5813181400299072, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8915491700172424, + "num_tokens": 393536916.0, + "step": 10136 + }, + { + "epoch": 1.289530594072001, + "grad_norm": 2.696209192276001, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8912546038627625, + "num_tokens": 393576356.0, + "step": 10137 + }, + { + "epoch": 1.2896578043505915, + "grad_norm": 2.4013264179229736, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9015477895736694, + "num_tokens": 393613801.0, + "step": 10138 + }, + { + "epoch": 1.289785014629182, + "grad_norm": 2.539268970489502, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8902354836463928, + "num_tokens": 393655847.0, + "step": 10139 + }, + { + "epoch": 1.2899122249077726, + "grad_norm": 2.613476276397705, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8898462057113647, + "num_tokens": 393697474.0, + "step": 10140 + }, + { + "epoch": 1.290039435186363, + "grad_norm": 2.5927038192749023, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.893622875213623, + "num_tokens": 393736360.0, + "step": 10141 + }, + { + "epoch": 1.2901666454649536, + "grad_norm": 2.4887161254882812, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8949453234672546, + "num_tokens": 393777962.0, + "step": 10142 + }, + { + "epoch": 1.2902938557435442, + "grad_norm": 2.8330395221710205, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.89061439037323, + "num_tokens": 393812749.0, + "step": 10143 + }, + { + "epoch": 1.2904210660221347, + "grad_norm": 2.8542683124542236, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8929589986801147, + "num_tokens": 393846147.0, + "step": 10144 + }, + { + "epoch": 1.2905482763007252, + "grad_norm": 2.573256254196167, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8787814378738403, + "num_tokens": 393890500.0, + "step": 10145 + }, + { + "epoch": 1.2906754865793157, + "grad_norm": 2.815070390701294, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8817166090011597, + "num_tokens": 393928875.0, + "step": 10146 + }, + { + "epoch": 1.290802696857906, + "grad_norm": 2.6609983444213867, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8939711451530457, + "num_tokens": 393965448.0, + "step": 10147 + }, + { + "epoch": 1.2909299071364966, + "grad_norm": 2.463940382003784, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.897294282913208, + "num_tokens": 394007541.0, + "step": 10148 + }, + { + "epoch": 1.291057117415087, + "grad_norm": 2.6262834072113037, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8868242502212524, + "num_tokens": 394052587.0, + "step": 10149 + }, + { + "epoch": 1.2911843276936776, + "grad_norm": 2.6738967895507812, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8827850222587585, + "num_tokens": 394091391.0, + "step": 10150 + }, + { + "epoch": 1.2913115379722682, + "grad_norm": 2.410601854324341, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.895991861820221, + "num_tokens": 394132353.0, + "step": 10151 + }, + { + "epoch": 1.2914387482508587, + "grad_norm": 2.5468788146972656, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8915845155715942, + "num_tokens": 394174634.0, + "step": 10152 + }, + { + "epoch": 1.2915659585294492, + "grad_norm": 2.5230300426483154, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8864050507545471, + "num_tokens": 394218351.0, + "step": 10153 + }, + { + "epoch": 1.2916931688080397, + "grad_norm": 2.7241499423980713, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8882378935813904, + "num_tokens": 394256096.0, + "step": 10154 + }, + { + "epoch": 1.2918203790866303, + "grad_norm": 2.685582160949707, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9032285213470459, + "num_tokens": 394290348.0, + "step": 10155 + }, + { + "epoch": 1.2919475893652206, + "grad_norm": 2.4851863384246826, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8966948986053467, + "num_tokens": 394331284.0, + "step": 10156 + }, + { + "epoch": 1.2920747996438111, + "grad_norm": 2.619201898574829, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8956416845321655, + "num_tokens": 394368572.0, + "step": 10157 + }, + { + "epoch": 1.2922020099224016, + "grad_norm": 2.8797361850738525, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8956888914108276, + "num_tokens": 394404227.0, + "step": 10158 + }, + { + "epoch": 1.2923292202009922, + "grad_norm": 2.5826737880706787, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.896487832069397, + "num_tokens": 394442343.0, + "step": 10159 + }, + { + "epoch": 1.2924564304795827, + "grad_norm": 2.780898094177246, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8821049928665161, + "num_tokens": 394480482.0, + "step": 10160 + }, + { + "epoch": 1.2925836407581732, + "grad_norm": 2.8909354209899902, + "learning_rate": 1e-06, + "loss": 0.3677, + "mean_token_accuracy": 0.8695811629295349, + "num_tokens": 394517285.0, + "step": 10161 + }, + { + "epoch": 1.2927108510367638, + "grad_norm": 2.848381996154785, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8913947939872742, + "num_tokens": 394551602.0, + "step": 10162 + }, + { + "epoch": 1.2928380613153543, + "grad_norm": 2.78550124168396, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8730279803276062, + "num_tokens": 394593534.0, + "step": 10163 + }, + { + "epoch": 1.2929652715939448, + "grad_norm": 2.615015983581543, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.885077714920044, + "num_tokens": 394637633.0, + "step": 10164 + }, + { + "epoch": 1.2930924818725353, + "grad_norm": 2.6952319145202637, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.888422966003418, + "num_tokens": 394673475.0, + "step": 10165 + }, + { + "epoch": 1.2932196921511259, + "grad_norm": 2.591036558151245, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8859227895736694, + "num_tokens": 394717609.0, + "step": 10166 + }, + { + "epoch": 1.2933469024297164, + "grad_norm": 2.755511522293091, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8968045711517334, + "num_tokens": 394755772.0, + "step": 10167 + }, + { + "epoch": 1.293474112708307, + "grad_norm": 2.8202264308929443, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8939913511276245, + "num_tokens": 394792335.0, + "step": 10168 + }, + { + "epoch": 1.2936013229868975, + "grad_norm": 2.8299872875213623, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8768798112869263, + "num_tokens": 394833302.0, + "step": 10169 + }, + { + "epoch": 1.293728533265488, + "grad_norm": 2.68662166595459, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9022778272628784, + "num_tokens": 394870165.0, + "step": 10170 + }, + { + "epoch": 1.2938557435440783, + "grad_norm": 2.4552323818206787, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8960518836975098, + "num_tokens": 394915067.0, + "step": 10171 + }, + { + "epoch": 1.2939829538226688, + "grad_norm": 2.7896225452423096, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8828099966049194, + "num_tokens": 394951068.0, + "step": 10172 + }, + { + "epoch": 1.2941101641012593, + "grad_norm": 2.8129334449768066, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8960326313972473, + "num_tokens": 394987759.0, + "step": 10173 + }, + { + "epoch": 1.2942373743798499, + "grad_norm": 2.8914289474487305, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8789651393890381, + "num_tokens": 395027663.0, + "step": 10174 + }, + { + "epoch": 1.2943645846584404, + "grad_norm": 2.595684051513672, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8943817615509033, + "num_tokens": 395067543.0, + "step": 10175 + }, + { + "epoch": 1.294491794937031, + "grad_norm": 2.6798129081726074, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8856917023658752, + "num_tokens": 395107845.0, + "step": 10176 + }, + { + "epoch": 1.2946190052156215, + "grad_norm": 3.0396182537078857, + "learning_rate": 1e-06, + "loss": 0.3236, + "mean_token_accuracy": 0.8836333155632019, + "num_tokens": 395138348.0, + "step": 10177 + }, + { + "epoch": 1.294746215494212, + "grad_norm": 2.767512798309326, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8887639045715332, + "num_tokens": 395178194.0, + "step": 10178 + }, + { + "epoch": 1.2948734257728025, + "grad_norm": 2.6302428245544434, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8846717476844788, + "num_tokens": 395220313.0, + "step": 10179 + }, + { + "epoch": 1.2950006360513928, + "grad_norm": 2.453791379928589, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.9024255275726318, + "num_tokens": 395259259.0, + "step": 10180 + }, + { + "epoch": 1.2951278463299833, + "grad_norm": 2.935363531112671, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8854025602340698, + "num_tokens": 395294267.0, + "step": 10181 + }, + { + "epoch": 1.2952550566085739, + "grad_norm": 2.4704229831695557, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8978895545005798, + "num_tokens": 395333267.0, + "step": 10182 + }, + { + "epoch": 1.2953822668871644, + "grad_norm": 2.552846908569336, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8911661505699158, + "num_tokens": 395375759.0, + "step": 10183 + }, + { + "epoch": 1.295509477165755, + "grad_norm": 2.6205782890319824, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8809463977813721, + "num_tokens": 395414684.0, + "step": 10184 + }, + { + "epoch": 1.2956366874443455, + "grad_norm": 2.744967460632324, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8947648406028748, + "num_tokens": 395449664.0, + "step": 10185 + }, + { + "epoch": 1.295763897722936, + "grad_norm": 2.4155218601226807, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.901252806186676, + "num_tokens": 395495358.0, + "step": 10186 + }, + { + "epoch": 1.2958911080015265, + "grad_norm": 2.682356357574463, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8958643674850464, + "num_tokens": 395528489.0, + "step": 10187 + }, + { + "epoch": 1.296018318280117, + "grad_norm": 2.702393054962158, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.894623339176178, + "num_tokens": 395564388.0, + "step": 10188 + }, + { + "epoch": 1.2961455285587076, + "grad_norm": 2.6947247982025146, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.8980177640914917, + "num_tokens": 395598065.0, + "step": 10189 + }, + { + "epoch": 1.296272738837298, + "grad_norm": 2.8127388954162598, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8900747299194336, + "num_tokens": 395637827.0, + "step": 10190 + }, + { + "epoch": 1.2963999491158886, + "grad_norm": 2.809892177581787, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8863243460655212, + "num_tokens": 395673871.0, + "step": 10191 + }, + { + "epoch": 1.2965271593944792, + "grad_norm": 2.8905858993530273, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.889564573764801, + "num_tokens": 395707123.0, + "step": 10192 + }, + { + "epoch": 1.2966543696730697, + "grad_norm": 2.870516777038574, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8825896978378296, + "num_tokens": 395742383.0, + "step": 10193 + }, + { + "epoch": 1.2967815799516602, + "grad_norm": 2.577096700668335, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8992111682891846, + "num_tokens": 395780679.0, + "step": 10194 + }, + { + "epoch": 1.2969087902302507, + "grad_norm": 2.741706609725952, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8890109062194824, + "num_tokens": 395819604.0, + "step": 10195 + }, + { + "epoch": 1.297036000508841, + "grad_norm": 2.913025379180908, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8801801204681396, + "num_tokens": 395854120.0, + "step": 10196 + }, + { + "epoch": 1.2971632107874316, + "grad_norm": 2.7726757526397705, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8956030607223511, + "num_tokens": 395889859.0, + "step": 10197 + }, + { + "epoch": 1.297290421066022, + "grad_norm": 2.6032304763793945, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8963183164596558, + "num_tokens": 395930658.0, + "step": 10198 + }, + { + "epoch": 1.2974176313446126, + "grad_norm": 2.6200110912323, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8947032690048218, + "num_tokens": 395969999.0, + "step": 10199 + }, + { + "epoch": 1.2975448416232032, + "grad_norm": 2.5099544525146484, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.898802638053894, + "num_tokens": 396012168.0, + "step": 10200 + }, + { + "epoch": 1.2976720519017937, + "grad_norm": 2.6905574798583984, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8900269865989685, + "num_tokens": 396051919.0, + "step": 10201 + }, + { + "epoch": 1.2977992621803842, + "grad_norm": 2.6704914569854736, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8953326940536499, + "num_tokens": 396090444.0, + "step": 10202 + }, + { + "epoch": 1.2979264724589747, + "grad_norm": 2.6211977005004883, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8903191089630127, + "num_tokens": 396129213.0, + "step": 10203 + }, + { + "epoch": 1.2980536827375653, + "grad_norm": 2.6981964111328125, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8805472254753113, + "num_tokens": 396169847.0, + "step": 10204 + }, + { + "epoch": 1.2981808930161556, + "grad_norm": 2.4810352325439453, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8948512077331543, + "num_tokens": 396212996.0, + "step": 10205 + }, + { + "epoch": 1.298308103294746, + "grad_norm": 2.6492931842803955, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8886048793792725, + "num_tokens": 396254147.0, + "step": 10206 + }, + { + "epoch": 1.2984353135733366, + "grad_norm": 2.514183282852173, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9022903442382812, + "num_tokens": 396290679.0, + "step": 10207 + }, + { + "epoch": 1.2985625238519272, + "grad_norm": 2.849066734313965, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8840436935424805, + "num_tokens": 396325659.0, + "step": 10208 + }, + { + "epoch": 1.2986897341305177, + "grad_norm": 2.609417676925659, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9046643376350403, + "num_tokens": 396357258.0, + "step": 10209 + }, + { + "epoch": 1.2988169444091082, + "grad_norm": 2.617128610610962, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8935403823852539, + "num_tokens": 396395238.0, + "step": 10210 + }, + { + "epoch": 1.2989441546876987, + "grad_norm": 2.5435101985931396, + "learning_rate": 1e-06, + "loss": 0.3667, + "mean_token_accuracy": 0.8680585622787476, + "num_tokens": 396440645.0, + "step": 10211 + }, + { + "epoch": 1.2990713649662893, + "grad_norm": 2.699819326400757, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.892692506313324, + "num_tokens": 396476200.0, + "step": 10212 + }, + { + "epoch": 1.2991985752448798, + "grad_norm": 2.7571468353271484, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8917974233627319, + "num_tokens": 396510801.0, + "step": 10213 + }, + { + "epoch": 1.2993257855234703, + "grad_norm": 2.535648822784424, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9033374190330505, + "num_tokens": 396546780.0, + "step": 10214 + }, + { + "epoch": 1.2994529958020609, + "grad_norm": 2.903646230697632, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8835157155990601, + "num_tokens": 396582364.0, + "step": 10215 + }, + { + "epoch": 1.2995802060806514, + "grad_norm": 2.747906446456909, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8867557048797607, + "num_tokens": 396619244.0, + "step": 10216 + }, + { + "epoch": 1.299707416359242, + "grad_norm": 2.5866291522979736, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9075980186462402, + "num_tokens": 396657992.0, + "step": 10217 + }, + { + "epoch": 1.2998346266378324, + "grad_norm": 2.486388921737671, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8953601121902466, + "num_tokens": 396699745.0, + "step": 10218 + }, + { + "epoch": 1.299961836916423, + "grad_norm": 2.77413010597229, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8978245258331299, + "num_tokens": 396734742.0, + "step": 10219 + }, + { + "epoch": 1.3000890471950133, + "grad_norm": 2.6834604740142822, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8975197672843933, + "num_tokens": 396770223.0, + "step": 10220 + }, + { + "epoch": 1.3002162574736038, + "grad_norm": 2.620983600616455, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8809136152267456, + "num_tokens": 396811738.0, + "step": 10221 + }, + { + "epoch": 1.3003434677521943, + "grad_norm": 2.788536787033081, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8943537473678589, + "num_tokens": 396844328.0, + "step": 10222 + }, + { + "epoch": 1.3004706780307849, + "grad_norm": 2.9033169746398926, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8823392391204834, + "num_tokens": 396880183.0, + "step": 10223 + }, + { + "epoch": 1.3005978883093754, + "grad_norm": 2.7943179607391357, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8999863862991333, + "num_tokens": 396913132.0, + "step": 10224 + }, + { + "epoch": 1.300725098587966, + "grad_norm": 2.633193016052246, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8862254619598389, + "num_tokens": 396951085.0, + "step": 10225 + }, + { + "epoch": 1.3008523088665565, + "grad_norm": 2.673123598098755, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8939822316169739, + "num_tokens": 396988093.0, + "step": 10226 + }, + { + "epoch": 1.300979519145147, + "grad_norm": 2.505748748779297, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8857009410858154, + "num_tokens": 397032943.0, + "step": 10227 + }, + { + "epoch": 1.3011067294237375, + "grad_norm": 2.756061315536499, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8800742626190186, + "num_tokens": 397069367.0, + "step": 10228 + }, + { + "epoch": 1.3012339397023278, + "grad_norm": 2.662407398223877, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8832236528396606, + "num_tokens": 397109636.0, + "step": 10229 + }, + { + "epoch": 1.3013611499809183, + "grad_norm": 2.7486891746520996, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8906390070915222, + "num_tokens": 397145630.0, + "step": 10230 + }, + { + "epoch": 1.3014883602595089, + "grad_norm": 2.5589282512664795, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8885043859481812, + "num_tokens": 397188405.0, + "step": 10231 + }, + { + "epoch": 1.3016155705380994, + "grad_norm": 2.6551384925842285, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8887569308280945, + "num_tokens": 397234328.0, + "step": 10232 + }, + { + "epoch": 1.30174278081669, + "grad_norm": 2.4598777294158936, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8965574502944946, + "num_tokens": 397278116.0, + "step": 10233 + }, + { + "epoch": 1.3018699910952805, + "grad_norm": 2.748884916305542, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8905354738235474, + "num_tokens": 397314561.0, + "step": 10234 + }, + { + "epoch": 1.301997201373871, + "grad_norm": 2.808863401412964, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8970124125480652, + "num_tokens": 397351686.0, + "step": 10235 + }, + { + "epoch": 1.3021244116524615, + "grad_norm": 2.804277181625366, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8865785598754883, + "num_tokens": 397388142.0, + "step": 10236 + }, + { + "epoch": 1.302251621931052, + "grad_norm": 2.5935564041137695, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.8788213729858398, + "num_tokens": 397431318.0, + "step": 10237 + }, + { + "epoch": 1.3023788322096426, + "grad_norm": 2.715867280960083, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8832714557647705, + "num_tokens": 397470411.0, + "step": 10238 + }, + { + "epoch": 1.302506042488233, + "grad_norm": 2.6817569732666016, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8851615190505981, + "num_tokens": 397511720.0, + "step": 10239 + }, + { + "epoch": 1.3026332527668236, + "grad_norm": 2.532231569290161, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8883029222488403, + "num_tokens": 397553954.0, + "step": 10240 + }, + { + "epoch": 1.3027604630454142, + "grad_norm": 2.670713186264038, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.899488091468811, + "num_tokens": 397585719.0, + "step": 10241 + }, + { + "epoch": 1.3028876733240047, + "grad_norm": 2.4585704803466797, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9001549482345581, + "num_tokens": 397625912.0, + "step": 10242 + }, + { + "epoch": 1.3030148836025952, + "grad_norm": 2.8144938945770264, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8969998359680176, + "num_tokens": 397659708.0, + "step": 10243 + }, + { + "epoch": 1.3031420938811857, + "grad_norm": 2.7663021087646484, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8812419176101685, + "num_tokens": 397697053.0, + "step": 10244 + }, + { + "epoch": 1.303269304159776, + "grad_norm": 2.719688653945923, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8892982006072998, + "num_tokens": 397732214.0, + "step": 10245 + }, + { + "epoch": 1.3033965144383666, + "grad_norm": 2.722419261932373, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.900909960269928, + "num_tokens": 397766110.0, + "step": 10246 + }, + { + "epoch": 1.303523724716957, + "grad_norm": 2.5540859699249268, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8879467248916626, + "num_tokens": 397808646.0, + "step": 10247 + }, + { + "epoch": 1.3036509349955476, + "grad_norm": 2.597238063812256, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8852751851081848, + "num_tokens": 397854941.0, + "step": 10248 + }, + { + "epoch": 1.3037781452741382, + "grad_norm": 2.603839874267578, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8951272964477539, + "num_tokens": 397890441.0, + "step": 10249 + }, + { + "epoch": 1.3039053555527287, + "grad_norm": 2.9574897289276123, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8967937231063843, + "num_tokens": 397923140.0, + "step": 10250 + }, + { + "epoch": 1.3040325658313192, + "grad_norm": 2.5056264400482178, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8903830051422119, + "num_tokens": 397964514.0, + "step": 10251 + }, + { + "epoch": 1.3041597761099097, + "grad_norm": 2.420393228530884, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9004492163658142, + "num_tokens": 398005268.0, + "step": 10252 + }, + { + "epoch": 1.3042869863885003, + "grad_norm": 2.657382011413574, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8838661909103394, + "num_tokens": 398045390.0, + "step": 10253 + }, + { + "epoch": 1.3044141966670906, + "grad_norm": 2.7148852348327637, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.886858344078064, + "num_tokens": 398084131.0, + "step": 10254 + }, + { + "epoch": 1.304541406945681, + "grad_norm": 2.4466516971588135, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8967183828353882, + "num_tokens": 398123720.0, + "step": 10255 + }, + { + "epoch": 1.3046686172242716, + "grad_norm": 2.778351306915283, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8937000632286072, + "num_tokens": 398159724.0, + "step": 10256 + }, + { + "epoch": 1.3047958275028622, + "grad_norm": 2.6762855052948, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8941627740859985, + "num_tokens": 398196876.0, + "step": 10257 + }, + { + "epoch": 1.3049230377814527, + "grad_norm": 2.791887044906616, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8876914978027344, + "num_tokens": 398233700.0, + "step": 10258 + }, + { + "epoch": 1.3050502480600432, + "grad_norm": 2.853182315826416, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.889132022857666, + "num_tokens": 398271447.0, + "step": 10259 + }, + { + "epoch": 1.3051774583386337, + "grad_norm": 2.6021387577056885, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8898434042930603, + "num_tokens": 398310657.0, + "step": 10260 + }, + { + "epoch": 1.3053046686172243, + "grad_norm": 2.6706483364105225, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8884730935096741, + "num_tokens": 398351875.0, + "step": 10261 + }, + { + "epoch": 1.3054318788958148, + "grad_norm": 2.5611279010772705, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8929241895675659, + "num_tokens": 398393388.0, + "step": 10262 + }, + { + "epoch": 1.3055590891744053, + "grad_norm": 2.5383400917053223, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8889773488044739, + "num_tokens": 398438348.0, + "step": 10263 + }, + { + "epoch": 1.3056862994529959, + "grad_norm": 2.528581142425537, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8921000957489014, + "num_tokens": 398481303.0, + "step": 10264 + }, + { + "epoch": 1.3058135097315864, + "grad_norm": 2.889871597290039, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8856099247932434, + "num_tokens": 398515933.0, + "step": 10265 + }, + { + "epoch": 1.305940720010177, + "grad_norm": 2.6019949913024902, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8884422779083252, + "num_tokens": 398556925.0, + "step": 10266 + }, + { + "epoch": 1.3060679302887674, + "grad_norm": 2.5809242725372314, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8876457214355469, + "num_tokens": 398598567.0, + "step": 10267 + }, + { + "epoch": 1.306195140567358, + "grad_norm": 2.823988437652588, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.888822078704834, + "num_tokens": 398633810.0, + "step": 10268 + }, + { + "epoch": 1.3063223508459483, + "grad_norm": 2.7695281505584717, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8853116035461426, + "num_tokens": 398671061.0, + "step": 10269 + }, + { + "epoch": 1.3064495611245388, + "grad_norm": 2.6891024112701416, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8900980949401855, + "num_tokens": 398710756.0, + "step": 10270 + }, + { + "epoch": 1.3065767714031293, + "grad_norm": 2.6450142860412598, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.9009814262390137, + "num_tokens": 398748091.0, + "step": 10271 + }, + { + "epoch": 1.3067039816817199, + "grad_norm": 2.6067187786102295, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8983148336410522, + "num_tokens": 398787003.0, + "step": 10272 + }, + { + "epoch": 1.3068311919603104, + "grad_norm": 2.359632968902588, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8861557841300964, + "num_tokens": 398836561.0, + "step": 10273 + }, + { + "epoch": 1.306958402238901, + "grad_norm": 2.9249916076660156, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8899054527282715, + "num_tokens": 398872390.0, + "step": 10274 + }, + { + "epoch": 1.3070856125174914, + "grad_norm": 2.4830636978149414, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8974418640136719, + "num_tokens": 398914376.0, + "step": 10275 + }, + { + "epoch": 1.307212822796082, + "grad_norm": 2.6650285720825195, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8799502849578857, + "num_tokens": 398959010.0, + "step": 10276 + }, + { + "epoch": 1.3073400330746725, + "grad_norm": 2.6558780670166016, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8894830942153931, + "num_tokens": 398997861.0, + "step": 10277 + }, + { + "epoch": 1.3074672433532628, + "grad_norm": 2.602487564086914, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.889161229133606, + "num_tokens": 399039559.0, + "step": 10278 + }, + { + "epoch": 1.3075944536318533, + "grad_norm": 2.5888333320617676, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8961877822875977, + "num_tokens": 399079638.0, + "step": 10279 + }, + { + "epoch": 1.3077216639104439, + "grad_norm": 2.5542123317718506, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8874841332435608, + "num_tokens": 399119161.0, + "step": 10280 + }, + { + "epoch": 1.3078488741890344, + "grad_norm": 2.716466188430786, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.898037314414978, + "num_tokens": 399155389.0, + "step": 10281 + }, + { + "epoch": 1.307976084467625, + "grad_norm": 2.8577284812927246, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8790745139122009, + "num_tokens": 399192663.0, + "step": 10282 + }, + { + "epoch": 1.3081032947462155, + "grad_norm": 2.690531015396118, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8877307772636414, + "num_tokens": 399230562.0, + "step": 10283 + }, + { + "epoch": 1.308230505024806, + "grad_norm": 2.576514720916748, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8877395391464233, + "num_tokens": 399271127.0, + "step": 10284 + }, + { + "epoch": 1.3083577153033965, + "grad_norm": 2.620396614074707, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8745254278182983, + "num_tokens": 399314744.0, + "step": 10285 + }, + { + "epoch": 1.308484925581987, + "grad_norm": 2.5117483139038086, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8886365294456482, + "num_tokens": 399354254.0, + "step": 10286 + }, + { + "epoch": 1.3086121358605776, + "grad_norm": 2.800748825073242, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9050913453102112, + "num_tokens": 399386978.0, + "step": 10287 + }, + { + "epoch": 1.308739346139168, + "grad_norm": 2.701951026916504, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8902902603149414, + "num_tokens": 399424696.0, + "step": 10288 + }, + { + "epoch": 1.3088665564177586, + "grad_norm": 2.699604034423828, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.885507345199585, + "num_tokens": 399462566.0, + "step": 10289 + }, + { + "epoch": 1.3089937666963491, + "grad_norm": 2.6403191089630127, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8885201811790466, + "num_tokens": 399499748.0, + "step": 10290 + }, + { + "epoch": 1.3091209769749397, + "grad_norm": 2.711594343185425, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8904464244842529, + "num_tokens": 399537655.0, + "step": 10291 + }, + { + "epoch": 1.3092481872535302, + "grad_norm": 2.8784871101379395, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8801062703132629, + "num_tokens": 399573800.0, + "step": 10292 + }, + { + "epoch": 1.3093753975321207, + "grad_norm": 2.64801025390625, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8875119090080261, + "num_tokens": 399614375.0, + "step": 10293 + }, + { + "epoch": 1.309502607810711, + "grad_norm": 2.694654703140259, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8777393102645874, + "num_tokens": 399656916.0, + "step": 10294 + }, + { + "epoch": 1.3096298180893016, + "grad_norm": 3.0110065937042236, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8865549564361572, + "num_tokens": 399686310.0, + "step": 10295 + }, + { + "epoch": 1.309757028367892, + "grad_norm": 2.6935787200927734, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8939913511276245, + "num_tokens": 399725662.0, + "step": 10296 + }, + { + "epoch": 1.3098842386464826, + "grad_norm": 2.8138484954833984, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8756110668182373, + "num_tokens": 399765957.0, + "step": 10297 + }, + { + "epoch": 1.3100114489250732, + "grad_norm": 2.724393129348755, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8905750513076782, + "num_tokens": 399801867.0, + "step": 10298 + }, + { + "epoch": 1.3101386592036637, + "grad_norm": 2.513326406478882, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8867475986480713, + "num_tokens": 399845002.0, + "step": 10299 + }, + { + "epoch": 1.3102658694822542, + "grad_norm": 2.5202250480651855, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8913683891296387, + "num_tokens": 399886429.0, + "step": 10300 + }, + { + "epoch": 1.3103930797608447, + "grad_norm": 3.1346330642700195, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8746199607849121, + "num_tokens": 399918215.0, + "step": 10301 + }, + { + "epoch": 1.3105202900394353, + "grad_norm": 2.5162699222564697, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8962805271148682, + "num_tokens": 399956238.0, + "step": 10302 + }, + { + "epoch": 1.3106475003180256, + "grad_norm": 2.7237839698791504, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8743289709091187, + "num_tokens": 399997279.0, + "step": 10303 + }, + { + "epoch": 1.310774710596616, + "grad_norm": 2.8920211791992188, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8758779764175415, + "num_tokens": 400035746.0, + "step": 10304 + }, + { + "epoch": 1.3109019208752066, + "grad_norm": 2.4313719272613525, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8896009922027588, + "num_tokens": 400082066.0, + "step": 10305 + }, + { + "epoch": 1.3110291311537972, + "grad_norm": 2.5024783611297607, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8857908844947815, + "num_tokens": 400125097.0, + "step": 10306 + }, + { + "epoch": 1.3111563414323877, + "grad_norm": 2.5280089378356934, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8759772777557373, + "num_tokens": 400169720.0, + "step": 10307 + }, + { + "epoch": 1.3112835517109782, + "grad_norm": 2.666919708251953, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8894705772399902, + "num_tokens": 400207827.0, + "step": 10308 + }, + { + "epoch": 1.3114107619895687, + "grad_norm": 2.9006617069244385, + "learning_rate": 1e-06, + "loss": 0.3497, + "mean_token_accuracy": 0.8785304427146912, + "num_tokens": 400246837.0, + "step": 10309 + }, + { + "epoch": 1.3115379722681593, + "grad_norm": 2.7121994495391846, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8915104866027832, + "num_tokens": 400281810.0, + "step": 10310 + }, + { + "epoch": 1.3116651825467498, + "grad_norm": 2.5531651973724365, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9022066593170166, + "num_tokens": 400318446.0, + "step": 10311 + }, + { + "epoch": 1.3117923928253403, + "grad_norm": 2.802919626235962, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.888332724571228, + "num_tokens": 400351271.0, + "step": 10312 + }, + { + "epoch": 1.3119196031039309, + "grad_norm": 2.660501003265381, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8912428617477417, + "num_tokens": 400390902.0, + "step": 10313 + }, + { + "epoch": 1.3120468133825214, + "grad_norm": 2.2527060508728027, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.902086079120636, + "num_tokens": 400438314.0, + "step": 10314 + }, + { + "epoch": 1.312174023661112, + "grad_norm": 3.054032325744629, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8912439942359924, + "num_tokens": 400473987.0, + "step": 10315 + }, + { + "epoch": 1.3123012339397024, + "grad_norm": 2.741426706314087, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.898098349571228, + "num_tokens": 400508540.0, + "step": 10316 + }, + { + "epoch": 1.312428444218293, + "grad_norm": 2.734717845916748, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8857349157333374, + "num_tokens": 400549828.0, + "step": 10317 + }, + { + "epoch": 1.3125556544968833, + "grad_norm": 2.8443217277526855, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8935966491699219, + "num_tokens": 400582732.0, + "step": 10318 + }, + { + "epoch": 1.3126828647754738, + "grad_norm": 2.6933867931365967, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8936997652053833, + "num_tokens": 400621595.0, + "step": 10319 + }, + { + "epoch": 1.3128100750540643, + "grad_norm": 3.01228928565979, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8869093656539917, + "num_tokens": 400656288.0, + "step": 10320 + }, + { + "epoch": 1.3129372853326549, + "grad_norm": 3.1000537872314453, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8785438537597656, + "num_tokens": 400691896.0, + "step": 10321 + }, + { + "epoch": 1.3130644956112454, + "grad_norm": 2.8833630084991455, + "learning_rate": 1e-06, + "loss": 0.361, + "mean_token_accuracy": 0.8711611032485962, + "num_tokens": 400736544.0, + "step": 10322 + }, + { + "epoch": 1.313191705889836, + "grad_norm": 2.6740429401397705, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8843916654586792, + "num_tokens": 400778056.0, + "step": 10323 + }, + { + "epoch": 1.3133189161684264, + "grad_norm": 2.5050365924835205, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8931089639663696, + "num_tokens": 400821157.0, + "step": 10324 + }, + { + "epoch": 1.313446126447017, + "grad_norm": 2.418999433517456, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.900981068611145, + "num_tokens": 400863699.0, + "step": 10325 + }, + { + "epoch": 1.3135733367256075, + "grad_norm": 2.7243940830230713, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8902933597564697, + "num_tokens": 400904942.0, + "step": 10326 + }, + { + "epoch": 1.3137005470041978, + "grad_norm": 2.6505236625671387, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8996565937995911, + "num_tokens": 400942637.0, + "step": 10327 + }, + { + "epoch": 1.3138277572827883, + "grad_norm": 2.4887633323669434, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8968398571014404, + "num_tokens": 400982074.0, + "step": 10328 + }, + { + "epoch": 1.3139549675613789, + "grad_norm": 2.7131474018096924, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8854358792304993, + "num_tokens": 401023308.0, + "step": 10329 + }, + { + "epoch": 1.3140821778399694, + "grad_norm": 2.863212823867798, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8772695064544678, + "num_tokens": 401063528.0, + "step": 10330 + }, + { + "epoch": 1.31420938811856, + "grad_norm": 2.498363971710205, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8879077434539795, + "num_tokens": 401107570.0, + "step": 10331 + }, + { + "epoch": 1.3143365983971504, + "grad_norm": 2.5486137866973877, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8901311755180359, + "num_tokens": 401150583.0, + "step": 10332 + }, + { + "epoch": 1.314463808675741, + "grad_norm": 2.479548454284668, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9008809328079224, + "num_tokens": 401190355.0, + "step": 10333 + }, + { + "epoch": 1.3145910189543315, + "grad_norm": 2.526764392852783, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9016733169555664, + "num_tokens": 401229602.0, + "step": 10334 + }, + { + "epoch": 1.314718229232922, + "grad_norm": 2.535212755203247, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8894115686416626, + "num_tokens": 401268927.0, + "step": 10335 + }, + { + "epoch": 1.3148454395115126, + "grad_norm": 2.565739154815674, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.8994433879852295, + "num_tokens": 401307443.0, + "step": 10336 + }, + { + "epoch": 1.314972649790103, + "grad_norm": 3.2074086666107178, + "learning_rate": 1e-06, + "loss": 0.3856, + "mean_token_accuracy": 0.8650447726249695, + "num_tokens": 401344225.0, + "step": 10337 + }, + { + "epoch": 1.3150998600686936, + "grad_norm": 2.7817885875701904, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8750723600387573, + "num_tokens": 401385276.0, + "step": 10338 + }, + { + "epoch": 1.3152270703472841, + "grad_norm": 2.6338963508605957, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8933192491531372, + "num_tokens": 401425549.0, + "step": 10339 + }, + { + "epoch": 1.3153542806258747, + "grad_norm": 2.6911439895629883, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8897504210472107, + "num_tokens": 401469390.0, + "step": 10340 + }, + { + "epoch": 1.3154814909044652, + "grad_norm": 2.5813040733337402, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8917586207389832, + "num_tokens": 401507518.0, + "step": 10341 + }, + { + "epoch": 1.3156087011830557, + "grad_norm": 2.7860546112060547, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8830560445785522, + "num_tokens": 401548154.0, + "step": 10342 + }, + { + "epoch": 1.315735911461646, + "grad_norm": 2.520099639892578, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8852533102035522, + "num_tokens": 401593987.0, + "step": 10343 + }, + { + "epoch": 1.3158631217402366, + "grad_norm": 2.7087559700012207, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8951627016067505, + "num_tokens": 401630586.0, + "step": 10344 + }, + { + "epoch": 1.315990332018827, + "grad_norm": 2.5505075454711914, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8943759202957153, + "num_tokens": 401668047.0, + "step": 10345 + }, + { + "epoch": 1.3161175422974176, + "grad_norm": 2.5740721225738525, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9039389491081238, + "num_tokens": 401704322.0, + "step": 10346 + }, + { + "epoch": 1.3162447525760081, + "grad_norm": 2.4001739025115967, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.8997646570205688, + "num_tokens": 401743977.0, + "step": 10347 + }, + { + "epoch": 1.3163719628545987, + "grad_norm": 2.738649845123291, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8920692205429077, + "num_tokens": 401776184.0, + "step": 10348 + }, + { + "epoch": 1.3164991731331892, + "grad_norm": 2.5237624645233154, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8860669732093811, + "num_tokens": 401819704.0, + "step": 10349 + }, + { + "epoch": 1.3166263834117797, + "grad_norm": 2.8064262866973877, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8869293928146362, + "num_tokens": 401855098.0, + "step": 10350 + }, + { + "epoch": 1.3167535936903703, + "grad_norm": 3.079719066619873, + "learning_rate": 1e-06, + "loss": 0.3754, + "mean_token_accuracy": 0.867403507232666, + "num_tokens": 401888485.0, + "step": 10351 + }, + { + "epoch": 1.3168808039689606, + "grad_norm": 2.999337911605835, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8818262815475464, + "num_tokens": 401923295.0, + "step": 10352 + }, + { + "epoch": 1.317008014247551, + "grad_norm": 2.8236141204833984, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8879470825195312, + "num_tokens": 401958938.0, + "step": 10353 + }, + { + "epoch": 1.3171352245261416, + "grad_norm": 2.543144702911377, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8908044695854187, + "num_tokens": 402001356.0, + "step": 10354 + }, + { + "epoch": 1.3172624348047322, + "grad_norm": 2.6845476627349854, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8882321119308472, + "num_tokens": 402040306.0, + "step": 10355 + }, + { + "epoch": 1.3173896450833227, + "grad_norm": 2.4599533081054688, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9023945331573486, + "num_tokens": 402078756.0, + "step": 10356 + }, + { + "epoch": 1.3175168553619132, + "grad_norm": 2.475886106491089, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8958886861801147, + "num_tokens": 402117705.0, + "step": 10357 + }, + { + "epoch": 1.3176440656405037, + "grad_norm": 2.649453639984131, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8893135786056519, + "num_tokens": 402159243.0, + "step": 10358 + }, + { + "epoch": 1.3177712759190943, + "grad_norm": 2.647268056869507, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8822882175445557, + "num_tokens": 402199883.0, + "step": 10359 + }, + { + "epoch": 1.3178984861976848, + "grad_norm": 2.836637020111084, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8884856700897217, + "num_tokens": 402233249.0, + "step": 10360 + }, + { + "epoch": 1.3180256964762753, + "grad_norm": 2.8418045043945312, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8919504880905151, + "num_tokens": 402268997.0, + "step": 10361 + }, + { + "epoch": 1.3181529067548658, + "grad_norm": 2.589698314666748, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8932397365570068, + "num_tokens": 402311869.0, + "step": 10362 + }, + { + "epoch": 1.3182801170334564, + "grad_norm": 2.7482857704162598, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.897983968257904, + "num_tokens": 402347087.0, + "step": 10363 + }, + { + "epoch": 1.318407327312047, + "grad_norm": 3.1034812927246094, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8803442716598511, + "num_tokens": 402381095.0, + "step": 10364 + }, + { + "epoch": 1.3185345375906374, + "grad_norm": 2.826113700866699, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8955404758453369, + "num_tokens": 402413005.0, + "step": 10365 + }, + { + "epoch": 1.318661747869228, + "grad_norm": 2.4317872524261475, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8974679112434387, + "num_tokens": 402454044.0, + "step": 10366 + }, + { + "epoch": 1.3187889581478183, + "grad_norm": 2.584228992462158, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8834449648857117, + "num_tokens": 402496059.0, + "step": 10367 + }, + { + "epoch": 1.3189161684264088, + "grad_norm": 2.4673960208892822, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8955667018890381, + "num_tokens": 402535906.0, + "step": 10368 + }, + { + "epoch": 1.3190433787049993, + "grad_norm": 2.5081663131713867, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8937365412712097, + "num_tokens": 402581385.0, + "step": 10369 + }, + { + "epoch": 1.3191705889835899, + "grad_norm": 2.8251216411590576, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8889769315719604, + "num_tokens": 402617147.0, + "step": 10370 + }, + { + "epoch": 1.3192977992621804, + "grad_norm": 2.710735321044922, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8954963684082031, + "num_tokens": 402654356.0, + "step": 10371 + }, + { + "epoch": 1.319425009540771, + "grad_norm": 2.6596803665161133, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8935562968254089, + "num_tokens": 402692822.0, + "step": 10372 + }, + { + "epoch": 1.3195522198193614, + "grad_norm": 2.748530864715576, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8998367786407471, + "num_tokens": 402728566.0, + "step": 10373 + }, + { + "epoch": 1.319679430097952, + "grad_norm": 2.5187132358551025, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.9001429677009583, + "num_tokens": 402767054.0, + "step": 10374 + }, + { + "epoch": 1.3198066403765425, + "grad_norm": 2.7497503757476807, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8861932754516602, + "num_tokens": 402805513.0, + "step": 10375 + }, + { + "epoch": 1.3199338506551328, + "grad_norm": 2.6892950534820557, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8950223922729492, + "num_tokens": 402841185.0, + "step": 10376 + }, + { + "epoch": 1.3200610609337233, + "grad_norm": 2.7649755477905273, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.888201117515564, + "num_tokens": 402878417.0, + "step": 10377 + }, + { + "epoch": 1.3201882712123139, + "grad_norm": 2.840381145477295, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8844084739685059, + "num_tokens": 402916378.0, + "step": 10378 + }, + { + "epoch": 1.3203154814909044, + "grad_norm": 2.5418097972869873, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8942068219184875, + "num_tokens": 402956594.0, + "step": 10379 + }, + { + "epoch": 1.320442691769495, + "grad_norm": 2.5947513580322266, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8969055414199829, + "num_tokens": 402994112.0, + "step": 10380 + }, + { + "epoch": 1.3205699020480854, + "grad_norm": 2.4622228145599365, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8933281898498535, + "num_tokens": 403036298.0, + "step": 10381 + }, + { + "epoch": 1.320697112326676, + "grad_norm": 2.4144906997680664, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8950690627098083, + "num_tokens": 403079719.0, + "step": 10382 + }, + { + "epoch": 1.3208243226052665, + "grad_norm": 2.7557461261749268, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8857735991477966, + "num_tokens": 403118576.0, + "step": 10383 + }, + { + "epoch": 1.320951532883857, + "grad_norm": 2.87606143951416, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8772820234298706, + "num_tokens": 403157440.0, + "step": 10384 + }, + { + "epoch": 1.3210787431624476, + "grad_norm": 2.7132866382598877, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8968888521194458, + "num_tokens": 403192849.0, + "step": 10385 + }, + { + "epoch": 1.321205953441038, + "grad_norm": 2.761038303375244, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8929548859596252, + "num_tokens": 403226504.0, + "step": 10386 + }, + { + "epoch": 1.3213331637196286, + "grad_norm": 2.471998453140259, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9004168510437012, + "num_tokens": 403267671.0, + "step": 10387 + }, + { + "epoch": 1.3214603739982191, + "grad_norm": 3.127318859100342, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8859729766845703, + "num_tokens": 403297909.0, + "step": 10388 + }, + { + "epoch": 1.3215875842768097, + "grad_norm": 2.6613659858703613, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8946185111999512, + "num_tokens": 403334342.0, + "step": 10389 + }, + { + "epoch": 1.3217147945554002, + "grad_norm": 2.5593700408935547, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8961091637611389, + "num_tokens": 403374968.0, + "step": 10390 + }, + { + "epoch": 1.3218420048339907, + "grad_norm": 2.533144235610962, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.9013096690177917, + "num_tokens": 403414545.0, + "step": 10391 + }, + { + "epoch": 1.321969215112581, + "grad_norm": 2.6747395992279053, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8867717981338501, + "num_tokens": 403454026.0, + "step": 10392 + }, + { + "epoch": 1.3220964253911716, + "grad_norm": 2.7386345863342285, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8921495676040649, + "num_tokens": 403485533.0, + "step": 10393 + }, + { + "epoch": 1.322223635669762, + "grad_norm": 2.9911692142486572, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8877462148666382, + "num_tokens": 403518781.0, + "step": 10394 + }, + { + "epoch": 1.3223508459483526, + "grad_norm": 2.660139560699463, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.9007396101951599, + "num_tokens": 403555969.0, + "step": 10395 + }, + { + "epoch": 1.3224780562269431, + "grad_norm": 2.49776291847229, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.8985956907272339, + "num_tokens": 403595085.0, + "step": 10396 + }, + { + "epoch": 1.3226052665055337, + "grad_norm": 2.8765673637390137, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8896419405937195, + "num_tokens": 403629132.0, + "step": 10397 + }, + { + "epoch": 1.3227324767841242, + "grad_norm": 2.721083402633667, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8838216066360474, + "num_tokens": 403666070.0, + "step": 10398 + }, + { + "epoch": 1.3228596870627147, + "grad_norm": 2.6336495876312256, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8977077007293701, + "num_tokens": 403701680.0, + "step": 10399 + }, + { + "epoch": 1.3229868973413053, + "grad_norm": 2.671654224395752, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8931031227111816, + "num_tokens": 403740836.0, + "step": 10400 + }, + { + "epoch": 1.3231141076198956, + "grad_norm": 3.0251238346099854, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8845351934432983, + "num_tokens": 403774042.0, + "step": 10401 + }, + { + "epoch": 1.323241317898486, + "grad_norm": 2.5655176639556885, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9021375775337219, + "num_tokens": 403810683.0, + "step": 10402 + }, + { + "epoch": 1.3233685281770766, + "grad_norm": 2.642197847366333, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.888727068901062, + "num_tokens": 403852691.0, + "step": 10403 + }, + { + "epoch": 1.3234957384556671, + "grad_norm": 2.698667287826538, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8860447406768799, + "num_tokens": 403893667.0, + "step": 10404 + }, + { + "epoch": 1.3236229487342577, + "grad_norm": 2.6781599521636963, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8980576992034912, + "num_tokens": 403928627.0, + "step": 10405 + }, + { + "epoch": 1.3237501590128482, + "grad_norm": 2.6337382793426514, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8935519456863403, + "num_tokens": 403968349.0, + "step": 10406 + }, + { + "epoch": 1.3238773692914387, + "grad_norm": 2.6029398441314697, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8903152942657471, + "num_tokens": 404009659.0, + "step": 10407 + }, + { + "epoch": 1.3240045795700293, + "grad_norm": 2.4471685886383057, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.898094117641449, + "num_tokens": 404053688.0, + "step": 10408 + }, + { + "epoch": 1.3241317898486198, + "grad_norm": 2.4627130031585693, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8924800157546997, + "num_tokens": 404096271.0, + "step": 10409 + }, + { + "epoch": 1.3242590001272103, + "grad_norm": 2.472132921218872, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8984955549240112, + "num_tokens": 404138482.0, + "step": 10410 + }, + { + "epoch": 1.3243862104058008, + "grad_norm": 2.613780975341797, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8909889459609985, + "num_tokens": 404180458.0, + "step": 10411 + }, + { + "epoch": 1.3245134206843914, + "grad_norm": 2.611934185028076, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8842395544052124, + "num_tokens": 404217429.0, + "step": 10412 + }, + { + "epoch": 1.324640630962982, + "grad_norm": 2.5432140827178955, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.902614951133728, + "num_tokens": 404252801.0, + "step": 10413 + }, + { + "epoch": 1.3247678412415724, + "grad_norm": 2.5229687690734863, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8891211748123169, + "num_tokens": 404294895.0, + "step": 10414 + }, + { + "epoch": 1.324895051520163, + "grad_norm": 2.446192502975464, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9005997776985168, + "num_tokens": 404334492.0, + "step": 10415 + }, + { + "epoch": 1.3250222617987533, + "grad_norm": 2.6781678199768066, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8913556933403015, + "num_tokens": 404372899.0, + "step": 10416 + }, + { + "epoch": 1.3251494720773438, + "grad_norm": 2.5203046798706055, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9003045558929443, + "num_tokens": 404413224.0, + "step": 10417 + }, + { + "epoch": 1.3252766823559343, + "grad_norm": 2.6821823120117188, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8861821889877319, + "num_tokens": 404453247.0, + "step": 10418 + }, + { + "epoch": 1.3254038926345248, + "grad_norm": 2.6159181594848633, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9060080051422119, + "num_tokens": 404487341.0, + "step": 10419 + }, + { + "epoch": 1.3255311029131154, + "grad_norm": 2.722627878189087, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8814525604248047, + "num_tokens": 404528756.0, + "step": 10420 + }, + { + "epoch": 1.325658313191706, + "grad_norm": 2.7061431407928467, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9033197164535522, + "num_tokens": 404564105.0, + "step": 10421 + }, + { + "epoch": 1.3257855234702964, + "grad_norm": 2.730574607849121, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8794069290161133, + "num_tokens": 404605774.0, + "step": 10422 + }, + { + "epoch": 1.325912733748887, + "grad_norm": 2.496267557144165, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9129072427749634, + "num_tokens": 404642200.0, + "step": 10423 + }, + { + "epoch": 1.3260399440274775, + "grad_norm": 3.0221657752990723, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8850204348564148, + "num_tokens": 404674877.0, + "step": 10424 + }, + { + "epoch": 1.3261671543060678, + "grad_norm": 2.505732536315918, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.8987979888916016, + "num_tokens": 404713057.0, + "step": 10425 + }, + { + "epoch": 1.3262943645846583, + "grad_norm": 2.7275643348693848, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8803372383117676, + "num_tokens": 404755972.0, + "step": 10426 + }, + { + "epoch": 1.3264215748632489, + "grad_norm": 2.4955241680145264, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8923813700675964, + "num_tokens": 404797432.0, + "step": 10427 + }, + { + "epoch": 1.3265487851418394, + "grad_norm": 2.5082480907440186, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8843625783920288, + "num_tokens": 404840810.0, + "step": 10428 + }, + { + "epoch": 1.32667599542043, + "grad_norm": 2.469177007675171, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.891048789024353, + "num_tokens": 404881500.0, + "step": 10429 + }, + { + "epoch": 1.3268032056990204, + "grad_norm": 2.704503059387207, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8893769979476929, + "num_tokens": 404921080.0, + "step": 10430 + }, + { + "epoch": 1.326930415977611, + "grad_norm": 2.556398630142212, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8965516090393066, + "num_tokens": 404958469.0, + "step": 10431 + }, + { + "epoch": 1.3270576262562015, + "grad_norm": 2.781873941421509, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8897237777709961, + "num_tokens": 404992723.0, + "step": 10432 + }, + { + "epoch": 1.327184836534792, + "grad_norm": 2.6685707569122314, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8889902234077454, + "num_tokens": 405032852.0, + "step": 10433 + }, + { + "epoch": 1.3273120468133826, + "grad_norm": 2.6894032955169678, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8873063325881958, + "num_tokens": 405071934.0, + "step": 10434 + }, + { + "epoch": 1.327439257091973, + "grad_norm": 2.862658739089966, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8739734888076782, + "num_tokens": 405110681.0, + "step": 10435 + }, + { + "epoch": 1.3275664673705636, + "grad_norm": 2.675196886062622, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8880297541618347, + "num_tokens": 405149687.0, + "step": 10436 + }, + { + "epoch": 1.3276936776491541, + "grad_norm": 2.53393292427063, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8844337463378906, + "num_tokens": 405191937.0, + "step": 10437 + }, + { + "epoch": 1.3278208879277447, + "grad_norm": 2.6010076999664307, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8937004804611206, + "num_tokens": 405232278.0, + "step": 10438 + }, + { + "epoch": 1.3279480982063352, + "grad_norm": 2.8071985244750977, + "learning_rate": 1e-06, + "loss": 0.3383, + "mean_token_accuracy": 0.8823176026344299, + "num_tokens": 405269748.0, + "step": 10439 + }, + { + "epoch": 1.3280753084849257, + "grad_norm": 2.9476656913757324, + "learning_rate": 1e-06, + "loss": 0.3473, + "mean_token_accuracy": 0.8759392499923706, + "num_tokens": 405306775.0, + "step": 10440 + }, + { + "epoch": 1.328202518763516, + "grad_norm": 2.6703739166259766, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8933438658714294, + "num_tokens": 405342284.0, + "step": 10441 + }, + { + "epoch": 1.3283297290421066, + "grad_norm": 2.777846574783325, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8991101980209351, + "num_tokens": 405378368.0, + "step": 10442 + }, + { + "epoch": 1.328456939320697, + "grad_norm": 2.5539464950561523, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8948541879653931, + "num_tokens": 405417349.0, + "step": 10443 + }, + { + "epoch": 1.3285841495992876, + "grad_norm": 2.586744546890259, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8928990364074707, + "num_tokens": 405455931.0, + "step": 10444 + }, + { + "epoch": 1.3287113598778781, + "grad_norm": 2.284001350402832, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9049049615859985, + "num_tokens": 405497094.0, + "step": 10445 + }, + { + "epoch": 1.3288385701564687, + "grad_norm": 2.629765033721924, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8855857849121094, + "num_tokens": 405535434.0, + "step": 10446 + }, + { + "epoch": 1.3289657804350592, + "grad_norm": 2.6068670749664307, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8790475726127625, + "num_tokens": 405581219.0, + "step": 10447 + }, + { + "epoch": 1.3290929907136497, + "grad_norm": 2.8584892749786377, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8857939839363098, + "num_tokens": 405619560.0, + "step": 10448 + }, + { + "epoch": 1.3292202009922403, + "grad_norm": 2.3458168506622314, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9054104089736938, + "num_tokens": 405660490.0, + "step": 10449 + }, + { + "epoch": 1.3293474112708306, + "grad_norm": 2.674358367919922, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8948124051094055, + "num_tokens": 405698349.0, + "step": 10450 + }, + { + "epoch": 1.329474621549421, + "grad_norm": 2.8739876747131348, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8925337791442871, + "num_tokens": 405731043.0, + "step": 10451 + }, + { + "epoch": 1.3296018318280116, + "grad_norm": 2.537472724914551, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8861908912658691, + "num_tokens": 405773548.0, + "step": 10452 + }, + { + "epoch": 1.3297290421066021, + "grad_norm": 2.6369588375091553, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8944779634475708, + "num_tokens": 405812316.0, + "step": 10453 + }, + { + "epoch": 1.3298562523851927, + "grad_norm": 2.926348924636841, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8947852849960327, + "num_tokens": 405848703.0, + "step": 10454 + }, + { + "epoch": 1.3299834626637832, + "grad_norm": 2.629970073699951, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8917363882064819, + "num_tokens": 405886783.0, + "step": 10455 + }, + { + "epoch": 1.3301106729423737, + "grad_norm": 2.9579062461853027, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8854491710662842, + "num_tokens": 405920008.0, + "step": 10456 + }, + { + "epoch": 1.3302378832209643, + "grad_norm": 2.7272164821624756, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8910143971443176, + "num_tokens": 405956752.0, + "step": 10457 + }, + { + "epoch": 1.3303650934995548, + "grad_norm": 2.825308084487915, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8954263925552368, + "num_tokens": 405990825.0, + "step": 10458 + }, + { + "epoch": 1.3304923037781453, + "grad_norm": 2.7258756160736084, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8824654817581177, + "num_tokens": 406030562.0, + "step": 10459 + }, + { + "epoch": 1.3306195140567358, + "grad_norm": 2.674651861190796, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9021514058113098, + "num_tokens": 406066757.0, + "step": 10460 + }, + { + "epoch": 1.3307467243353264, + "grad_norm": 2.734508752822876, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8901911973953247, + "num_tokens": 406105772.0, + "step": 10461 + }, + { + "epoch": 1.330873934613917, + "grad_norm": 2.781238079071045, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8970946073532104, + "num_tokens": 406139420.0, + "step": 10462 + }, + { + "epoch": 1.3310011448925074, + "grad_norm": 2.5851588249206543, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8925105333328247, + "num_tokens": 406177305.0, + "step": 10463 + }, + { + "epoch": 1.331128355171098, + "grad_norm": 2.426708459854126, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8980087041854858, + "num_tokens": 406221090.0, + "step": 10464 + }, + { + "epoch": 1.3312555654496883, + "grad_norm": 2.8520560264587402, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8894743919372559, + "num_tokens": 406257531.0, + "step": 10465 + }, + { + "epoch": 1.3313827757282788, + "grad_norm": 2.5002427101135254, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8968778848648071, + "num_tokens": 406298671.0, + "step": 10466 + }, + { + "epoch": 1.3315099860068693, + "grad_norm": 2.536651849746704, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8905568718910217, + "num_tokens": 406340148.0, + "step": 10467 + }, + { + "epoch": 1.3316371962854598, + "grad_norm": 2.59635853767395, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8855524063110352, + "num_tokens": 406383264.0, + "step": 10468 + }, + { + "epoch": 1.3317644065640504, + "grad_norm": 2.637479305267334, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.880428671836853, + "num_tokens": 406422864.0, + "step": 10469 + }, + { + "epoch": 1.331891616842641, + "grad_norm": 2.613790988922119, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8807233572006226, + "num_tokens": 406464381.0, + "step": 10470 + }, + { + "epoch": 1.3320188271212314, + "grad_norm": 2.381683349609375, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8963174819946289, + "num_tokens": 406505686.0, + "step": 10471 + }, + { + "epoch": 1.332146037399822, + "grad_norm": 2.473093032836914, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.883669376373291, + "num_tokens": 406550884.0, + "step": 10472 + }, + { + "epoch": 1.3322732476784125, + "grad_norm": 2.4689669609069824, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8967591524124146, + "num_tokens": 406592867.0, + "step": 10473 + }, + { + "epoch": 1.3324004579570028, + "grad_norm": 2.5228233337402344, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8996945023536682, + "num_tokens": 406629574.0, + "step": 10474 + }, + { + "epoch": 1.3325276682355933, + "grad_norm": 2.6280972957611084, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8909333944320679, + "num_tokens": 406668988.0, + "step": 10475 + }, + { + "epoch": 1.3326548785141838, + "grad_norm": 2.479987144470215, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8946503400802612, + "num_tokens": 406715635.0, + "step": 10476 + }, + { + "epoch": 1.3327820887927744, + "grad_norm": 2.6939034461975098, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8720378875732422, + "num_tokens": 406756897.0, + "step": 10477 + }, + { + "epoch": 1.332909299071365, + "grad_norm": 2.9943182468414307, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8845518827438354, + "num_tokens": 406786213.0, + "step": 10478 + }, + { + "epoch": 1.3330365093499554, + "grad_norm": 2.6815104484558105, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8943262696266174, + "num_tokens": 406822173.0, + "step": 10479 + }, + { + "epoch": 1.333163719628546, + "grad_norm": 2.4242491722106934, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9024296998977661, + "num_tokens": 406861654.0, + "step": 10480 + }, + { + "epoch": 1.3332909299071365, + "grad_norm": 2.372701406478882, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8910901546478271, + "num_tokens": 406907907.0, + "step": 10481 + }, + { + "epoch": 1.333418140185727, + "grad_norm": 2.4807140827178955, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8999673128128052, + "num_tokens": 406946406.0, + "step": 10482 + }, + { + "epoch": 1.3335453504643175, + "grad_norm": 3.490840196609497, + "learning_rate": 1e-06, + "loss": 0.4004, + "mean_token_accuracy": 0.8585423231124878, + "num_tokens": 406979974.0, + "step": 10483 + }, + { + "epoch": 1.333672560742908, + "grad_norm": 2.3917396068573, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8983288407325745, + "num_tokens": 407021368.0, + "step": 10484 + }, + { + "epoch": 1.3337997710214986, + "grad_norm": 2.584266185760498, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8885528445243835, + "num_tokens": 407066584.0, + "step": 10485 + }, + { + "epoch": 1.3339269813000891, + "grad_norm": 2.764036178588867, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.8841335773468018, + "num_tokens": 407104870.0, + "step": 10486 + }, + { + "epoch": 1.3340541915786797, + "grad_norm": 2.7226343154907227, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8809115290641785, + "num_tokens": 407144337.0, + "step": 10487 + }, + { + "epoch": 1.3341814018572702, + "grad_norm": 2.711834192276001, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8940163254737854, + "num_tokens": 407181181.0, + "step": 10488 + }, + { + "epoch": 1.3343086121358605, + "grad_norm": 2.7417056560516357, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8815295696258545, + "num_tokens": 407224624.0, + "step": 10489 + }, + { + "epoch": 1.334435822414451, + "grad_norm": 2.416203022003174, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8878365755081177, + "num_tokens": 407270194.0, + "step": 10490 + }, + { + "epoch": 1.3345630326930416, + "grad_norm": 2.784663200378418, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8928974866867065, + "num_tokens": 407310001.0, + "step": 10491 + }, + { + "epoch": 1.334690242971632, + "grad_norm": 2.8904664516448975, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8832917213439941, + "num_tokens": 407343456.0, + "step": 10492 + }, + { + "epoch": 1.3348174532502226, + "grad_norm": 2.6786861419677734, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8808756470680237, + "num_tokens": 407383947.0, + "step": 10493 + }, + { + "epoch": 1.3349446635288131, + "grad_norm": 2.5361368656158447, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8861116170883179, + "num_tokens": 407427876.0, + "step": 10494 + }, + { + "epoch": 1.3350718738074037, + "grad_norm": 2.6095919609069824, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.8761948347091675, + "num_tokens": 407472446.0, + "step": 10495 + }, + { + "epoch": 1.3351990840859942, + "grad_norm": 2.7742135524749756, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8912965059280396, + "num_tokens": 407509191.0, + "step": 10496 + }, + { + "epoch": 1.3353262943645847, + "grad_norm": 2.598381519317627, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8843669891357422, + "num_tokens": 407549777.0, + "step": 10497 + }, + { + "epoch": 1.3354535046431752, + "grad_norm": 2.730527400970459, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8838515281677246, + "num_tokens": 407588616.0, + "step": 10498 + }, + { + "epoch": 1.3355807149217656, + "grad_norm": 2.669275999069214, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8798518180847168, + "num_tokens": 407630236.0, + "step": 10499 + }, + { + "epoch": 1.335707925200356, + "grad_norm": 2.8331425189971924, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8988988399505615, + "num_tokens": 407662273.0, + "step": 10500 + }, + { + "epoch": 1.3358351354789466, + "grad_norm": 2.744685411453247, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8916207551956177, + "num_tokens": 407699176.0, + "step": 10501 + }, + { + "epoch": 1.3359623457575371, + "grad_norm": 2.5314228534698486, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8885222673416138, + "num_tokens": 407742721.0, + "step": 10502 + }, + { + "epoch": 1.3360895560361277, + "grad_norm": 2.535125970840454, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8871347308158875, + "num_tokens": 407786483.0, + "step": 10503 + }, + { + "epoch": 1.3362167663147182, + "grad_norm": 2.7314236164093018, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8946880102157593, + "num_tokens": 407820404.0, + "step": 10504 + }, + { + "epoch": 1.3363439765933087, + "grad_norm": 2.8409183025360107, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8887491822242737, + "num_tokens": 407852040.0, + "step": 10505 + }, + { + "epoch": 1.3364711868718993, + "grad_norm": 2.4755492210388184, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8938471674919128, + "num_tokens": 407893754.0, + "step": 10506 + }, + { + "epoch": 1.3365983971504898, + "grad_norm": 2.681001663208008, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8837579488754272, + "num_tokens": 407933102.0, + "step": 10507 + }, + { + "epoch": 1.3367256074290803, + "grad_norm": 2.5903990268707275, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9036633968353271, + "num_tokens": 407970677.0, + "step": 10508 + }, + { + "epoch": 1.3368528177076708, + "grad_norm": 2.5315542221069336, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8895695209503174, + "num_tokens": 408009818.0, + "step": 10509 + }, + { + "epoch": 1.3369800279862614, + "grad_norm": 2.7252426147460938, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8977901935577393, + "num_tokens": 408046145.0, + "step": 10510 + }, + { + "epoch": 1.337107238264852, + "grad_norm": 2.6326403617858887, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.888239324092865, + "num_tokens": 408085197.0, + "step": 10511 + }, + { + "epoch": 1.3372344485434424, + "grad_norm": 2.573455572128296, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8982158899307251, + "num_tokens": 408127601.0, + "step": 10512 + }, + { + "epoch": 1.337361658822033, + "grad_norm": 2.4612436294555664, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9073860049247742, + "num_tokens": 408168555.0, + "step": 10513 + }, + { + "epoch": 1.3374888691006233, + "grad_norm": 2.831726551055908, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8839845657348633, + "num_tokens": 408204732.0, + "step": 10514 + }, + { + "epoch": 1.3376160793792138, + "grad_norm": 2.5432381629943848, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8934841156005859, + "num_tokens": 408245679.0, + "step": 10515 + }, + { + "epoch": 1.3377432896578043, + "grad_norm": 2.6914665699005127, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8790180087089539, + "num_tokens": 408290597.0, + "step": 10516 + }, + { + "epoch": 1.3378704999363948, + "grad_norm": 2.744663715362549, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8820509314537048, + "num_tokens": 408328943.0, + "step": 10517 + }, + { + "epoch": 1.3379977102149854, + "grad_norm": 2.698667526245117, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.879784345626831, + "num_tokens": 408370936.0, + "step": 10518 + }, + { + "epoch": 1.338124920493576, + "grad_norm": 2.539174795150757, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8980783820152283, + "num_tokens": 408408787.0, + "step": 10519 + }, + { + "epoch": 1.3382521307721664, + "grad_norm": 2.604703664779663, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8756237626075745, + "num_tokens": 408453443.0, + "step": 10520 + }, + { + "epoch": 1.338379341050757, + "grad_norm": 2.8384947776794434, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8874104022979736, + "num_tokens": 408490981.0, + "step": 10521 + }, + { + "epoch": 1.3385065513293475, + "grad_norm": 2.9476122856140137, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.887049674987793, + "num_tokens": 408523004.0, + "step": 10522 + }, + { + "epoch": 1.3386337616079378, + "grad_norm": 2.5237646102905273, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8964514136314392, + "num_tokens": 408563506.0, + "step": 10523 + }, + { + "epoch": 1.3387609718865283, + "grad_norm": 2.613006114959717, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8879744410514832, + "num_tokens": 408603490.0, + "step": 10524 + }, + { + "epoch": 1.3388881821651188, + "grad_norm": 2.5877745151519775, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8947057723999023, + "num_tokens": 408643370.0, + "step": 10525 + }, + { + "epoch": 1.3390153924437094, + "grad_norm": 2.535508632659912, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8872905969619751, + "num_tokens": 408685579.0, + "step": 10526 + }, + { + "epoch": 1.3391426027223, + "grad_norm": 2.5290112495422363, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9006646871566772, + "num_tokens": 408723329.0, + "step": 10527 + }, + { + "epoch": 1.3392698130008904, + "grad_norm": 2.672811269760132, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.890403687953949, + "num_tokens": 408758663.0, + "step": 10528 + }, + { + "epoch": 1.339397023279481, + "grad_norm": 2.7601211071014404, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8855265378952026, + "num_tokens": 408794098.0, + "step": 10529 + }, + { + "epoch": 1.3395242335580715, + "grad_norm": 2.393399953842163, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8903226852416992, + "num_tokens": 408839592.0, + "step": 10530 + }, + { + "epoch": 1.339651443836662, + "grad_norm": 2.6540255546569824, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9027875661849976, + "num_tokens": 408872291.0, + "step": 10531 + }, + { + "epoch": 1.3397786541152525, + "grad_norm": 2.4837725162506104, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9013872146606445, + "num_tokens": 408914924.0, + "step": 10532 + }, + { + "epoch": 1.339905864393843, + "grad_norm": 2.586841106414795, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8995699882507324, + "num_tokens": 408951606.0, + "step": 10533 + }, + { + "epoch": 1.3400330746724336, + "grad_norm": 2.6464385986328125, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8946129083633423, + "num_tokens": 408987642.0, + "step": 10534 + }, + { + "epoch": 1.3401602849510241, + "grad_norm": 2.501420259475708, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8986700773239136, + "num_tokens": 409029341.0, + "step": 10535 + }, + { + "epoch": 1.3402874952296147, + "grad_norm": 2.829702854156494, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8903834819793701, + "num_tokens": 409062671.0, + "step": 10536 + }, + { + "epoch": 1.3404147055082052, + "grad_norm": 2.6604690551757812, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8935743570327759, + "num_tokens": 409109863.0, + "step": 10537 + }, + { + "epoch": 1.3405419157867955, + "grad_norm": 2.4992709159851074, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8916136026382446, + "num_tokens": 409155843.0, + "step": 10538 + }, + { + "epoch": 1.340669126065386, + "grad_norm": 2.722447156906128, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8855127096176147, + "num_tokens": 409198240.0, + "step": 10539 + }, + { + "epoch": 1.3407963363439765, + "grad_norm": 2.8641974925994873, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8879586458206177, + "num_tokens": 409233280.0, + "step": 10540 + }, + { + "epoch": 1.340923546622567, + "grad_norm": 2.7157411575317383, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8847270011901855, + "num_tokens": 409273230.0, + "step": 10541 + }, + { + "epoch": 1.3410507569011576, + "grad_norm": 2.703951120376587, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8932076096534729, + "num_tokens": 409306153.0, + "step": 10542 + }, + { + "epoch": 1.3411779671797481, + "grad_norm": 2.8453142642974854, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8880942463874817, + "num_tokens": 409340750.0, + "step": 10543 + }, + { + "epoch": 1.3413051774583387, + "grad_norm": 2.833468437194824, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8835309743881226, + "num_tokens": 409379506.0, + "step": 10544 + }, + { + "epoch": 1.3414323877369292, + "grad_norm": 2.990251064300537, + "learning_rate": 1e-06, + "loss": 0.3643, + "mean_token_accuracy": 0.869813084602356, + "num_tokens": 409415766.0, + "step": 10545 + }, + { + "epoch": 1.3415595980155197, + "grad_norm": 2.794239044189453, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8794365525245667, + "num_tokens": 409454210.0, + "step": 10546 + }, + { + "epoch": 1.3416868082941102, + "grad_norm": 2.6393020153045654, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8807204961776733, + "num_tokens": 409495502.0, + "step": 10547 + }, + { + "epoch": 1.3418140185727006, + "grad_norm": 2.76863694190979, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.891609787940979, + "num_tokens": 409529860.0, + "step": 10548 + }, + { + "epoch": 1.341941228851291, + "grad_norm": 2.5419702529907227, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8965000510215759, + "num_tokens": 409571338.0, + "step": 10549 + }, + { + "epoch": 1.3420684391298816, + "grad_norm": 2.7297582626342773, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.893954873085022, + "num_tokens": 409605872.0, + "step": 10550 + }, + { + "epoch": 1.3421956494084721, + "grad_norm": 2.716078758239746, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8902658820152283, + "num_tokens": 409644398.0, + "step": 10551 + }, + { + "epoch": 1.3423228596870627, + "grad_norm": 2.557349443435669, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8895177245140076, + "num_tokens": 409688079.0, + "step": 10552 + }, + { + "epoch": 1.3424500699656532, + "grad_norm": 2.8879566192626953, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8869202136993408, + "num_tokens": 409722297.0, + "step": 10553 + }, + { + "epoch": 1.3425772802442437, + "grad_norm": 2.47568416595459, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8998773097991943, + "num_tokens": 409761508.0, + "step": 10554 + }, + { + "epoch": 1.3427044905228342, + "grad_norm": 2.470731496810913, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9011903405189514, + "num_tokens": 409803938.0, + "step": 10555 + }, + { + "epoch": 1.3428317008014248, + "grad_norm": 3.1747047901153564, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8818936944007874, + "num_tokens": 409842341.0, + "step": 10556 + }, + { + "epoch": 1.3429589110800153, + "grad_norm": 2.897395610809326, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.887171745300293, + "num_tokens": 409878308.0, + "step": 10557 + }, + { + "epoch": 1.3430861213586058, + "grad_norm": 2.4084396362304688, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8898290991783142, + "num_tokens": 409920254.0, + "step": 10558 + }, + { + "epoch": 1.3432133316371964, + "grad_norm": 2.739229202270508, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8935812711715698, + "num_tokens": 409957921.0, + "step": 10559 + }, + { + "epoch": 1.343340541915787, + "grad_norm": 3.0810813903808594, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8892093896865845, + "num_tokens": 409991107.0, + "step": 10560 + }, + { + "epoch": 1.3434677521943774, + "grad_norm": 2.9269917011260986, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8949207067489624, + "num_tokens": 410022658.0, + "step": 10561 + }, + { + "epoch": 1.343594962472968, + "grad_norm": 2.7034430503845215, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8828162550926208, + "num_tokens": 410061237.0, + "step": 10562 + }, + { + "epoch": 1.3437221727515583, + "grad_norm": 2.714693784713745, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8841481804847717, + "num_tokens": 410102724.0, + "step": 10563 + }, + { + "epoch": 1.3438493830301488, + "grad_norm": 2.678539991378784, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8957477807998657, + "num_tokens": 410143882.0, + "step": 10564 + }, + { + "epoch": 1.3439765933087393, + "grad_norm": 2.549579381942749, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8947691917419434, + "num_tokens": 410182204.0, + "step": 10565 + }, + { + "epoch": 1.3441038035873298, + "grad_norm": 2.672652244567871, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8936522006988525, + "num_tokens": 410218066.0, + "step": 10566 + }, + { + "epoch": 1.3442310138659204, + "grad_norm": 2.639439582824707, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8792147636413574, + "num_tokens": 410258715.0, + "step": 10567 + }, + { + "epoch": 1.344358224144511, + "grad_norm": 2.605736494064331, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8962216377258301, + "num_tokens": 410297000.0, + "step": 10568 + }, + { + "epoch": 1.3444854344231014, + "grad_norm": 2.694396734237671, + "learning_rate": 1e-06, + "loss": 0.3589, + "mean_token_accuracy": 0.8732216358184814, + "num_tokens": 410342580.0, + "step": 10569 + }, + { + "epoch": 1.344612644701692, + "grad_norm": 2.8104214668273926, + "learning_rate": 1e-06, + "loss": 0.3612, + "mean_token_accuracy": 0.8710341453552246, + "num_tokens": 410384407.0, + "step": 10570 + }, + { + "epoch": 1.3447398549802825, + "grad_norm": 2.497343063354492, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8977587223052979, + "num_tokens": 410423045.0, + "step": 10571 + }, + { + "epoch": 1.3448670652588728, + "grad_norm": 2.3732423782348633, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9025461673736572, + "num_tokens": 410464897.0, + "step": 10572 + }, + { + "epoch": 1.3449942755374633, + "grad_norm": 2.4021525382995605, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.8996905088424683, + "num_tokens": 410509003.0, + "step": 10573 + }, + { + "epoch": 1.3451214858160538, + "grad_norm": 2.9592907428741455, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8917607665061951, + "num_tokens": 410541978.0, + "step": 10574 + }, + { + "epoch": 1.3452486960946444, + "grad_norm": 2.7237460613250732, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8791242837905884, + "num_tokens": 410581541.0, + "step": 10575 + }, + { + "epoch": 1.345375906373235, + "grad_norm": 2.770418643951416, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8793286681175232, + "num_tokens": 410620816.0, + "step": 10576 + }, + { + "epoch": 1.3455031166518254, + "grad_norm": 2.588564395904541, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8921574354171753, + "num_tokens": 410662225.0, + "step": 10577 + }, + { + "epoch": 1.345630326930416, + "grad_norm": 3.1330456733703613, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8810861706733704, + "num_tokens": 410692374.0, + "step": 10578 + }, + { + "epoch": 1.3457575372090065, + "grad_norm": 2.7939295768737793, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8906036019325256, + "num_tokens": 410729309.0, + "step": 10579 + }, + { + "epoch": 1.345884747487597, + "grad_norm": 2.536107301712036, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8909604549407959, + "num_tokens": 410775779.0, + "step": 10580 + }, + { + "epoch": 1.3460119577661875, + "grad_norm": 2.805013656616211, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8779398202896118, + "num_tokens": 410815820.0, + "step": 10581 + }, + { + "epoch": 1.346139168044778, + "grad_norm": 2.6467678546905518, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.884846568107605, + "num_tokens": 410855944.0, + "step": 10582 + }, + { + "epoch": 1.3462663783233686, + "grad_norm": 2.9387288093566895, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8837989568710327, + "num_tokens": 410889534.0, + "step": 10583 + }, + { + "epoch": 1.3463935886019591, + "grad_norm": 2.6107170581817627, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8986126184463501, + "num_tokens": 410928192.0, + "step": 10584 + }, + { + "epoch": 1.3465207988805497, + "grad_norm": 2.259732961654663, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8920307159423828, + "num_tokens": 410980746.0, + "step": 10585 + }, + { + "epoch": 1.3466480091591402, + "grad_norm": 2.646794080734253, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8930819034576416, + "num_tokens": 411019442.0, + "step": 10586 + }, + { + "epoch": 1.3467752194377305, + "grad_norm": 2.3906972408294678, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8985300064086914, + "num_tokens": 411066674.0, + "step": 10587 + }, + { + "epoch": 1.346902429716321, + "grad_norm": 2.6822426319122314, + "learning_rate": 1e-06, + "loss": 0.346, + "mean_token_accuracy": 0.8765692710876465, + "num_tokens": 411110321.0, + "step": 10588 + }, + { + "epoch": 1.3470296399949115, + "grad_norm": 2.5352253913879395, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8950703740119934, + "num_tokens": 411151902.0, + "step": 10589 + }, + { + "epoch": 1.347156850273502, + "grad_norm": 2.627000093460083, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8947422504425049, + "num_tokens": 411188870.0, + "step": 10590 + }, + { + "epoch": 1.3472840605520926, + "grad_norm": 2.676286220550537, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8967589139938354, + "num_tokens": 411226020.0, + "step": 10591 + }, + { + "epoch": 1.3474112708306831, + "grad_norm": 2.6502134799957275, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8838851451873779, + "num_tokens": 411267517.0, + "step": 10592 + }, + { + "epoch": 1.3475384811092737, + "grad_norm": 2.656827688217163, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8926634192466736, + "num_tokens": 411302711.0, + "step": 10593 + }, + { + "epoch": 1.3476656913878642, + "grad_norm": 2.475271701812744, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8975589871406555, + "num_tokens": 411343773.0, + "step": 10594 + }, + { + "epoch": 1.3477929016664547, + "grad_norm": 2.726022720336914, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.883484959602356, + "num_tokens": 411384410.0, + "step": 10595 + }, + { + "epoch": 1.3479201119450452, + "grad_norm": 3.168503999710083, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8916303515434265, + "num_tokens": 411413393.0, + "step": 10596 + }, + { + "epoch": 1.3480473222236355, + "grad_norm": 2.184715509414673, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9041035175323486, + "num_tokens": 411462742.0, + "step": 10597 + }, + { + "epoch": 1.348174532502226, + "grad_norm": 2.761605739593506, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8980225324630737, + "num_tokens": 411498055.0, + "step": 10598 + }, + { + "epoch": 1.3483017427808166, + "grad_norm": 2.9875030517578125, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8774797916412354, + "num_tokens": 411533612.0, + "step": 10599 + }, + { + "epoch": 1.3484289530594071, + "grad_norm": 2.585726261138916, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8866746425628662, + "num_tokens": 411576750.0, + "step": 10600 + }, + { + "epoch": 1.3485561633379977, + "grad_norm": 2.636995792388916, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9031091332435608, + "num_tokens": 411613560.0, + "step": 10601 + }, + { + "epoch": 1.3486833736165882, + "grad_norm": 2.6857008934020996, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.885291576385498, + "num_tokens": 411652382.0, + "step": 10602 + }, + { + "epoch": 1.3488105838951787, + "grad_norm": 2.6581368446350098, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8934829831123352, + "num_tokens": 411691115.0, + "step": 10603 + }, + { + "epoch": 1.3489377941737692, + "grad_norm": 3.0572214126586914, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8863739967346191, + "num_tokens": 411723965.0, + "step": 10604 + }, + { + "epoch": 1.3490650044523598, + "grad_norm": 2.766303300857544, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8758233785629272, + "num_tokens": 411763334.0, + "step": 10605 + }, + { + "epoch": 1.3491922147309503, + "grad_norm": 2.6545794010162354, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.901902437210083, + "num_tokens": 411798761.0, + "step": 10606 + }, + { + "epoch": 1.3493194250095408, + "grad_norm": 2.663477897644043, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8904396295547485, + "num_tokens": 411837191.0, + "step": 10607 + }, + { + "epoch": 1.3494466352881314, + "grad_norm": 2.5741519927978516, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8936138153076172, + "num_tokens": 411880331.0, + "step": 10608 + }, + { + "epoch": 1.3495738455667219, + "grad_norm": 2.6092021465301514, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8947177529335022, + "num_tokens": 411918655.0, + "step": 10609 + }, + { + "epoch": 1.3497010558453124, + "grad_norm": 2.5511233806610107, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.894589364528656, + "num_tokens": 411957759.0, + "step": 10610 + }, + { + "epoch": 1.349828266123903, + "grad_norm": 2.7189888954162598, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8828892707824707, + "num_tokens": 411996780.0, + "step": 10611 + }, + { + "epoch": 1.3499554764024932, + "grad_norm": 2.555269718170166, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8922711610794067, + "num_tokens": 412035209.0, + "step": 10612 + }, + { + "epoch": 1.3500826866810838, + "grad_norm": 2.562487840652466, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9005295038223267, + "num_tokens": 412072001.0, + "step": 10613 + }, + { + "epoch": 1.3502098969596743, + "grad_norm": 2.485172748565674, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8990050554275513, + "num_tokens": 412111735.0, + "step": 10614 + }, + { + "epoch": 1.3503371072382648, + "grad_norm": 2.409017562866211, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.894845187664032, + "num_tokens": 412155911.0, + "step": 10615 + }, + { + "epoch": 1.3504643175168554, + "grad_norm": 2.6056859493255615, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8911617398262024, + "num_tokens": 412196720.0, + "step": 10616 + }, + { + "epoch": 1.350591527795446, + "grad_norm": 3.14688777923584, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.872206449508667, + "num_tokens": 412228987.0, + "step": 10617 + }, + { + "epoch": 1.3507187380740364, + "grad_norm": 2.487741708755493, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8928874731063843, + "num_tokens": 412270821.0, + "step": 10618 + }, + { + "epoch": 1.350845948352627, + "grad_norm": 3.1160600185394287, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8844549655914307, + "num_tokens": 412300115.0, + "step": 10619 + }, + { + "epoch": 1.3509731586312175, + "grad_norm": 2.6832239627838135, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8991407155990601, + "num_tokens": 412334254.0, + "step": 10620 + }, + { + "epoch": 1.3511003689098078, + "grad_norm": 2.4976468086242676, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8989038467407227, + "num_tokens": 412375627.0, + "step": 10621 + }, + { + "epoch": 1.3512275791883983, + "grad_norm": 2.7964155673980713, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8905207514762878, + "num_tokens": 412414739.0, + "step": 10622 + }, + { + "epoch": 1.3513547894669888, + "grad_norm": 2.7998046875, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8868840932846069, + "num_tokens": 412452875.0, + "step": 10623 + }, + { + "epoch": 1.3514819997455794, + "grad_norm": 2.8833792209625244, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8926443457603455, + "num_tokens": 412488335.0, + "step": 10624 + }, + { + "epoch": 1.35160921002417, + "grad_norm": 2.839467763900757, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8882080316543579, + "num_tokens": 412524620.0, + "step": 10625 + }, + { + "epoch": 1.3517364203027604, + "grad_norm": 2.724047899246216, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8910284042358398, + "num_tokens": 412560641.0, + "step": 10626 + }, + { + "epoch": 1.351863630581351, + "grad_norm": 2.787848711013794, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8879610896110535, + "num_tokens": 412597716.0, + "step": 10627 + }, + { + "epoch": 1.3519908408599415, + "grad_norm": 2.855057716369629, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8952009081840515, + "num_tokens": 412630240.0, + "step": 10628 + }, + { + "epoch": 1.352118051138532, + "grad_norm": 2.709395170211792, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.871712327003479, + "num_tokens": 412670778.0, + "step": 10629 + }, + { + "epoch": 1.3522452614171225, + "grad_norm": 2.8612868785858154, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8941015005111694, + "num_tokens": 412705355.0, + "step": 10630 + }, + { + "epoch": 1.352372471695713, + "grad_norm": 2.4805779457092285, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9000481367111206, + "num_tokens": 412743498.0, + "step": 10631 + }, + { + "epoch": 1.3524996819743036, + "grad_norm": 2.7134907245635986, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8889308571815491, + "num_tokens": 412781498.0, + "step": 10632 + }, + { + "epoch": 1.3526268922528941, + "grad_norm": 2.8156254291534424, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8752503395080566, + "num_tokens": 412820989.0, + "step": 10633 + }, + { + "epoch": 1.3527541025314846, + "grad_norm": 2.7310824394226074, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8824649453163147, + "num_tokens": 412858134.0, + "step": 10634 + }, + { + "epoch": 1.3528813128100752, + "grad_norm": 2.797013998031616, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8898985385894775, + "num_tokens": 412893044.0, + "step": 10635 + }, + { + "epoch": 1.3530085230886655, + "grad_norm": 2.365696907043457, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8915693163871765, + "num_tokens": 412942294.0, + "step": 10636 + }, + { + "epoch": 1.353135733367256, + "grad_norm": 2.6667654514312744, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8894407749176025, + "num_tokens": 412982222.0, + "step": 10637 + }, + { + "epoch": 1.3532629436458465, + "grad_norm": 2.6375415325164795, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8890494108200073, + "num_tokens": 413017551.0, + "step": 10638 + }, + { + "epoch": 1.353390153924437, + "grad_norm": 2.7461540699005127, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8922823071479797, + "num_tokens": 413054407.0, + "step": 10639 + }, + { + "epoch": 1.3535173642030276, + "grad_norm": 2.6281635761260986, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8990240097045898, + "num_tokens": 413091914.0, + "step": 10640 + }, + { + "epoch": 1.3536445744816181, + "grad_norm": 2.476015090942383, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9059624075889587, + "num_tokens": 413129564.0, + "step": 10641 + }, + { + "epoch": 1.3537717847602087, + "grad_norm": 2.6408567428588867, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8956747651100159, + "num_tokens": 413167471.0, + "step": 10642 + }, + { + "epoch": 1.3538989950387992, + "grad_norm": 2.8013696670532227, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.886906623840332, + "num_tokens": 413205218.0, + "step": 10643 + }, + { + "epoch": 1.3540262053173897, + "grad_norm": 2.912208318710327, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8830744624137878, + "num_tokens": 413241126.0, + "step": 10644 + }, + { + "epoch": 1.3541534155959802, + "grad_norm": 2.5096633434295654, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8932809829711914, + "num_tokens": 413282659.0, + "step": 10645 + }, + { + "epoch": 1.3542806258745705, + "grad_norm": 2.5751724243164062, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8920471668243408, + "num_tokens": 413324302.0, + "step": 10646 + }, + { + "epoch": 1.354407836153161, + "grad_norm": 2.604851722717285, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8924636244773865, + "num_tokens": 413362828.0, + "step": 10647 + }, + { + "epoch": 1.3545350464317516, + "grad_norm": 2.674346685409546, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8965405821800232, + "num_tokens": 413397991.0, + "step": 10648 + }, + { + "epoch": 1.3546622567103421, + "grad_norm": 2.906069278717041, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8938978910446167, + "num_tokens": 413430051.0, + "step": 10649 + }, + { + "epoch": 1.3547894669889327, + "grad_norm": 2.6870126724243164, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8952507972717285, + "num_tokens": 413464760.0, + "step": 10650 + }, + { + "epoch": 1.3549166772675232, + "grad_norm": 2.323856830596924, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8991372585296631, + "num_tokens": 413509358.0, + "step": 10651 + }, + { + "epoch": 1.3550438875461137, + "grad_norm": 2.7590231895446777, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8881152272224426, + "num_tokens": 413546096.0, + "step": 10652 + }, + { + "epoch": 1.3551710978247042, + "grad_norm": 2.647643804550171, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8914586305618286, + "num_tokens": 413586103.0, + "step": 10653 + }, + { + "epoch": 1.3552983081032948, + "grad_norm": 3.009040355682373, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8934911489486694, + "num_tokens": 413621815.0, + "step": 10654 + }, + { + "epoch": 1.3554255183818853, + "grad_norm": 2.7588465213775635, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8967753052711487, + "num_tokens": 413657090.0, + "step": 10655 + }, + { + "epoch": 1.3555527286604758, + "grad_norm": 2.6174991130828857, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8788719773292542, + "num_tokens": 413699340.0, + "step": 10656 + }, + { + "epoch": 1.3556799389390664, + "grad_norm": 2.3400723934173584, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8925089240074158, + "num_tokens": 413745255.0, + "step": 10657 + }, + { + "epoch": 1.3558071492176569, + "grad_norm": 2.7775025367736816, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8934524059295654, + "num_tokens": 413781250.0, + "step": 10658 + }, + { + "epoch": 1.3559343594962474, + "grad_norm": 2.6713058948516846, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8791087865829468, + "num_tokens": 413822142.0, + "step": 10659 + }, + { + "epoch": 1.356061569774838, + "grad_norm": 2.6223878860473633, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8872816562652588, + "num_tokens": 413863910.0, + "step": 10660 + }, + { + "epoch": 1.3561887800534282, + "grad_norm": 2.8927838802337646, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8849111795425415, + "num_tokens": 413899760.0, + "step": 10661 + }, + { + "epoch": 1.3563159903320188, + "grad_norm": 2.828956365585327, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.882741391658783, + "num_tokens": 413937479.0, + "step": 10662 + }, + { + "epoch": 1.3564432006106093, + "grad_norm": 2.6923253536224365, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8869665265083313, + "num_tokens": 413975737.0, + "step": 10663 + }, + { + "epoch": 1.3565704108891998, + "grad_norm": 2.58009672164917, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8968455791473389, + "num_tokens": 414012868.0, + "step": 10664 + }, + { + "epoch": 1.3566976211677904, + "grad_norm": 2.806325674057007, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9016074538230896, + "num_tokens": 414043012.0, + "step": 10665 + }, + { + "epoch": 1.3568248314463809, + "grad_norm": 2.898361921310425, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8766725063323975, + "num_tokens": 414080765.0, + "step": 10666 + }, + { + "epoch": 1.3569520417249714, + "grad_norm": 2.568268299102783, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8951660990715027, + "num_tokens": 414118156.0, + "step": 10667 + }, + { + "epoch": 1.357079252003562, + "grad_norm": 2.3524508476257324, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9086240530014038, + "num_tokens": 414159314.0, + "step": 10668 + }, + { + "epoch": 1.3572064622821525, + "grad_norm": 2.546485424041748, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8895454406738281, + "num_tokens": 414200876.0, + "step": 10669 + }, + { + "epoch": 1.3573336725607428, + "grad_norm": 2.7937912940979004, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8796895742416382, + "num_tokens": 414240419.0, + "step": 10670 + }, + { + "epoch": 1.3574608828393333, + "grad_norm": 2.6489384174346924, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8957201242446899, + "num_tokens": 414278540.0, + "step": 10671 + }, + { + "epoch": 1.3575880931179238, + "grad_norm": 3.416863203048706, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8993213176727295, + "num_tokens": 414317316.0, + "step": 10672 + }, + { + "epoch": 1.3577153033965144, + "grad_norm": 2.8898303508758545, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9007817506790161, + "num_tokens": 414347774.0, + "step": 10673 + }, + { + "epoch": 1.357842513675105, + "grad_norm": 2.6052658557891846, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8882337808609009, + "num_tokens": 414390867.0, + "step": 10674 + }, + { + "epoch": 1.3579697239536954, + "grad_norm": 2.622864007949829, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8919253349304199, + "num_tokens": 414433122.0, + "step": 10675 + }, + { + "epoch": 1.358096934232286, + "grad_norm": 2.5892045497894287, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.894531786441803, + "num_tokens": 414473766.0, + "step": 10676 + }, + { + "epoch": 1.3582241445108765, + "grad_norm": 2.6121726036071777, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8809733986854553, + "num_tokens": 414518850.0, + "step": 10677 + }, + { + "epoch": 1.358351354789467, + "grad_norm": 2.432741641998291, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8914934396743774, + "num_tokens": 414564311.0, + "step": 10678 + }, + { + "epoch": 1.3584785650680575, + "grad_norm": 2.6791090965270996, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8979581594467163, + "num_tokens": 414599508.0, + "step": 10679 + }, + { + "epoch": 1.358605775346648, + "grad_norm": 2.5463595390319824, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8871276378631592, + "num_tokens": 414638639.0, + "step": 10680 + }, + { + "epoch": 1.3587329856252386, + "grad_norm": 2.596419095993042, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8931592106819153, + "num_tokens": 414677107.0, + "step": 10681 + }, + { + "epoch": 1.3588601959038291, + "grad_norm": 2.7322864532470703, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8772323131561279, + "num_tokens": 414713829.0, + "step": 10682 + }, + { + "epoch": 1.3589874061824196, + "grad_norm": 2.8105831146240234, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8740378618240356, + "num_tokens": 414756388.0, + "step": 10683 + }, + { + "epoch": 1.3591146164610102, + "grad_norm": 2.486907720565796, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8877749443054199, + "num_tokens": 414798363.0, + "step": 10684 + }, + { + "epoch": 1.3592418267396005, + "grad_norm": 2.5870838165283203, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8956420421600342, + "num_tokens": 414836584.0, + "step": 10685 + }, + { + "epoch": 1.359369037018191, + "grad_norm": 2.5867302417755127, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8817257285118103, + "num_tokens": 414877650.0, + "step": 10686 + }, + { + "epoch": 1.3594962472967815, + "grad_norm": 2.624432325363159, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8961453437805176, + "num_tokens": 414914086.0, + "step": 10687 + }, + { + "epoch": 1.359623457575372, + "grad_norm": 2.5330324172973633, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8931502103805542, + "num_tokens": 414957834.0, + "step": 10688 + }, + { + "epoch": 1.3597506678539626, + "grad_norm": 2.604764223098755, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9031386375427246, + "num_tokens": 414995569.0, + "step": 10689 + }, + { + "epoch": 1.3598778781325531, + "grad_norm": 2.614797353744507, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8983380198478699, + "num_tokens": 415032890.0, + "step": 10690 + }, + { + "epoch": 1.3600050884111436, + "grad_norm": 2.8763039112091064, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8801515102386475, + "num_tokens": 415067733.0, + "step": 10691 + }, + { + "epoch": 1.3601322986897342, + "grad_norm": 2.525181770324707, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9003517031669617, + "num_tokens": 415105453.0, + "step": 10692 + }, + { + "epoch": 1.3602595089683247, + "grad_norm": 2.5574746131896973, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9008409380912781, + "num_tokens": 415144770.0, + "step": 10693 + }, + { + "epoch": 1.3603867192469152, + "grad_norm": 2.7489192485809326, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8865333795547485, + "num_tokens": 415180739.0, + "step": 10694 + }, + { + "epoch": 1.3605139295255055, + "grad_norm": 2.8797247409820557, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8931312561035156, + "num_tokens": 415213447.0, + "step": 10695 + }, + { + "epoch": 1.360641139804096, + "grad_norm": 2.6723616123199463, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8897719383239746, + "num_tokens": 415250041.0, + "step": 10696 + }, + { + "epoch": 1.3607683500826866, + "grad_norm": 2.782626152038574, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8950860500335693, + "num_tokens": 415284963.0, + "step": 10697 + }, + { + "epoch": 1.3608955603612771, + "grad_norm": 2.996961832046509, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.878696084022522, + "num_tokens": 415322764.0, + "step": 10698 + }, + { + "epoch": 1.3610227706398677, + "grad_norm": 2.6962530612945557, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8900211453437805, + "num_tokens": 415362671.0, + "step": 10699 + }, + { + "epoch": 1.3611499809184582, + "grad_norm": 2.627230167388916, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.892958402633667, + "num_tokens": 415400895.0, + "step": 10700 + }, + { + "epoch": 1.3612771911970487, + "grad_norm": 2.3596577644348145, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9007444381713867, + "num_tokens": 415443528.0, + "step": 10701 + }, + { + "epoch": 1.3614044014756392, + "grad_norm": 2.6073684692382812, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8932953476905823, + "num_tokens": 415485744.0, + "step": 10702 + }, + { + "epoch": 1.3615316117542298, + "grad_norm": 2.8734490871429443, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8880503177642822, + "num_tokens": 415519521.0, + "step": 10703 + }, + { + "epoch": 1.3616588220328203, + "grad_norm": 2.6646840572357178, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9065409898757935, + "num_tokens": 415556441.0, + "step": 10704 + }, + { + "epoch": 1.3617860323114108, + "grad_norm": 2.550283670425415, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8886196613311768, + "num_tokens": 415597992.0, + "step": 10705 + }, + { + "epoch": 1.3619132425900013, + "grad_norm": 2.787850856781006, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8938971757888794, + "num_tokens": 415632004.0, + "step": 10706 + }, + { + "epoch": 1.3620404528685919, + "grad_norm": 2.901336193084717, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8925352096557617, + "num_tokens": 415662068.0, + "step": 10707 + }, + { + "epoch": 1.3621676631471824, + "grad_norm": 2.8523526191711426, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8941370248794556, + "num_tokens": 415696464.0, + "step": 10708 + }, + { + "epoch": 1.362294873425773, + "grad_norm": 2.9405105113983154, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8909862041473389, + "num_tokens": 415727959.0, + "step": 10709 + }, + { + "epoch": 1.3624220837043632, + "grad_norm": 2.8377485275268555, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8743401765823364, + "num_tokens": 415763874.0, + "step": 10710 + }, + { + "epoch": 1.3625492939829538, + "grad_norm": 2.59061598777771, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.896308183670044, + "num_tokens": 415800503.0, + "step": 10711 + }, + { + "epoch": 1.3626765042615443, + "grad_norm": 2.7824857234954834, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8861863613128662, + "num_tokens": 415834749.0, + "step": 10712 + }, + { + "epoch": 1.3628037145401348, + "grad_norm": 2.850651264190674, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8947643041610718, + "num_tokens": 415866889.0, + "step": 10713 + }, + { + "epoch": 1.3629309248187254, + "grad_norm": 2.663316249847412, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.889305830001831, + "num_tokens": 415908275.0, + "step": 10714 + }, + { + "epoch": 1.3630581350973159, + "grad_norm": 2.591811418533325, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8916652798652649, + "num_tokens": 415948414.0, + "step": 10715 + }, + { + "epoch": 1.3631853453759064, + "grad_norm": 2.4952750205993652, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8892388343811035, + "num_tokens": 415992533.0, + "step": 10716 + }, + { + "epoch": 1.363312555654497, + "grad_norm": 2.5508267879486084, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8940523266792297, + "num_tokens": 416034848.0, + "step": 10717 + }, + { + "epoch": 1.3634397659330875, + "grad_norm": 2.7625927925109863, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8892842531204224, + "num_tokens": 416069785.0, + "step": 10718 + }, + { + "epoch": 1.3635669762116778, + "grad_norm": 2.826521158218384, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8894070982933044, + "num_tokens": 416108076.0, + "step": 10719 + }, + { + "epoch": 1.3636941864902683, + "grad_norm": 2.7604615688323975, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8834829330444336, + "num_tokens": 416149788.0, + "step": 10720 + }, + { + "epoch": 1.3638213967688588, + "grad_norm": 2.793654203414917, + "learning_rate": 1e-06, + "loss": 0.3239, + "mean_token_accuracy": 0.887197732925415, + "num_tokens": 416188119.0, + "step": 10721 + }, + { + "epoch": 1.3639486070474494, + "grad_norm": 2.776977777481079, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8878647089004517, + "num_tokens": 416223279.0, + "step": 10722 + }, + { + "epoch": 1.3640758173260399, + "grad_norm": 2.6173133850097656, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8959835767745972, + "num_tokens": 416263289.0, + "step": 10723 + }, + { + "epoch": 1.3642030276046304, + "grad_norm": 2.7306337356567383, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.8965811133384705, + "num_tokens": 416299563.0, + "step": 10724 + }, + { + "epoch": 1.364330237883221, + "grad_norm": 2.7783172130584717, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8875101804733276, + "num_tokens": 416334451.0, + "step": 10725 + }, + { + "epoch": 1.3644574481618115, + "grad_norm": 2.7678372859954834, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8952153921127319, + "num_tokens": 416370315.0, + "step": 10726 + }, + { + "epoch": 1.364584658440402, + "grad_norm": 2.7898542881011963, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8809599876403809, + "num_tokens": 416408028.0, + "step": 10727 + }, + { + "epoch": 1.3647118687189925, + "grad_norm": 2.5884671211242676, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8943464756011963, + "num_tokens": 416445825.0, + "step": 10728 + }, + { + "epoch": 1.364839078997583, + "grad_norm": 2.4991958141326904, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8857284784317017, + "num_tokens": 416490404.0, + "step": 10729 + }, + { + "epoch": 1.3649662892761736, + "grad_norm": 2.603181838989258, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8918484449386597, + "num_tokens": 416528003.0, + "step": 10730 + }, + { + "epoch": 1.3650934995547641, + "grad_norm": 2.8055474758148193, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8837438821792603, + "num_tokens": 416563808.0, + "step": 10731 + }, + { + "epoch": 1.3652207098333546, + "grad_norm": 2.406181573867798, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9069365859031677, + "num_tokens": 416600057.0, + "step": 10732 + }, + { + "epoch": 1.3653479201119452, + "grad_norm": 2.563260078430176, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.8965186476707458, + "num_tokens": 416638136.0, + "step": 10733 + }, + { + "epoch": 1.3654751303905355, + "grad_norm": 2.675549268722534, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8864555954933167, + "num_tokens": 416679470.0, + "step": 10734 + }, + { + "epoch": 1.365602340669126, + "grad_norm": 2.43485164642334, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8960745334625244, + "num_tokens": 416722145.0, + "step": 10735 + }, + { + "epoch": 1.3657295509477165, + "grad_norm": 2.4448153972625732, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8983080387115479, + "num_tokens": 416764125.0, + "step": 10736 + }, + { + "epoch": 1.365856761226307, + "grad_norm": 2.8481850624084473, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8893054127693176, + "num_tokens": 416798257.0, + "step": 10737 + }, + { + "epoch": 1.3659839715048976, + "grad_norm": 2.6327755451202393, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8947722911834717, + "num_tokens": 416837101.0, + "step": 10738 + }, + { + "epoch": 1.3661111817834881, + "grad_norm": 2.800748825073242, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8908306360244751, + "num_tokens": 416869835.0, + "step": 10739 + }, + { + "epoch": 1.3662383920620786, + "grad_norm": 2.6176629066467285, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.893271267414093, + "num_tokens": 416909030.0, + "step": 10740 + }, + { + "epoch": 1.3663656023406692, + "grad_norm": 2.5640745162963867, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8890013694763184, + "num_tokens": 416951201.0, + "step": 10741 + }, + { + "epoch": 1.3664928126192597, + "grad_norm": 2.659210443496704, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8908679485321045, + "num_tokens": 416990016.0, + "step": 10742 + }, + { + "epoch": 1.3666200228978502, + "grad_norm": 2.4466326236724854, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8972028493881226, + "num_tokens": 417031745.0, + "step": 10743 + }, + { + "epoch": 1.3667472331764405, + "grad_norm": 2.5095503330230713, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.89094477891922, + "num_tokens": 417078239.0, + "step": 10744 + }, + { + "epoch": 1.366874443455031, + "grad_norm": 2.603663682937622, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8990472555160522, + "num_tokens": 417116070.0, + "step": 10745 + }, + { + "epoch": 1.3670016537336216, + "grad_norm": 2.8848299980163574, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8945263624191284, + "num_tokens": 417147690.0, + "step": 10746 + }, + { + "epoch": 1.3671288640122121, + "grad_norm": 2.642874240875244, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8917152881622314, + "num_tokens": 417186101.0, + "step": 10747 + }, + { + "epoch": 1.3672560742908026, + "grad_norm": 2.391807794570923, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9061002731323242, + "num_tokens": 417223941.0, + "step": 10748 + }, + { + "epoch": 1.3673832845693932, + "grad_norm": 2.5889668464660645, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8942192792892456, + "num_tokens": 417261099.0, + "step": 10749 + }, + { + "epoch": 1.3675104948479837, + "grad_norm": 2.479367256164551, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.8739671111106873, + "num_tokens": 417310153.0, + "step": 10750 + }, + { + "epoch": 1.3676377051265742, + "grad_norm": 2.686145067214966, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8900483846664429, + "num_tokens": 417350693.0, + "step": 10751 + }, + { + "epoch": 1.3677649154051648, + "grad_norm": 2.9518425464630127, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8804247975349426, + "num_tokens": 417388308.0, + "step": 10752 + }, + { + "epoch": 1.3678921256837553, + "grad_norm": 2.781043767929077, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8750618696212769, + "num_tokens": 417426355.0, + "step": 10753 + }, + { + "epoch": 1.3680193359623458, + "grad_norm": 2.5948963165283203, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8890461921691895, + "num_tokens": 417464074.0, + "step": 10754 + }, + { + "epoch": 1.3681465462409363, + "grad_norm": 2.5939221382141113, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.886211633682251, + "num_tokens": 417504422.0, + "step": 10755 + }, + { + "epoch": 1.3682737565195269, + "grad_norm": 2.6696431636810303, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9011462926864624, + "num_tokens": 417538575.0, + "step": 10756 + }, + { + "epoch": 1.3684009667981174, + "grad_norm": 2.6687443256378174, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8871251940727234, + "num_tokens": 417579117.0, + "step": 10757 + }, + { + "epoch": 1.368528177076708, + "grad_norm": 2.7217063903808594, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8930553197860718, + "num_tokens": 417615836.0, + "step": 10758 + }, + { + "epoch": 1.3686553873552982, + "grad_norm": 2.7714767456054688, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8886075019836426, + "num_tokens": 417651635.0, + "step": 10759 + }, + { + "epoch": 1.3687825976338888, + "grad_norm": 2.713313341140747, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8865182399749756, + "num_tokens": 417695894.0, + "step": 10760 + }, + { + "epoch": 1.3689098079124793, + "grad_norm": 2.6471493244171143, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8783082962036133, + "num_tokens": 417737198.0, + "step": 10761 + }, + { + "epoch": 1.3690370181910698, + "grad_norm": 2.7524564266204834, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.902185320854187, + "num_tokens": 417771629.0, + "step": 10762 + }, + { + "epoch": 1.3691642284696603, + "grad_norm": 2.651038885116577, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.883151113986969, + "num_tokens": 417813255.0, + "step": 10763 + }, + { + "epoch": 1.3692914387482509, + "grad_norm": 2.8440141677856445, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8890411853790283, + "num_tokens": 417850233.0, + "step": 10764 + }, + { + "epoch": 1.3694186490268414, + "grad_norm": 2.766622304916382, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8826825618743896, + "num_tokens": 417889468.0, + "step": 10765 + }, + { + "epoch": 1.369545859305432, + "grad_norm": 2.7036850452423096, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8920766711235046, + "num_tokens": 417929385.0, + "step": 10766 + }, + { + "epoch": 1.3696730695840225, + "grad_norm": 2.8598594665527344, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8849022388458252, + "num_tokens": 417963917.0, + "step": 10767 + }, + { + "epoch": 1.3698002798626128, + "grad_norm": 2.7322466373443604, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8885936737060547, + "num_tokens": 418003439.0, + "step": 10768 + }, + { + "epoch": 1.3699274901412033, + "grad_norm": 2.7133100032806396, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8938913345336914, + "num_tokens": 418039973.0, + "step": 10769 + }, + { + "epoch": 1.3700547004197938, + "grad_norm": 2.643190383911133, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8865170478820801, + "num_tokens": 418080210.0, + "step": 10770 + }, + { + "epoch": 1.3701819106983844, + "grad_norm": 2.656470537185669, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9047602415084839, + "num_tokens": 418116849.0, + "step": 10771 + }, + { + "epoch": 1.3703091209769749, + "grad_norm": 2.5624313354492188, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9036505222320557, + "num_tokens": 418155792.0, + "step": 10772 + }, + { + "epoch": 1.3704363312555654, + "grad_norm": 2.5121374130249023, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8903833627700806, + "num_tokens": 418197444.0, + "step": 10773 + }, + { + "epoch": 1.370563541534156, + "grad_norm": 2.427494764328003, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9056346416473389, + "num_tokens": 418235972.0, + "step": 10774 + }, + { + "epoch": 1.3706907518127465, + "grad_norm": 2.53108286857605, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8959286212921143, + "num_tokens": 418276642.0, + "step": 10775 + }, + { + "epoch": 1.370817962091337, + "grad_norm": 2.7630341053009033, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8843021392822266, + "num_tokens": 418315214.0, + "step": 10776 + }, + { + "epoch": 1.3709451723699275, + "grad_norm": 2.555537700653076, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8912166357040405, + "num_tokens": 418351810.0, + "step": 10777 + }, + { + "epoch": 1.371072382648518, + "grad_norm": 2.678924083709717, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8826178908348083, + "num_tokens": 418396462.0, + "step": 10778 + }, + { + "epoch": 1.3711995929271086, + "grad_norm": 2.4306795597076416, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8946376442909241, + "num_tokens": 418439160.0, + "step": 10779 + }, + { + "epoch": 1.371326803205699, + "grad_norm": 2.529059410095215, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8915119767189026, + "num_tokens": 418479723.0, + "step": 10780 + }, + { + "epoch": 1.3714540134842896, + "grad_norm": 2.4253673553466797, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9003337621688843, + "num_tokens": 418519668.0, + "step": 10781 + }, + { + "epoch": 1.3715812237628802, + "grad_norm": 2.5435385704040527, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8963595628738403, + "num_tokens": 418556880.0, + "step": 10782 + }, + { + "epoch": 1.3717084340414705, + "grad_norm": 2.7034733295440674, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.881757378578186, + "num_tokens": 418599035.0, + "step": 10783 + }, + { + "epoch": 1.371835644320061, + "grad_norm": 2.913982629776001, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9035199880599976, + "num_tokens": 418626973.0, + "step": 10784 + }, + { + "epoch": 1.3719628545986515, + "grad_norm": 2.740554094314575, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9118621349334717, + "num_tokens": 418658102.0, + "step": 10785 + }, + { + "epoch": 1.372090064877242, + "grad_norm": 2.490675210952759, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8930114507675171, + "num_tokens": 418701712.0, + "step": 10786 + }, + { + "epoch": 1.3722172751558326, + "grad_norm": 2.3830769062042236, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8979439735412598, + "num_tokens": 418748408.0, + "step": 10787 + }, + { + "epoch": 1.3723444854344231, + "grad_norm": 2.5227084159851074, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8881070017814636, + "num_tokens": 418795016.0, + "step": 10788 + }, + { + "epoch": 1.3724716957130136, + "grad_norm": 2.478538751602173, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8924934267997742, + "num_tokens": 418836339.0, + "step": 10789 + }, + { + "epoch": 1.3725989059916042, + "grad_norm": 2.7114317417144775, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8939496278762817, + "num_tokens": 418872207.0, + "step": 10790 + }, + { + "epoch": 1.3727261162701947, + "grad_norm": 2.638692855834961, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8890171051025391, + "num_tokens": 418911685.0, + "step": 10791 + }, + { + "epoch": 1.3728533265487852, + "grad_norm": 2.357114791870117, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9097837209701538, + "num_tokens": 418952391.0, + "step": 10792 + }, + { + "epoch": 1.3729805368273755, + "grad_norm": 2.6966617107391357, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.882174551486969, + "num_tokens": 418991562.0, + "step": 10793 + }, + { + "epoch": 1.373107747105966, + "grad_norm": 2.921410083770752, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8936710357666016, + "num_tokens": 419023157.0, + "step": 10794 + }, + { + "epoch": 1.3732349573845566, + "grad_norm": 3.038358449935913, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8895317912101746, + "num_tokens": 419055033.0, + "step": 10795 + }, + { + "epoch": 1.3733621676631471, + "grad_norm": 2.8727340698242188, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8843196630477905, + "num_tokens": 419090429.0, + "step": 10796 + }, + { + "epoch": 1.3734893779417376, + "grad_norm": 2.539210557937622, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8852550387382507, + "num_tokens": 419135668.0, + "step": 10797 + }, + { + "epoch": 1.3736165882203282, + "grad_norm": 3.0975780487060547, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8949079513549805, + "num_tokens": 419169115.0, + "step": 10798 + }, + { + "epoch": 1.3737437984989187, + "grad_norm": 2.6337764263153076, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8891445398330688, + "num_tokens": 419210145.0, + "step": 10799 + }, + { + "epoch": 1.3738710087775092, + "grad_norm": 2.94250750541687, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8929625749588013, + "num_tokens": 419245519.0, + "step": 10800 + }, + { + "epoch": 1.3739982190560998, + "grad_norm": 2.4726428985595703, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8975604772567749, + "num_tokens": 419287006.0, + "step": 10801 + }, + { + "epoch": 1.3741254293346903, + "grad_norm": 2.578460931777954, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8843560218811035, + "num_tokens": 419329747.0, + "step": 10802 + }, + { + "epoch": 1.3742526396132808, + "grad_norm": 2.6320533752441406, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9016493558883667, + "num_tokens": 419364036.0, + "step": 10803 + }, + { + "epoch": 1.3743798498918713, + "grad_norm": 2.454292058944702, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8983700275421143, + "num_tokens": 419408345.0, + "step": 10804 + }, + { + "epoch": 1.3745070601704619, + "grad_norm": 2.5329480171203613, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9007608890533447, + "num_tokens": 419444575.0, + "step": 10805 + }, + { + "epoch": 1.3746342704490524, + "grad_norm": 2.5652101039886475, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8901282548904419, + "num_tokens": 419482421.0, + "step": 10806 + }, + { + "epoch": 1.374761480727643, + "grad_norm": 2.7961761951446533, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8789703845977783, + "num_tokens": 419520176.0, + "step": 10807 + }, + { + "epoch": 1.3748886910062332, + "grad_norm": 2.546631097793579, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8870595693588257, + "num_tokens": 419562524.0, + "step": 10808 + }, + { + "epoch": 1.3750159012848238, + "grad_norm": 2.815732479095459, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8749085664749146, + "num_tokens": 419601319.0, + "step": 10809 + }, + { + "epoch": 1.3751431115634143, + "grad_norm": 2.420077323913574, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9075459241867065, + "num_tokens": 419638969.0, + "step": 10810 + }, + { + "epoch": 1.3752703218420048, + "grad_norm": 2.6507046222686768, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8857041597366333, + "num_tokens": 419680970.0, + "step": 10811 + }, + { + "epoch": 1.3753975321205953, + "grad_norm": 2.4388163089752197, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8923965692520142, + "num_tokens": 419723848.0, + "step": 10812 + }, + { + "epoch": 1.3755247423991859, + "grad_norm": 2.6038360595703125, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8996282815933228, + "num_tokens": 419761416.0, + "step": 10813 + }, + { + "epoch": 1.3756519526777764, + "grad_norm": 2.8616764545440674, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8831816911697388, + "num_tokens": 419795361.0, + "step": 10814 + }, + { + "epoch": 1.375779162956367, + "grad_norm": 2.6450603008270264, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8882790207862854, + "num_tokens": 419837116.0, + "step": 10815 + }, + { + "epoch": 1.3759063732349575, + "grad_norm": 2.5263445377349854, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8829635977745056, + "num_tokens": 419882714.0, + "step": 10816 + }, + { + "epoch": 1.3760335835135478, + "grad_norm": 4.4818243980407715, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8829141855239868, + "num_tokens": 419915539.0, + "step": 10817 + }, + { + "epoch": 1.3761607937921383, + "grad_norm": 2.573467969894409, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8949612379074097, + "num_tokens": 419955460.0, + "step": 10818 + }, + { + "epoch": 1.3762880040707288, + "grad_norm": 2.8718292713165283, + "learning_rate": 1e-06, + "loss": 0.3605, + "mean_token_accuracy": 0.8758300542831421, + "num_tokens": 419995079.0, + "step": 10819 + }, + { + "epoch": 1.3764152143493193, + "grad_norm": 3.1950066089630127, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.8977599143981934, + "num_tokens": 420027989.0, + "step": 10820 + }, + { + "epoch": 1.3765424246279099, + "grad_norm": 2.525911808013916, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8854482173919678, + "num_tokens": 420072875.0, + "step": 10821 + }, + { + "epoch": 1.3766696349065004, + "grad_norm": 2.5387163162231445, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8984851241111755, + "num_tokens": 420114363.0, + "step": 10822 + }, + { + "epoch": 1.376796845185091, + "grad_norm": 2.573662757873535, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9057259559631348, + "num_tokens": 420149176.0, + "step": 10823 + }, + { + "epoch": 1.3769240554636815, + "grad_norm": 2.6065967082977295, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8907324075698853, + "num_tokens": 420187328.0, + "step": 10824 + }, + { + "epoch": 1.377051265742272, + "grad_norm": 2.7473971843719482, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8882511854171753, + "num_tokens": 420224000.0, + "step": 10825 + }, + { + "epoch": 1.3771784760208625, + "grad_norm": 2.424910068511963, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8934661149978638, + "num_tokens": 420264137.0, + "step": 10826 + }, + { + "epoch": 1.377305686299453, + "grad_norm": 2.5641913414001465, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.895027220249176, + "num_tokens": 420304773.0, + "step": 10827 + }, + { + "epoch": 1.3774328965780436, + "grad_norm": 2.5851430892944336, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.880953848361969, + "num_tokens": 420346106.0, + "step": 10828 + }, + { + "epoch": 1.377560106856634, + "grad_norm": 2.3816323280334473, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9009549021720886, + "num_tokens": 420389656.0, + "step": 10829 + }, + { + "epoch": 1.3776873171352246, + "grad_norm": 2.950782299041748, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8769885897636414, + "num_tokens": 420424894.0, + "step": 10830 + }, + { + "epoch": 1.3778145274138152, + "grad_norm": 2.3457539081573486, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9025795459747314, + "num_tokens": 420467411.0, + "step": 10831 + }, + { + "epoch": 1.3779417376924055, + "grad_norm": 2.439150810241699, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8982932567596436, + "num_tokens": 420510543.0, + "step": 10832 + }, + { + "epoch": 1.378068947970996, + "grad_norm": 2.8668935298919678, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.883762538433075, + "num_tokens": 420545340.0, + "step": 10833 + }, + { + "epoch": 1.3781961582495865, + "grad_norm": 2.763052463531494, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8767006397247314, + "num_tokens": 420582230.0, + "step": 10834 + }, + { + "epoch": 1.378323368528177, + "grad_norm": 2.6737112998962402, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8888764381408691, + "num_tokens": 420620283.0, + "step": 10835 + }, + { + "epoch": 1.3784505788067676, + "grad_norm": 2.6524288654327393, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8954338431358337, + "num_tokens": 420660584.0, + "step": 10836 + }, + { + "epoch": 1.378577789085358, + "grad_norm": 2.531827211380005, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8967856168746948, + "num_tokens": 420701217.0, + "step": 10837 + }, + { + "epoch": 1.3787049993639486, + "grad_norm": 2.704045534133911, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.8969119191169739, + "num_tokens": 420738187.0, + "step": 10838 + }, + { + "epoch": 1.3788322096425392, + "grad_norm": 2.8102974891662598, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.889471173286438, + "num_tokens": 420775697.0, + "step": 10839 + }, + { + "epoch": 1.3789594199211297, + "grad_norm": 2.5754284858703613, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8821390271186829, + "num_tokens": 420817341.0, + "step": 10840 + }, + { + "epoch": 1.3790866301997202, + "grad_norm": 2.675706148147583, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8890762329101562, + "num_tokens": 420858373.0, + "step": 10841 + }, + { + "epoch": 1.3792138404783105, + "grad_norm": 2.5338690280914307, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8974226713180542, + "num_tokens": 420897282.0, + "step": 10842 + }, + { + "epoch": 1.379341050756901, + "grad_norm": 2.625887632369995, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8911948204040527, + "num_tokens": 420937370.0, + "step": 10843 + }, + { + "epoch": 1.3794682610354916, + "grad_norm": 2.7148995399475098, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8913661241531372, + "num_tokens": 420975931.0, + "step": 10844 + }, + { + "epoch": 1.3795954713140821, + "grad_norm": 2.7082669734954834, + "learning_rate": 1e-06, + "loss": 0.3712, + "mean_token_accuracy": 0.8697439432144165, + "num_tokens": 421017063.0, + "step": 10845 + }, + { + "epoch": 1.3797226815926726, + "grad_norm": 2.411832809448242, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9026080965995789, + "num_tokens": 421057287.0, + "step": 10846 + }, + { + "epoch": 1.3798498918712632, + "grad_norm": 2.5867714881896973, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8990124464035034, + "num_tokens": 421092896.0, + "step": 10847 + }, + { + "epoch": 1.3799771021498537, + "grad_norm": 2.8884522914886475, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8811885714530945, + "num_tokens": 421128790.0, + "step": 10848 + }, + { + "epoch": 1.3801043124284442, + "grad_norm": 2.5353622436523438, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.886088490486145, + "num_tokens": 421175492.0, + "step": 10849 + }, + { + "epoch": 1.3802315227070348, + "grad_norm": 2.667264223098755, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8816348314285278, + "num_tokens": 421214802.0, + "step": 10850 + }, + { + "epoch": 1.3803587329856253, + "grad_norm": 3.0147812366485596, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8879274129867554, + "num_tokens": 421245369.0, + "step": 10851 + }, + { + "epoch": 1.3804859432642158, + "grad_norm": 2.9131641387939453, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.888946533203125, + "num_tokens": 421281732.0, + "step": 10852 + }, + { + "epoch": 1.3806131535428063, + "grad_norm": 2.594444751739502, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8910098671913147, + "num_tokens": 421321607.0, + "step": 10853 + }, + { + "epoch": 1.3807403638213969, + "grad_norm": 2.6059603691101074, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8779425621032715, + "num_tokens": 421363881.0, + "step": 10854 + }, + { + "epoch": 1.3808675740999874, + "grad_norm": 2.5175905227661133, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.895667314529419, + "num_tokens": 421405989.0, + "step": 10855 + }, + { + "epoch": 1.380994784378578, + "grad_norm": 2.6769766807556152, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9008138179779053, + "num_tokens": 421441201.0, + "step": 10856 + }, + { + "epoch": 1.3811219946571682, + "grad_norm": 2.561331033706665, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.894864559173584, + "num_tokens": 421480550.0, + "step": 10857 + }, + { + "epoch": 1.3812492049357588, + "grad_norm": 2.665128231048584, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8882510662078857, + "num_tokens": 421521181.0, + "step": 10858 + }, + { + "epoch": 1.3813764152143493, + "grad_norm": 2.574320077896118, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9007647633552551, + "num_tokens": 421559838.0, + "step": 10859 + }, + { + "epoch": 1.3815036254929398, + "grad_norm": 2.6467318534851074, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8996757864952087, + "num_tokens": 421595865.0, + "step": 10860 + }, + { + "epoch": 1.3816308357715303, + "grad_norm": 2.564896583557129, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8954023122787476, + "num_tokens": 421634808.0, + "step": 10861 + }, + { + "epoch": 1.3817580460501209, + "grad_norm": 3.172924280166626, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.8788334131240845, + "num_tokens": 421670445.0, + "step": 10862 + }, + { + "epoch": 1.3818852563287114, + "grad_norm": 2.6675286293029785, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8949576616287231, + "num_tokens": 421708965.0, + "step": 10863 + }, + { + "epoch": 1.382012466607302, + "grad_norm": 2.6708598136901855, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8855775594711304, + "num_tokens": 421747406.0, + "step": 10864 + }, + { + "epoch": 1.3821396768858925, + "grad_norm": 2.7660772800445557, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8930940628051758, + "num_tokens": 421783257.0, + "step": 10865 + }, + { + "epoch": 1.3822668871644828, + "grad_norm": 2.6238720417022705, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8798391819000244, + "num_tokens": 421830420.0, + "step": 10866 + }, + { + "epoch": 1.3823940974430733, + "grad_norm": 2.422166109085083, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.896226704120636, + "num_tokens": 421872070.0, + "step": 10867 + }, + { + "epoch": 1.3825213077216638, + "grad_norm": 2.545200824737549, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8963562250137329, + "num_tokens": 421913478.0, + "step": 10868 + }, + { + "epoch": 1.3826485180002543, + "grad_norm": 2.5556585788726807, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8929638862609863, + "num_tokens": 421955451.0, + "step": 10869 + }, + { + "epoch": 1.3827757282788449, + "grad_norm": 2.9729363918304443, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8900025486946106, + "num_tokens": 421988586.0, + "step": 10870 + }, + { + "epoch": 1.3829029385574354, + "grad_norm": 2.567242383956909, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8859373927116394, + "num_tokens": 422030488.0, + "step": 10871 + }, + { + "epoch": 1.383030148836026, + "grad_norm": 2.7902796268463135, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8793482184410095, + "num_tokens": 422069182.0, + "step": 10872 + }, + { + "epoch": 1.3831573591146165, + "grad_norm": 2.695725440979004, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8881961107254028, + "num_tokens": 422105624.0, + "step": 10873 + }, + { + "epoch": 1.383284569393207, + "grad_norm": 2.5992863178253174, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8944346308708191, + "num_tokens": 422144773.0, + "step": 10874 + }, + { + "epoch": 1.3834117796717975, + "grad_norm": 2.5260133743286133, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8933968544006348, + "num_tokens": 422189504.0, + "step": 10875 + }, + { + "epoch": 1.383538989950388, + "grad_norm": 2.959575891494751, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8787747025489807, + "num_tokens": 422225180.0, + "step": 10876 + }, + { + "epoch": 1.3836662002289786, + "grad_norm": 2.649733066558838, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.89506995677948, + "num_tokens": 422261070.0, + "step": 10877 + }, + { + "epoch": 1.383793410507569, + "grad_norm": 2.484503984451294, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8849743008613586, + "num_tokens": 422304542.0, + "step": 10878 + }, + { + "epoch": 1.3839206207861596, + "grad_norm": 2.6773862838745117, + "learning_rate": 1e-06, + "loss": 0.3346, + "mean_token_accuracy": 0.8780202865600586, + "num_tokens": 422344702.0, + "step": 10879 + }, + { + "epoch": 1.3840478310647502, + "grad_norm": 2.8876900672912598, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8850983381271362, + "num_tokens": 422378925.0, + "step": 10880 + }, + { + "epoch": 1.3841750413433405, + "grad_norm": 2.5898356437683105, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9048027992248535, + "num_tokens": 422414130.0, + "step": 10881 + }, + { + "epoch": 1.384302251621931, + "grad_norm": 2.7233893871307373, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8830263614654541, + "num_tokens": 422455928.0, + "step": 10882 + }, + { + "epoch": 1.3844294619005215, + "grad_norm": 3.0082526206970215, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8696607947349548, + "num_tokens": 422490791.0, + "step": 10883 + }, + { + "epoch": 1.384556672179112, + "grad_norm": 2.711834669113159, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8839718103408813, + "num_tokens": 422530343.0, + "step": 10884 + }, + { + "epoch": 1.3846838824577026, + "grad_norm": 2.6859543323516846, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8917291164398193, + "num_tokens": 422567862.0, + "step": 10885 + }, + { + "epoch": 1.384811092736293, + "grad_norm": 2.563819408416748, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.898929238319397, + "num_tokens": 422604302.0, + "step": 10886 + }, + { + "epoch": 1.3849383030148836, + "grad_norm": 2.753772258758545, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8864008188247681, + "num_tokens": 422641227.0, + "step": 10887 + }, + { + "epoch": 1.3850655132934742, + "grad_norm": 2.6230967044830322, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8837642073631287, + "num_tokens": 422680955.0, + "step": 10888 + }, + { + "epoch": 1.3851927235720647, + "grad_norm": 2.8134307861328125, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8886125087738037, + "num_tokens": 422719737.0, + "step": 10889 + }, + { + "epoch": 1.385319933850655, + "grad_norm": 2.6683125495910645, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8765701055526733, + "num_tokens": 422766507.0, + "step": 10890 + }, + { + "epoch": 1.3854471441292455, + "grad_norm": 2.7635185718536377, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8921387791633606, + "num_tokens": 422803915.0, + "step": 10891 + }, + { + "epoch": 1.385574354407836, + "grad_norm": 2.4278669357299805, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8872891664505005, + "num_tokens": 422847840.0, + "step": 10892 + }, + { + "epoch": 1.3857015646864266, + "grad_norm": 2.5867550373077393, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8912619948387146, + "num_tokens": 422887512.0, + "step": 10893 + }, + { + "epoch": 1.385828774965017, + "grad_norm": 2.6164140701293945, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8918492197990417, + "num_tokens": 422923046.0, + "step": 10894 + }, + { + "epoch": 1.3859559852436076, + "grad_norm": 2.549966812133789, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8885436058044434, + "num_tokens": 422963567.0, + "step": 10895 + }, + { + "epoch": 1.3860831955221982, + "grad_norm": 2.7241897583007812, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8988296985626221, + "num_tokens": 422999361.0, + "step": 10896 + }, + { + "epoch": 1.3862104058007887, + "grad_norm": 2.6809964179992676, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.897894024848938, + "num_tokens": 423035430.0, + "step": 10897 + }, + { + "epoch": 1.3863376160793792, + "grad_norm": 2.6353886127471924, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8951332569122314, + "num_tokens": 423073647.0, + "step": 10898 + }, + { + "epoch": 1.3864648263579697, + "grad_norm": 2.340826988220215, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8976653814315796, + "num_tokens": 423116964.0, + "step": 10899 + }, + { + "epoch": 1.3865920366365603, + "grad_norm": 2.7791285514831543, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8912748694419861, + "num_tokens": 423151650.0, + "step": 10900 + }, + { + "epoch": 1.3867192469151508, + "grad_norm": 2.7774829864501953, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8965733051300049, + "num_tokens": 423184173.0, + "step": 10901 + }, + { + "epoch": 1.3868464571937413, + "grad_norm": 2.3844032287597656, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8829396367073059, + "num_tokens": 423237212.0, + "step": 10902 + }, + { + "epoch": 1.3869736674723319, + "grad_norm": 2.7021396160125732, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.8969929218292236, + "num_tokens": 423272911.0, + "step": 10903 + }, + { + "epoch": 1.3871008777509224, + "grad_norm": 2.576521396636963, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8902204036712646, + "num_tokens": 423313904.0, + "step": 10904 + }, + { + "epoch": 1.387228088029513, + "grad_norm": 2.72963547706604, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.877427875995636, + "num_tokens": 423356226.0, + "step": 10905 + }, + { + "epoch": 1.3873552983081032, + "grad_norm": 2.625166177749634, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8924949169158936, + "num_tokens": 423394945.0, + "step": 10906 + }, + { + "epoch": 1.3874825085866938, + "grad_norm": 2.549689292907715, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8898917436599731, + "num_tokens": 423437626.0, + "step": 10907 + }, + { + "epoch": 1.3876097188652843, + "grad_norm": 2.8707709312438965, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8794268369674683, + "num_tokens": 423472529.0, + "step": 10908 + }, + { + "epoch": 1.3877369291438748, + "grad_norm": 2.84407114982605, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8701073527336121, + "num_tokens": 423512446.0, + "step": 10909 + }, + { + "epoch": 1.3878641394224653, + "grad_norm": 2.6283695697784424, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.889792799949646, + "num_tokens": 423554140.0, + "step": 10910 + }, + { + "epoch": 1.3879913497010559, + "grad_norm": 2.570258855819702, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8934853076934814, + "num_tokens": 423595328.0, + "step": 10911 + }, + { + "epoch": 1.3881185599796464, + "grad_norm": 2.537355899810791, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9040100574493408, + "num_tokens": 423630771.0, + "step": 10912 + }, + { + "epoch": 1.388245770258237, + "grad_norm": 2.755887508392334, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8765692710876465, + "num_tokens": 423673920.0, + "step": 10913 + }, + { + "epoch": 1.3883729805368275, + "grad_norm": 2.6813037395477295, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8956380486488342, + "num_tokens": 423710932.0, + "step": 10914 + }, + { + "epoch": 1.3885001908154178, + "grad_norm": 2.4576308727264404, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8861629366874695, + "num_tokens": 423754669.0, + "step": 10915 + }, + { + "epoch": 1.3886274010940083, + "grad_norm": 2.9346108436584473, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8841160535812378, + "num_tokens": 423789748.0, + "step": 10916 + }, + { + "epoch": 1.3887546113725988, + "grad_norm": 2.838513135910034, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8740406036376953, + "num_tokens": 423829693.0, + "step": 10917 + }, + { + "epoch": 1.3888818216511893, + "grad_norm": 2.789388418197632, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8940764665603638, + "num_tokens": 423865197.0, + "step": 10918 + }, + { + "epoch": 1.3890090319297799, + "grad_norm": 2.6558098793029785, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8936753869056702, + "num_tokens": 423901653.0, + "step": 10919 + }, + { + "epoch": 1.3891362422083704, + "grad_norm": 2.7358896732330322, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8949636220932007, + "num_tokens": 423934234.0, + "step": 10920 + }, + { + "epoch": 1.389263452486961, + "grad_norm": 2.613276720046997, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8979785442352295, + "num_tokens": 423976648.0, + "step": 10921 + }, + { + "epoch": 1.3893906627655515, + "grad_norm": 2.7395238876342773, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8905290365219116, + "num_tokens": 424012838.0, + "step": 10922 + }, + { + "epoch": 1.389517873044142, + "grad_norm": 2.5787291526794434, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8814067840576172, + "num_tokens": 424051214.0, + "step": 10923 + }, + { + "epoch": 1.3896450833227325, + "grad_norm": 2.5468263626098633, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.8818608522415161, + "num_tokens": 424095347.0, + "step": 10924 + }, + { + "epoch": 1.389772293601323, + "grad_norm": 2.5602872371673584, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8989739418029785, + "num_tokens": 424133896.0, + "step": 10925 + }, + { + "epoch": 1.3898995038799136, + "grad_norm": 2.6933908462524414, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8916254043579102, + "num_tokens": 424172400.0, + "step": 10926 + }, + { + "epoch": 1.390026714158504, + "grad_norm": 2.4243602752685547, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8897880911827087, + "num_tokens": 424215415.0, + "step": 10927 + }, + { + "epoch": 1.3901539244370946, + "grad_norm": 2.708354949951172, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8907651305198669, + "num_tokens": 424251711.0, + "step": 10928 + }, + { + "epoch": 1.3902811347156852, + "grad_norm": 2.7778022289276123, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.892376184463501, + "num_tokens": 424287641.0, + "step": 10929 + }, + { + "epoch": 1.3904083449942755, + "grad_norm": 2.7641682624816895, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8864696025848389, + "num_tokens": 424323592.0, + "step": 10930 + }, + { + "epoch": 1.390535555272866, + "grad_norm": 2.530653715133667, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8788431882858276, + "num_tokens": 424367994.0, + "step": 10931 + }, + { + "epoch": 1.3906627655514565, + "grad_norm": 2.4625887870788574, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8952350616455078, + "num_tokens": 424410503.0, + "step": 10932 + }, + { + "epoch": 1.390789975830047, + "grad_norm": 2.4845898151397705, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8901416063308716, + "num_tokens": 424450773.0, + "step": 10933 + }, + { + "epoch": 1.3909171861086376, + "grad_norm": 2.667342185974121, + "learning_rate": 1e-06, + "loss": 0.3387, + "mean_token_accuracy": 0.8798651695251465, + "num_tokens": 424495128.0, + "step": 10934 + }, + { + "epoch": 1.391044396387228, + "grad_norm": 2.5810494422912598, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8907867074012756, + "num_tokens": 424533083.0, + "step": 10935 + }, + { + "epoch": 1.3911716066658186, + "grad_norm": 2.8821818828582764, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8861870765686035, + "num_tokens": 424571172.0, + "step": 10936 + }, + { + "epoch": 1.3912988169444092, + "grad_norm": 2.6159098148345947, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8877074122428894, + "num_tokens": 424610485.0, + "step": 10937 + }, + { + "epoch": 1.3914260272229997, + "grad_norm": 2.704993963241577, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.882486879825592, + "num_tokens": 424651650.0, + "step": 10938 + }, + { + "epoch": 1.39155323750159, + "grad_norm": 2.6939244270324707, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8834978342056274, + "num_tokens": 424690688.0, + "step": 10939 + }, + { + "epoch": 1.3916804477801805, + "grad_norm": 2.5115466117858887, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.889899730682373, + "num_tokens": 424735656.0, + "step": 10940 + }, + { + "epoch": 1.391807658058771, + "grad_norm": 2.938763380050659, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8803687691688538, + "num_tokens": 424769010.0, + "step": 10941 + }, + { + "epoch": 1.3919348683373616, + "grad_norm": 2.682800054550171, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.884064793586731, + "num_tokens": 424802670.0, + "step": 10942 + }, + { + "epoch": 1.392062078615952, + "grad_norm": 2.67177677154541, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8900026082992554, + "num_tokens": 424841095.0, + "step": 10943 + }, + { + "epoch": 1.3921892888945426, + "grad_norm": 2.7274463176727295, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8842675089836121, + "num_tokens": 424881657.0, + "step": 10944 + }, + { + "epoch": 1.3923164991731332, + "grad_norm": 2.544881582260132, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8807603120803833, + "num_tokens": 424927693.0, + "step": 10945 + }, + { + "epoch": 1.3924437094517237, + "grad_norm": 2.594727039337158, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8798901438713074, + "num_tokens": 424966294.0, + "step": 10946 + }, + { + "epoch": 1.3925709197303142, + "grad_norm": 2.3781700134277344, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9043024182319641, + "num_tokens": 425006073.0, + "step": 10947 + }, + { + "epoch": 1.3926981300089047, + "grad_norm": 3.025916576385498, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8811807632446289, + "num_tokens": 425040821.0, + "step": 10948 + }, + { + "epoch": 1.3928253402874953, + "grad_norm": 2.401665210723877, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9051662683486938, + "num_tokens": 425082838.0, + "step": 10949 + }, + { + "epoch": 1.3929525505660858, + "grad_norm": 2.6007070541381836, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8895893096923828, + "num_tokens": 425121484.0, + "step": 10950 + }, + { + "epoch": 1.3930797608446763, + "grad_norm": 2.784566879272461, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8947649598121643, + "num_tokens": 425154584.0, + "step": 10951 + }, + { + "epoch": 1.3932069711232669, + "grad_norm": 2.674518346786499, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8976049423217773, + "num_tokens": 425193316.0, + "step": 10952 + }, + { + "epoch": 1.3933341814018574, + "grad_norm": 2.51422119140625, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8915125131607056, + "num_tokens": 425236269.0, + "step": 10953 + }, + { + "epoch": 1.393461391680448, + "grad_norm": 2.768742799758911, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8823111057281494, + "num_tokens": 425276314.0, + "step": 10954 + }, + { + "epoch": 1.3935886019590382, + "grad_norm": 2.5391359329223633, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8848789930343628, + "num_tokens": 425318483.0, + "step": 10955 + }, + { + "epoch": 1.3937158122376287, + "grad_norm": 2.40440034866333, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9077185988426208, + "num_tokens": 425361235.0, + "step": 10956 + }, + { + "epoch": 1.3938430225162193, + "grad_norm": 2.5882368087768555, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8948609828948975, + "num_tokens": 425400090.0, + "step": 10957 + }, + { + "epoch": 1.3939702327948098, + "grad_norm": 2.982546806335449, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8870913982391357, + "num_tokens": 425432362.0, + "step": 10958 + }, + { + "epoch": 1.3940974430734003, + "grad_norm": 2.490809679031372, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8912622928619385, + "num_tokens": 425476527.0, + "step": 10959 + }, + { + "epoch": 1.3942246533519909, + "grad_norm": 2.75538969039917, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8861173391342163, + "num_tokens": 425515800.0, + "step": 10960 + }, + { + "epoch": 1.3943518636305814, + "grad_norm": 2.8028767108917236, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8932809829711914, + "num_tokens": 425548636.0, + "step": 10961 + }, + { + "epoch": 1.394479073909172, + "grad_norm": 2.5872421264648438, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8964976072311401, + "num_tokens": 425586575.0, + "step": 10962 + }, + { + "epoch": 1.3946062841877624, + "grad_norm": 2.6797847747802734, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8934317827224731, + "num_tokens": 425624030.0, + "step": 10963 + }, + { + "epoch": 1.3947334944663528, + "grad_norm": 2.591456651687622, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.9001786708831787, + "num_tokens": 425662741.0, + "step": 10964 + }, + { + "epoch": 1.3948607047449433, + "grad_norm": 2.7595226764678955, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.880579948425293, + "num_tokens": 425703535.0, + "step": 10965 + }, + { + "epoch": 1.3949879150235338, + "grad_norm": 2.6469926834106445, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8880798816680908, + "num_tokens": 425745252.0, + "step": 10966 + }, + { + "epoch": 1.3951151253021243, + "grad_norm": 2.6455767154693604, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9011335372924805, + "num_tokens": 425783218.0, + "step": 10967 + }, + { + "epoch": 1.3952423355807149, + "grad_norm": 2.8262059688568115, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8816498517990112, + "num_tokens": 425822423.0, + "step": 10968 + }, + { + "epoch": 1.3953695458593054, + "grad_norm": 2.6096067428588867, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8863484859466553, + "num_tokens": 425866010.0, + "step": 10969 + }, + { + "epoch": 1.395496756137896, + "grad_norm": 2.3613197803497314, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8982882499694824, + "num_tokens": 425911830.0, + "step": 10970 + }, + { + "epoch": 1.3956239664164865, + "grad_norm": 2.7984423637390137, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8908774852752686, + "num_tokens": 425946730.0, + "step": 10971 + }, + { + "epoch": 1.395751176695077, + "grad_norm": 2.7150051593780518, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8783022165298462, + "num_tokens": 425988243.0, + "step": 10972 + }, + { + "epoch": 1.3958783869736675, + "grad_norm": 2.4302120208740234, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8859405517578125, + "num_tokens": 426031412.0, + "step": 10973 + }, + { + "epoch": 1.396005597252258, + "grad_norm": 2.864889144897461, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8854935169219971, + "num_tokens": 426069261.0, + "step": 10974 + }, + { + "epoch": 1.3961328075308486, + "grad_norm": 2.56272292137146, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.886760950088501, + "num_tokens": 426111244.0, + "step": 10975 + }, + { + "epoch": 1.396260017809439, + "grad_norm": 2.4358489513397217, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9007938504219055, + "num_tokens": 426152579.0, + "step": 10976 + }, + { + "epoch": 1.3963872280880296, + "grad_norm": 2.7265872955322266, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8866541385650635, + "num_tokens": 426191658.0, + "step": 10977 + }, + { + "epoch": 1.3965144383666201, + "grad_norm": 2.509603500366211, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8995846509933472, + "num_tokens": 426233862.0, + "step": 10978 + }, + { + "epoch": 1.3966416486452105, + "grad_norm": 2.8808441162109375, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.875635027885437, + "num_tokens": 426270777.0, + "step": 10979 + }, + { + "epoch": 1.396768858923801, + "grad_norm": 2.5220885276794434, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.9004350304603577, + "num_tokens": 426312899.0, + "step": 10980 + }, + { + "epoch": 1.3968960692023915, + "grad_norm": 2.4810919761657715, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8945590257644653, + "num_tokens": 426351423.0, + "step": 10981 + }, + { + "epoch": 1.397023279480982, + "grad_norm": 2.7175486087799072, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8851129412651062, + "num_tokens": 426387554.0, + "step": 10982 + }, + { + "epoch": 1.3971504897595726, + "grad_norm": 2.783402681350708, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8833536505699158, + "num_tokens": 426427191.0, + "step": 10983 + }, + { + "epoch": 1.397277700038163, + "grad_norm": 2.3405563831329346, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9008294343948364, + "num_tokens": 426469457.0, + "step": 10984 + }, + { + "epoch": 1.3974049103167536, + "grad_norm": 2.611427068710327, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8871066570281982, + "num_tokens": 426512570.0, + "step": 10985 + }, + { + "epoch": 1.3975321205953442, + "grad_norm": 3.0247745513916016, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8986758589744568, + "num_tokens": 426544700.0, + "step": 10986 + }, + { + "epoch": 1.3976593308739347, + "grad_norm": 3.0588488578796387, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.889866828918457, + "num_tokens": 426579882.0, + "step": 10987 + }, + { + "epoch": 1.397786541152525, + "grad_norm": 2.608551263809204, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8898541927337646, + "num_tokens": 426625413.0, + "step": 10988 + }, + { + "epoch": 1.3979137514311155, + "grad_norm": 2.660294532775879, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8983298540115356, + "num_tokens": 426664199.0, + "step": 10989 + }, + { + "epoch": 1.398040961709706, + "grad_norm": 2.6721889972686768, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8761395215988159, + "num_tokens": 426706424.0, + "step": 10990 + }, + { + "epoch": 1.3981681719882966, + "grad_norm": 2.5114517211914062, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8988255858421326, + "num_tokens": 426748342.0, + "step": 10991 + }, + { + "epoch": 1.398295382266887, + "grad_norm": 2.8572492599487305, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8847814202308655, + "num_tokens": 426780381.0, + "step": 10992 + }, + { + "epoch": 1.3984225925454776, + "grad_norm": 2.4341042041778564, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.892734944820404, + "num_tokens": 426822525.0, + "step": 10993 + }, + { + "epoch": 1.3985498028240682, + "grad_norm": 2.52162504196167, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8953416347503662, + "num_tokens": 426861785.0, + "step": 10994 + }, + { + "epoch": 1.3986770131026587, + "grad_norm": 2.77823543548584, + "learning_rate": 1e-06, + "loss": 0.3687, + "mean_token_accuracy": 0.8655409812927246, + "num_tokens": 426903763.0, + "step": 10995 + }, + { + "epoch": 1.3988042233812492, + "grad_norm": 2.6579806804656982, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.891501247882843, + "num_tokens": 426940410.0, + "step": 10996 + }, + { + "epoch": 1.3989314336598397, + "grad_norm": 2.7495498657226562, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8939971327781677, + "num_tokens": 426978256.0, + "step": 10997 + }, + { + "epoch": 1.3990586439384303, + "grad_norm": 3.012638568878174, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8963536024093628, + "num_tokens": 427007869.0, + "step": 10998 + }, + { + "epoch": 1.3991858542170208, + "grad_norm": 2.592921018600464, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8804188966751099, + "num_tokens": 427054039.0, + "step": 10999 + }, + { + "epoch": 1.3993130644956113, + "grad_norm": 2.7528772354125977, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8836690187454224, + "num_tokens": 427093015.0, + "step": 11000 + }, + { + "epoch": 1.3994402747742019, + "grad_norm": 2.6167337894439697, + "learning_rate": 1e-06, + "loss": 0.3523, + "mean_token_accuracy": 0.873318076133728, + "num_tokens": 427141100.0, + "step": 11001 + }, + { + "epoch": 1.3995674850527924, + "grad_norm": 2.9486117362976074, + "learning_rate": 1e-06, + "loss": 0.3526, + "mean_token_accuracy": 0.876788318157196, + "num_tokens": 427176377.0, + "step": 11002 + }, + { + "epoch": 1.399694695331383, + "grad_norm": 2.5644874572753906, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.891830563545227, + "num_tokens": 427216686.0, + "step": 11003 + }, + { + "epoch": 1.3998219056099732, + "grad_norm": 2.5984904766082764, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.896574854850769, + "num_tokens": 427255807.0, + "step": 11004 + }, + { + "epoch": 1.3999491158885637, + "grad_norm": 2.6231396198272705, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8866894245147705, + "num_tokens": 427294087.0, + "step": 11005 + }, + { + "epoch": 1.4000763261671543, + "grad_norm": 2.7391862869262695, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8898505568504333, + "num_tokens": 427329862.0, + "step": 11006 + }, + { + "epoch": 1.4002035364457448, + "grad_norm": 2.568711519241333, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8904881477355957, + "num_tokens": 427372303.0, + "step": 11007 + }, + { + "epoch": 1.4003307467243353, + "grad_norm": 2.614957571029663, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8986815810203552, + "num_tokens": 427408885.0, + "step": 11008 + }, + { + "epoch": 1.4004579570029259, + "grad_norm": 2.4677295684814453, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8926352262496948, + "num_tokens": 427450389.0, + "step": 11009 + }, + { + "epoch": 1.4005851672815164, + "grad_norm": 2.5510354042053223, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8902212381362915, + "num_tokens": 427489280.0, + "step": 11010 + }, + { + "epoch": 1.400712377560107, + "grad_norm": 2.3807077407836914, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8896427154541016, + "num_tokens": 427532735.0, + "step": 11011 + }, + { + "epoch": 1.4008395878386974, + "grad_norm": 2.318289041519165, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.901341438293457, + "num_tokens": 427577262.0, + "step": 11012 + }, + { + "epoch": 1.4009667981172877, + "grad_norm": 2.703566312789917, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9038069248199463, + "num_tokens": 427611088.0, + "step": 11013 + }, + { + "epoch": 1.4010940083958783, + "grad_norm": 2.7740776538848877, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8923091888427734, + "num_tokens": 427646035.0, + "step": 11014 + }, + { + "epoch": 1.4012212186744688, + "grad_norm": 2.974442720413208, + "learning_rate": 1e-06, + "loss": 0.3717, + "mean_token_accuracy": 0.8691685199737549, + "num_tokens": 427679963.0, + "step": 11015 + }, + { + "epoch": 1.4013484289530593, + "grad_norm": 2.7483022212982178, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8924041390419006, + "num_tokens": 427715918.0, + "step": 11016 + }, + { + "epoch": 1.4014756392316499, + "grad_norm": 2.386439561843872, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8943544030189514, + "num_tokens": 427760795.0, + "step": 11017 + }, + { + "epoch": 1.4016028495102404, + "grad_norm": 2.8187718391418457, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8856866359710693, + "num_tokens": 427797355.0, + "step": 11018 + }, + { + "epoch": 1.401730059788831, + "grad_norm": 2.5782647132873535, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8886612057685852, + "num_tokens": 427839518.0, + "step": 11019 + }, + { + "epoch": 1.4018572700674214, + "grad_norm": 2.564389228820801, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8895564079284668, + "num_tokens": 427881876.0, + "step": 11020 + }, + { + "epoch": 1.401984480346012, + "grad_norm": 2.666746139526367, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8948763012886047, + "num_tokens": 427916156.0, + "step": 11021 + }, + { + "epoch": 1.4021116906246025, + "grad_norm": 2.9003241062164307, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8742539882659912, + "num_tokens": 427954807.0, + "step": 11022 + }, + { + "epoch": 1.402238900903193, + "grad_norm": 2.6750550270080566, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8906817436218262, + "num_tokens": 427993638.0, + "step": 11023 + }, + { + "epoch": 1.4023661111817836, + "grad_norm": 2.9774842262268066, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8898809552192688, + "num_tokens": 428027552.0, + "step": 11024 + }, + { + "epoch": 1.402493321460374, + "grad_norm": 2.59041166305542, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8925697803497314, + "num_tokens": 428068138.0, + "step": 11025 + }, + { + "epoch": 1.4026205317389646, + "grad_norm": 2.8601932525634766, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8837108612060547, + "num_tokens": 428103566.0, + "step": 11026 + }, + { + "epoch": 1.4027477420175551, + "grad_norm": 2.432196855545044, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8931745290756226, + "num_tokens": 428148976.0, + "step": 11027 + }, + { + "epoch": 1.4028749522961454, + "grad_norm": 2.611619234085083, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8838397264480591, + "num_tokens": 428189857.0, + "step": 11028 + }, + { + "epoch": 1.403002162574736, + "grad_norm": 2.7043566703796387, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8848263025283813, + "num_tokens": 428230174.0, + "step": 11029 + }, + { + "epoch": 1.4031293728533265, + "grad_norm": 2.3641915321350098, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.8972620964050293, + "num_tokens": 428274689.0, + "step": 11030 + }, + { + "epoch": 1.403256583131917, + "grad_norm": 2.6446406841278076, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8953895568847656, + "num_tokens": 428311760.0, + "step": 11031 + }, + { + "epoch": 1.4033837934105076, + "grad_norm": 2.5454533100128174, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8967061042785645, + "num_tokens": 428349602.0, + "step": 11032 + }, + { + "epoch": 1.403511003689098, + "grad_norm": 2.894501209259033, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.885829746723175, + "num_tokens": 428385592.0, + "step": 11033 + }, + { + "epoch": 1.4036382139676886, + "grad_norm": 2.7202365398406982, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8911950588226318, + "num_tokens": 428422926.0, + "step": 11034 + }, + { + "epoch": 1.4037654242462791, + "grad_norm": 2.8647563457489014, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8769304752349854, + "num_tokens": 428459859.0, + "step": 11035 + }, + { + "epoch": 1.4038926345248697, + "grad_norm": 2.4867117404937744, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9014613628387451, + "num_tokens": 428498095.0, + "step": 11036 + }, + { + "epoch": 1.40401984480346, + "grad_norm": 2.6452784538269043, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9020677208900452, + "num_tokens": 428534759.0, + "step": 11037 + }, + { + "epoch": 1.4041470550820505, + "grad_norm": 2.7287681102752686, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8836572766304016, + "num_tokens": 428573278.0, + "step": 11038 + }, + { + "epoch": 1.404274265360641, + "grad_norm": 2.805098295211792, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8887925744056702, + "num_tokens": 428608663.0, + "step": 11039 + }, + { + "epoch": 1.4044014756392316, + "grad_norm": 2.4275224208831787, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.900542676448822, + "num_tokens": 428652321.0, + "step": 11040 + }, + { + "epoch": 1.404528685917822, + "grad_norm": 2.6938626766204834, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.88927161693573, + "num_tokens": 428695210.0, + "step": 11041 + }, + { + "epoch": 1.4046558961964126, + "grad_norm": 2.5586490631103516, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8964921832084656, + "num_tokens": 428733555.0, + "step": 11042 + }, + { + "epoch": 1.4047831064750032, + "grad_norm": 2.726222038269043, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.884803295135498, + "num_tokens": 428773520.0, + "step": 11043 + }, + { + "epoch": 1.4049103167535937, + "grad_norm": 2.99147629737854, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8799203634262085, + "num_tokens": 428808227.0, + "step": 11044 + }, + { + "epoch": 1.4050375270321842, + "grad_norm": 2.6256866455078125, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.885995626449585, + "num_tokens": 428851981.0, + "step": 11045 + }, + { + "epoch": 1.4051647373107747, + "grad_norm": 2.876574993133545, + "learning_rate": 1e-06, + "loss": 0.3328, + "mean_token_accuracy": 0.8823029398918152, + "num_tokens": 428889569.0, + "step": 11046 + }, + { + "epoch": 1.4052919475893653, + "grad_norm": 2.8436481952667236, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8867838382720947, + "num_tokens": 428921083.0, + "step": 11047 + }, + { + "epoch": 1.4054191578679558, + "grad_norm": 2.560497760772705, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8986225724220276, + "num_tokens": 428960551.0, + "step": 11048 + }, + { + "epoch": 1.4055463681465463, + "grad_norm": 2.6206746101379395, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.903965175151825, + "num_tokens": 428994636.0, + "step": 11049 + }, + { + "epoch": 1.4056735784251368, + "grad_norm": 2.454758882522583, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8940250277519226, + "num_tokens": 429038426.0, + "step": 11050 + }, + { + "epoch": 1.4058007887037274, + "grad_norm": 2.499433994293213, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.894814670085907, + "num_tokens": 429079190.0, + "step": 11051 + }, + { + "epoch": 1.405927998982318, + "grad_norm": 2.7994425296783447, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9003097414970398, + "num_tokens": 429113052.0, + "step": 11052 + }, + { + "epoch": 1.4060552092609082, + "grad_norm": 2.375443696975708, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8945388197898865, + "num_tokens": 429157723.0, + "step": 11053 + }, + { + "epoch": 1.4061824195394987, + "grad_norm": 2.725112199783325, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8874033689498901, + "num_tokens": 429195851.0, + "step": 11054 + }, + { + "epoch": 1.4063096298180893, + "grad_norm": 2.3687400817871094, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9005725383758545, + "num_tokens": 429239933.0, + "step": 11055 + }, + { + "epoch": 1.4064368400966798, + "grad_norm": 2.407470941543579, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9000052213668823, + "num_tokens": 429279739.0, + "step": 11056 + }, + { + "epoch": 1.4065640503752703, + "grad_norm": 2.868777275085449, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8942214250564575, + "num_tokens": 429312316.0, + "step": 11057 + }, + { + "epoch": 1.4066912606538609, + "grad_norm": 2.883274555206299, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8820095062255859, + "num_tokens": 429346952.0, + "step": 11058 + }, + { + "epoch": 1.4068184709324514, + "grad_norm": 2.7157764434814453, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8967490196228027, + "num_tokens": 429385326.0, + "step": 11059 + }, + { + "epoch": 1.406945681211042, + "grad_norm": 2.7245266437530518, + "learning_rate": 1e-06, + "loss": 0.3577, + "mean_token_accuracy": 0.8741716146469116, + "num_tokens": 429428899.0, + "step": 11060 + }, + { + "epoch": 1.4070728914896324, + "grad_norm": 2.91444993019104, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8866638541221619, + "num_tokens": 429464364.0, + "step": 11061 + }, + { + "epoch": 1.4072001017682227, + "grad_norm": 2.7182087898254395, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8933271169662476, + "num_tokens": 429501987.0, + "step": 11062 + }, + { + "epoch": 1.4073273120468133, + "grad_norm": 2.8173928260803223, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8924544453620911, + "num_tokens": 429534654.0, + "step": 11063 + }, + { + "epoch": 1.4074545223254038, + "grad_norm": 3.3213021755218506, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8949645161628723, + "num_tokens": 429560173.0, + "step": 11064 + }, + { + "epoch": 1.4075817326039943, + "grad_norm": 2.9363515377044678, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8800259828567505, + "num_tokens": 429594781.0, + "step": 11065 + }, + { + "epoch": 1.4077089428825849, + "grad_norm": 2.4588866233825684, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8987116813659668, + "num_tokens": 429636967.0, + "step": 11066 + }, + { + "epoch": 1.4078361531611754, + "grad_norm": 2.654831647872925, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8943334817886353, + "num_tokens": 429673425.0, + "step": 11067 + }, + { + "epoch": 1.407963363439766, + "grad_norm": 2.4712913036346436, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8770982027053833, + "num_tokens": 429718474.0, + "step": 11068 + }, + { + "epoch": 1.4080905737183564, + "grad_norm": 2.7747178077697754, + "learning_rate": 1e-06, + "loss": 0.3507, + "mean_token_accuracy": 0.8796268701553345, + "num_tokens": 429756183.0, + "step": 11069 + }, + { + "epoch": 1.408217783996947, + "grad_norm": 2.8023390769958496, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8835859894752502, + "num_tokens": 429792775.0, + "step": 11070 + }, + { + "epoch": 1.4083449942755375, + "grad_norm": 2.484182834625244, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8998157978057861, + "num_tokens": 429832878.0, + "step": 11071 + }, + { + "epoch": 1.408472204554128, + "grad_norm": 2.794539213180542, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8913350701332092, + "num_tokens": 429870739.0, + "step": 11072 + }, + { + "epoch": 1.4085994148327186, + "grad_norm": 2.4962399005889893, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.889721155166626, + "num_tokens": 429914083.0, + "step": 11073 + }, + { + "epoch": 1.408726625111309, + "grad_norm": 2.352414846420288, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8928706049919128, + "num_tokens": 429964503.0, + "step": 11074 + }, + { + "epoch": 1.4088538353898996, + "grad_norm": 2.527740955352783, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9099566340446472, + "num_tokens": 429999041.0, + "step": 11075 + }, + { + "epoch": 1.4089810456684901, + "grad_norm": 2.3770875930786133, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9069423675537109, + "num_tokens": 430046224.0, + "step": 11076 + }, + { + "epoch": 1.4091082559470804, + "grad_norm": 2.9424731731414795, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8884192109107971, + "num_tokens": 430079820.0, + "step": 11077 + }, + { + "epoch": 1.409235466225671, + "grad_norm": 2.570716142654419, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8946597576141357, + "num_tokens": 430119400.0, + "step": 11078 + }, + { + "epoch": 1.4093626765042615, + "grad_norm": 2.768054723739624, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8998215198516846, + "num_tokens": 430152945.0, + "step": 11079 + }, + { + "epoch": 1.409489886782852, + "grad_norm": 2.4924426078796387, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9063481092453003, + "num_tokens": 430194651.0, + "step": 11080 + }, + { + "epoch": 1.4096170970614426, + "grad_norm": 2.208376407623291, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9018856883049011, + "num_tokens": 430241342.0, + "step": 11081 + }, + { + "epoch": 1.409744307340033, + "grad_norm": 2.8115110397338867, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8980796933174133, + "num_tokens": 430272837.0, + "step": 11082 + }, + { + "epoch": 1.4098715176186236, + "grad_norm": 2.6993701457977295, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8833756446838379, + "num_tokens": 430314123.0, + "step": 11083 + }, + { + "epoch": 1.4099987278972141, + "grad_norm": 2.710329532623291, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8991527557373047, + "num_tokens": 430348028.0, + "step": 11084 + }, + { + "epoch": 1.4101259381758047, + "grad_norm": 2.3999733924865723, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8943384885787964, + "num_tokens": 430393045.0, + "step": 11085 + }, + { + "epoch": 1.410253148454395, + "grad_norm": 2.5078766345977783, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8813489675521851, + "num_tokens": 430440530.0, + "step": 11086 + }, + { + "epoch": 1.4103803587329855, + "grad_norm": 2.6440343856811523, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8878492116928101, + "num_tokens": 430485797.0, + "step": 11087 + }, + { + "epoch": 1.410507569011576, + "grad_norm": 2.7022292613983154, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8888869285583496, + "num_tokens": 430523802.0, + "step": 11088 + }, + { + "epoch": 1.4106347792901666, + "grad_norm": 2.746161699295044, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8928753733634949, + "num_tokens": 430558740.0, + "step": 11089 + }, + { + "epoch": 1.410761989568757, + "grad_norm": 2.595526933670044, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8852368593215942, + "num_tokens": 430598116.0, + "step": 11090 + }, + { + "epoch": 1.4108891998473476, + "grad_norm": 3.1870644092559814, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8917229175567627, + "num_tokens": 430625716.0, + "step": 11091 + }, + { + "epoch": 1.4110164101259381, + "grad_norm": 2.800725221633911, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8736679553985596, + "num_tokens": 430665340.0, + "step": 11092 + }, + { + "epoch": 1.4111436204045287, + "grad_norm": 2.7948262691497803, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8895525932312012, + "num_tokens": 430700832.0, + "step": 11093 + }, + { + "epoch": 1.4112708306831192, + "grad_norm": 2.5159809589385986, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8877600431442261, + "num_tokens": 430744385.0, + "step": 11094 + }, + { + "epoch": 1.4113980409617097, + "grad_norm": 3.103447198867798, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.873043417930603, + "num_tokens": 430779691.0, + "step": 11095 + }, + { + "epoch": 1.4115252512403003, + "grad_norm": 2.7377166748046875, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8953975439071655, + "num_tokens": 430812993.0, + "step": 11096 + }, + { + "epoch": 1.4116524615188908, + "grad_norm": 2.4928762912750244, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8914716243743896, + "num_tokens": 430853641.0, + "step": 11097 + }, + { + "epoch": 1.4117796717974813, + "grad_norm": 2.5151805877685547, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8929107189178467, + "num_tokens": 430894421.0, + "step": 11098 + }, + { + "epoch": 1.4119068820760718, + "grad_norm": 2.56766676902771, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8903636336326599, + "num_tokens": 430935673.0, + "step": 11099 + }, + { + "epoch": 1.4120340923546624, + "grad_norm": 2.6993656158447266, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8879294991493225, + "num_tokens": 430975242.0, + "step": 11100 + }, + { + "epoch": 1.412161302633253, + "grad_norm": 2.7812163829803467, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8786655068397522, + "num_tokens": 431016329.0, + "step": 11101 + }, + { + "epoch": 1.4122885129118432, + "grad_norm": 2.672510862350464, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8943129777908325, + "num_tokens": 431051637.0, + "step": 11102 + }, + { + "epoch": 1.4124157231904337, + "grad_norm": 2.8458175659179688, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8783164620399475, + "num_tokens": 431091718.0, + "step": 11103 + }, + { + "epoch": 1.4125429334690243, + "grad_norm": 2.736095905303955, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8791680932044983, + "num_tokens": 431130864.0, + "step": 11104 + }, + { + "epoch": 1.4126701437476148, + "grad_norm": 2.634181261062622, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8843090534210205, + "num_tokens": 431169178.0, + "step": 11105 + }, + { + "epoch": 1.4127973540262053, + "grad_norm": 2.594320058822632, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.885887861251831, + "num_tokens": 431207948.0, + "step": 11106 + }, + { + "epoch": 1.4129245643047958, + "grad_norm": 2.570352554321289, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8999326229095459, + "num_tokens": 431243186.0, + "step": 11107 + }, + { + "epoch": 1.4130517745833864, + "grad_norm": 2.7465591430664062, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8933480978012085, + "num_tokens": 431280221.0, + "step": 11108 + }, + { + "epoch": 1.413178984861977, + "grad_norm": 2.707155704498291, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8933586478233337, + "num_tokens": 431317255.0, + "step": 11109 + }, + { + "epoch": 1.4133061951405674, + "grad_norm": 2.6040303707122803, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8862123489379883, + "num_tokens": 431356540.0, + "step": 11110 + }, + { + "epoch": 1.4134334054191577, + "grad_norm": 2.6313085556030273, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8919413089752197, + "num_tokens": 431394085.0, + "step": 11111 + }, + { + "epoch": 1.4135606156977483, + "grad_norm": 2.324920654296875, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8915809392929077, + "num_tokens": 431440009.0, + "step": 11112 + }, + { + "epoch": 1.4136878259763388, + "grad_norm": 2.9262545108795166, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8988048434257507, + "num_tokens": 431470069.0, + "step": 11113 + }, + { + "epoch": 1.4138150362549293, + "grad_norm": 2.569998025894165, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8839776515960693, + "num_tokens": 431510553.0, + "step": 11114 + }, + { + "epoch": 1.4139422465335199, + "grad_norm": 2.2002689838409424, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9048264026641846, + "num_tokens": 431557875.0, + "step": 11115 + }, + { + "epoch": 1.4140694568121104, + "grad_norm": 2.631373882293701, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8906632661819458, + "num_tokens": 431596555.0, + "step": 11116 + }, + { + "epoch": 1.414196667090701, + "grad_norm": 2.838775634765625, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.894955575466156, + "num_tokens": 431629596.0, + "step": 11117 + }, + { + "epoch": 1.4143238773692914, + "grad_norm": 2.564805269241333, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8981204628944397, + "num_tokens": 431665834.0, + "step": 11118 + }, + { + "epoch": 1.414451087647882, + "grad_norm": 2.5072879791259766, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.885485053062439, + "num_tokens": 431711479.0, + "step": 11119 + }, + { + "epoch": 1.4145782979264725, + "grad_norm": 2.5507760047912598, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.8996777534484863, + "num_tokens": 431751553.0, + "step": 11120 + }, + { + "epoch": 1.414705508205063, + "grad_norm": 2.6871719360351562, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8947780132293701, + "num_tokens": 431790451.0, + "step": 11121 + }, + { + "epoch": 1.4148327184836536, + "grad_norm": 2.8812038898468018, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.88720703125, + "num_tokens": 431825388.0, + "step": 11122 + }, + { + "epoch": 1.414959928762244, + "grad_norm": 2.6381518840789795, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8877692222595215, + "num_tokens": 431863982.0, + "step": 11123 + }, + { + "epoch": 1.4150871390408346, + "grad_norm": 2.520440101623535, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8883543610572815, + "num_tokens": 431907516.0, + "step": 11124 + }, + { + "epoch": 1.4152143493194251, + "grad_norm": 2.7283785343170166, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8770666718482971, + "num_tokens": 431946926.0, + "step": 11125 + }, + { + "epoch": 1.4153415595980154, + "grad_norm": 3.0323383808135986, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8875051736831665, + "num_tokens": 431978032.0, + "step": 11126 + }, + { + "epoch": 1.415468769876606, + "grad_norm": 2.801053524017334, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.886871337890625, + "num_tokens": 432016703.0, + "step": 11127 + }, + { + "epoch": 1.4155959801551965, + "grad_norm": 2.6451148986816406, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8941927552223206, + "num_tokens": 432054126.0, + "step": 11128 + }, + { + "epoch": 1.415723190433787, + "grad_norm": 2.6762897968292236, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.907024621963501, + "num_tokens": 432089505.0, + "step": 11129 + }, + { + "epoch": 1.4158504007123776, + "grad_norm": 2.432645797729492, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8887232542037964, + "num_tokens": 432135474.0, + "step": 11130 + }, + { + "epoch": 1.415977610990968, + "grad_norm": 2.4513509273529053, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.8991875648498535, + "num_tokens": 432177725.0, + "step": 11131 + }, + { + "epoch": 1.4161048212695586, + "grad_norm": 2.517056941986084, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8920013904571533, + "num_tokens": 432220670.0, + "step": 11132 + }, + { + "epoch": 1.4162320315481491, + "grad_norm": 2.819934129714966, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8958604335784912, + "num_tokens": 432256510.0, + "step": 11133 + }, + { + "epoch": 1.4163592418267397, + "grad_norm": 2.4472427368164062, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8870808482170105, + "num_tokens": 432300395.0, + "step": 11134 + }, + { + "epoch": 1.41648645210533, + "grad_norm": 2.556594133377075, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8978466987609863, + "num_tokens": 432341543.0, + "step": 11135 + }, + { + "epoch": 1.4166136623839205, + "grad_norm": 2.5948193073272705, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8800517320632935, + "num_tokens": 432384233.0, + "step": 11136 + }, + { + "epoch": 1.416740872662511, + "grad_norm": 2.5917959213256836, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8976157903671265, + "num_tokens": 432422241.0, + "step": 11137 + }, + { + "epoch": 1.4168680829411016, + "grad_norm": 2.3942949771881104, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8958420753479004, + "num_tokens": 432464608.0, + "step": 11138 + }, + { + "epoch": 1.416995293219692, + "grad_norm": 2.4679453372955322, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.897134006023407, + "num_tokens": 432505143.0, + "step": 11139 + }, + { + "epoch": 1.4171225034982826, + "grad_norm": 2.611323595046997, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8926562666893005, + "num_tokens": 432542734.0, + "step": 11140 + }, + { + "epoch": 1.4172497137768731, + "grad_norm": 2.893794536590576, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8845787048339844, + "num_tokens": 432577223.0, + "step": 11141 + }, + { + "epoch": 1.4173769240554637, + "grad_norm": 2.537902593612671, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8830500245094299, + "num_tokens": 432620152.0, + "step": 11142 + }, + { + "epoch": 1.4175041343340542, + "grad_norm": 2.7568814754486084, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8864091038703918, + "num_tokens": 432658911.0, + "step": 11143 + }, + { + "epoch": 1.4176313446126447, + "grad_norm": 2.4060113430023193, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8934330940246582, + "num_tokens": 432701398.0, + "step": 11144 + }, + { + "epoch": 1.4177585548912353, + "grad_norm": 2.7553975582122803, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.8959651589393616, + "num_tokens": 432735704.0, + "step": 11145 + }, + { + "epoch": 1.4178857651698258, + "grad_norm": 2.816920757293701, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.884046196937561, + "num_tokens": 432773288.0, + "step": 11146 + }, + { + "epoch": 1.4180129754484163, + "grad_norm": 2.604825973510742, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9003647565841675, + "num_tokens": 432811239.0, + "step": 11147 + }, + { + "epoch": 1.4181401857270068, + "grad_norm": 2.607741594314575, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.891305148601532, + "num_tokens": 432851500.0, + "step": 11148 + }, + { + "epoch": 1.4182673960055974, + "grad_norm": 2.6466245651245117, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8805285692214966, + "num_tokens": 432894143.0, + "step": 11149 + }, + { + "epoch": 1.418394606284188, + "grad_norm": 2.6849045753479004, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8933112621307373, + "num_tokens": 432935693.0, + "step": 11150 + }, + { + "epoch": 1.4185218165627782, + "grad_norm": 2.682572603225708, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9048402309417725, + "num_tokens": 432969664.0, + "step": 11151 + }, + { + "epoch": 1.4186490268413687, + "grad_norm": 2.749379873275757, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8810784816741943, + "num_tokens": 433010070.0, + "step": 11152 + }, + { + "epoch": 1.4187762371199593, + "grad_norm": 2.763620615005493, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8770643472671509, + "num_tokens": 433049881.0, + "step": 11153 + }, + { + "epoch": 1.4189034473985498, + "grad_norm": 2.38558292388916, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.895328164100647, + "num_tokens": 433096504.0, + "step": 11154 + }, + { + "epoch": 1.4190306576771403, + "grad_norm": 2.607818841934204, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8954236507415771, + "num_tokens": 433135982.0, + "step": 11155 + }, + { + "epoch": 1.4191578679557308, + "grad_norm": 2.990166425704956, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8841363191604614, + "num_tokens": 433169973.0, + "step": 11156 + }, + { + "epoch": 1.4192850782343214, + "grad_norm": 2.61859393119812, + "learning_rate": 1e-06, + "loss": 0.3415, + "mean_token_accuracy": 0.8790680170059204, + "num_tokens": 433213626.0, + "step": 11157 + }, + { + "epoch": 1.419412288512912, + "grad_norm": 2.766617774963379, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9018237590789795, + "num_tokens": 433245832.0, + "step": 11158 + }, + { + "epoch": 1.4195394987915024, + "grad_norm": 2.815603256225586, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.88994300365448, + "num_tokens": 433283308.0, + "step": 11159 + }, + { + "epoch": 1.4196667090700927, + "grad_norm": 2.901437759399414, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.892814040184021, + "num_tokens": 433322617.0, + "step": 11160 + }, + { + "epoch": 1.4197939193486833, + "grad_norm": 2.6403982639312744, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.8823736906051636, + "num_tokens": 433362193.0, + "step": 11161 + }, + { + "epoch": 1.4199211296272738, + "grad_norm": 2.719511032104492, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8914918899536133, + "num_tokens": 433400095.0, + "step": 11162 + }, + { + "epoch": 1.4200483399058643, + "grad_norm": 2.752018690109253, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8956291675567627, + "num_tokens": 433432273.0, + "step": 11163 + }, + { + "epoch": 1.4201755501844548, + "grad_norm": 2.587517023086548, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8895336389541626, + "num_tokens": 433468363.0, + "step": 11164 + }, + { + "epoch": 1.4203027604630454, + "grad_norm": 2.5222465991973877, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8968283534049988, + "num_tokens": 433513794.0, + "step": 11165 + }, + { + "epoch": 1.420429970741636, + "grad_norm": 3.0576891899108887, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8895294666290283, + "num_tokens": 433554418.0, + "step": 11166 + }, + { + "epoch": 1.4205571810202264, + "grad_norm": 2.7620341777801514, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8846790194511414, + "num_tokens": 433596095.0, + "step": 11167 + }, + { + "epoch": 1.420684391298817, + "grad_norm": 2.5966989994049072, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8957483768463135, + "num_tokens": 433634677.0, + "step": 11168 + }, + { + "epoch": 1.4208116015774075, + "grad_norm": 2.7661657333374023, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8829134702682495, + "num_tokens": 433674170.0, + "step": 11169 + }, + { + "epoch": 1.420938811855998, + "grad_norm": 2.573016405105591, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8863207101821899, + "num_tokens": 433718017.0, + "step": 11170 + }, + { + "epoch": 1.4210660221345885, + "grad_norm": 2.4319753646850586, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.8998984694480896, + "num_tokens": 433761083.0, + "step": 11171 + }, + { + "epoch": 1.421193232413179, + "grad_norm": 2.5137734413146973, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8926401138305664, + "num_tokens": 433803605.0, + "step": 11172 + }, + { + "epoch": 1.4213204426917696, + "grad_norm": 2.9090309143066406, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.893970787525177, + "num_tokens": 433836295.0, + "step": 11173 + }, + { + "epoch": 1.4214476529703601, + "grad_norm": 2.6942059993743896, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8885709643363953, + "num_tokens": 433878963.0, + "step": 11174 + }, + { + "epoch": 1.4215748632489504, + "grad_norm": 2.6088905334472656, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8862854242324829, + "num_tokens": 433918161.0, + "step": 11175 + }, + { + "epoch": 1.421702073527541, + "grad_norm": 2.707681179046631, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8826855421066284, + "num_tokens": 433956412.0, + "step": 11176 + }, + { + "epoch": 1.4218292838061315, + "grad_norm": 2.8207013607025146, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8763976693153381, + "num_tokens": 433996282.0, + "step": 11177 + }, + { + "epoch": 1.421956494084722, + "grad_norm": 2.5833613872528076, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.870050847530365, + "num_tokens": 434036337.0, + "step": 11178 + }, + { + "epoch": 1.4220837043633126, + "grad_norm": 2.7835915088653564, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8856490850448608, + "num_tokens": 434073644.0, + "step": 11179 + }, + { + "epoch": 1.422210914641903, + "grad_norm": 2.872969627380371, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8829096555709839, + "num_tokens": 434108724.0, + "step": 11180 + }, + { + "epoch": 1.4223381249204936, + "grad_norm": 2.8379409313201904, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8908184766769409, + "num_tokens": 434140042.0, + "step": 11181 + }, + { + "epoch": 1.4224653351990841, + "grad_norm": 2.388033628463745, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8952720165252686, + "num_tokens": 434182631.0, + "step": 11182 + }, + { + "epoch": 1.4225925454776747, + "grad_norm": 2.4084393978118896, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9039550423622131, + "num_tokens": 434222357.0, + "step": 11183 + }, + { + "epoch": 1.422719755756265, + "grad_norm": 2.7630012035369873, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8926443457603455, + "num_tokens": 434257610.0, + "step": 11184 + }, + { + "epoch": 1.4228469660348555, + "grad_norm": 2.88153076171875, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8768247961997986, + "num_tokens": 434293898.0, + "step": 11185 + }, + { + "epoch": 1.422974176313446, + "grad_norm": 2.5735976696014404, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8960216045379639, + "num_tokens": 434334797.0, + "step": 11186 + }, + { + "epoch": 1.4231013865920366, + "grad_norm": 2.5600852966308594, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8837690949440002, + "num_tokens": 434380023.0, + "step": 11187 + }, + { + "epoch": 1.423228596870627, + "grad_norm": 2.6968038082122803, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8859947919845581, + "num_tokens": 434424086.0, + "step": 11188 + }, + { + "epoch": 1.4233558071492176, + "grad_norm": 2.6306874752044678, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8981329202651978, + "num_tokens": 434460950.0, + "step": 11189 + }, + { + "epoch": 1.4234830174278081, + "grad_norm": 2.7212624549865723, + "learning_rate": 1e-06, + "loss": 0.3475, + "mean_token_accuracy": 0.8763723373413086, + "num_tokens": 434503427.0, + "step": 11190 + }, + { + "epoch": 1.4236102277063987, + "grad_norm": 2.727956533432007, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8917425870895386, + "num_tokens": 434539891.0, + "step": 11191 + }, + { + "epoch": 1.4237374379849892, + "grad_norm": 2.624821901321411, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8886090517044067, + "num_tokens": 434578998.0, + "step": 11192 + }, + { + "epoch": 1.4238646482635797, + "grad_norm": 2.9583616256713867, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8817178010940552, + "num_tokens": 434613607.0, + "step": 11193 + }, + { + "epoch": 1.4239918585421703, + "grad_norm": 2.5787997245788574, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8964758515357971, + "num_tokens": 434651769.0, + "step": 11194 + }, + { + "epoch": 1.4241190688207608, + "grad_norm": 2.619866132736206, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8968095779418945, + "num_tokens": 434691289.0, + "step": 11195 + }, + { + "epoch": 1.4242462790993513, + "grad_norm": 2.954650402069092, + "learning_rate": 1e-06, + "loss": 0.345, + "mean_token_accuracy": 0.8769327402114868, + "num_tokens": 434729477.0, + "step": 11196 + }, + { + "epoch": 1.4243734893779418, + "grad_norm": 2.8453376293182373, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.8779757022857666, + "num_tokens": 434766471.0, + "step": 11197 + }, + { + "epoch": 1.4245006996565324, + "grad_norm": 2.6421070098876953, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8922208547592163, + "num_tokens": 434806275.0, + "step": 11198 + }, + { + "epoch": 1.424627909935123, + "grad_norm": 2.549534559249878, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8990694880485535, + "num_tokens": 434844317.0, + "step": 11199 + }, + { + "epoch": 1.4247551202137132, + "grad_norm": 2.4672586917877197, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8990942239761353, + "num_tokens": 434887861.0, + "step": 11200 + }, + { + "epoch": 1.4248823304923037, + "grad_norm": 2.4992728233337402, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8972152471542358, + "num_tokens": 434932154.0, + "step": 11201 + }, + { + "epoch": 1.4250095407708943, + "grad_norm": 2.334613084793091, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9043067097663879, + "num_tokens": 434975802.0, + "step": 11202 + }, + { + "epoch": 1.4251367510494848, + "grad_norm": 2.556180238723755, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8997696042060852, + "num_tokens": 435015518.0, + "step": 11203 + }, + { + "epoch": 1.4252639613280753, + "grad_norm": 2.573341131210327, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8776007890701294, + "num_tokens": 435062671.0, + "step": 11204 + }, + { + "epoch": 1.4253911716066658, + "grad_norm": 2.638262987136841, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9080790281295776, + "num_tokens": 435094281.0, + "step": 11205 + }, + { + "epoch": 1.4255183818852564, + "grad_norm": 2.560258626937866, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8910836577415466, + "num_tokens": 435133262.0, + "step": 11206 + }, + { + "epoch": 1.425645592163847, + "grad_norm": 2.5385236740112305, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9004390835762024, + "num_tokens": 435171053.0, + "step": 11207 + }, + { + "epoch": 1.4257728024424374, + "grad_norm": 2.6070139408111572, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8844617605209351, + "num_tokens": 435212104.0, + "step": 11208 + }, + { + "epoch": 1.4259000127210277, + "grad_norm": 2.6450552940368652, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8939061760902405, + "num_tokens": 435251809.0, + "step": 11209 + }, + { + "epoch": 1.4260272229996183, + "grad_norm": 2.785179853439331, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8841133713722229, + "num_tokens": 435287791.0, + "step": 11210 + }, + { + "epoch": 1.4261544332782088, + "grad_norm": 2.8992645740509033, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8953389525413513, + "num_tokens": 435321965.0, + "step": 11211 + }, + { + "epoch": 1.4262816435567993, + "grad_norm": 2.7822015285491943, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8910421133041382, + "num_tokens": 435358507.0, + "step": 11212 + }, + { + "epoch": 1.4264088538353898, + "grad_norm": 2.5386581420898438, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8957101106643677, + "num_tokens": 435399977.0, + "step": 11213 + }, + { + "epoch": 1.4265360641139804, + "grad_norm": 2.613473653793335, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8865119814872742, + "num_tokens": 435442935.0, + "step": 11214 + }, + { + "epoch": 1.426663274392571, + "grad_norm": 2.7810943126678467, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8903070688247681, + "num_tokens": 435479259.0, + "step": 11215 + }, + { + "epoch": 1.4267904846711614, + "grad_norm": 2.9213478565216064, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8926230669021606, + "num_tokens": 435510129.0, + "step": 11216 + }, + { + "epoch": 1.426917694949752, + "grad_norm": 2.5652332305908203, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.890129566192627, + "num_tokens": 435549571.0, + "step": 11217 + }, + { + "epoch": 1.4270449052283425, + "grad_norm": 2.8316681385040283, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8888643980026245, + "num_tokens": 435584996.0, + "step": 11218 + }, + { + "epoch": 1.427172115506933, + "grad_norm": 2.6422536373138428, + "learning_rate": 1e-06, + "loss": 0.3454, + "mean_token_accuracy": 0.8787409663200378, + "num_tokens": 435635193.0, + "step": 11219 + }, + { + "epoch": 1.4272993257855235, + "grad_norm": 2.7608535289764404, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8804534077644348, + "num_tokens": 435674856.0, + "step": 11220 + }, + { + "epoch": 1.427426536064114, + "grad_norm": 2.515583038330078, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8992397785186768, + "num_tokens": 435713481.0, + "step": 11221 + }, + { + "epoch": 1.4275537463427046, + "grad_norm": 2.806806802749634, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8767989873886108, + "num_tokens": 435754511.0, + "step": 11222 + }, + { + "epoch": 1.4276809566212951, + "grad_norm": 2.894402503967285, + "learning_rate": 1e-06, + "loss": 0.3489, + "mean_token_accuracy": 0.8737862706184387, + "num_tokens": 435791521.0, + "step": 11223 + }, + { + "epoch": 1.4278081668998854, + "grad_norm": 2.296471357345581, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9063992500305176, + "num_tokens": 435837620.0, + "step": 11224 + }, + { + "epoch": 1.427935377178476, + "grad_norm": 3.0452041625976562, + "learning_rate": 1e-06, + "loss": 0.3567, + "mean_token_accuracy": 0.8723862171173096, + "num_tokens": 435871386.0, + "step": 11225 + }, + { + "epoch": 1.4280625874570665, + "grad_norm": 2.5899336338043213, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8820753693580627, + "num_tokens": 435911917.0, + "step": 11226 + }, + { + "epoch": 1.428189797735657, + "grad_norm": 2.9321582317352295, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.888315737247467, + "num_tokens": 435944478.0, + "step": 11227 + }, + { + "epoch": 1.4283170080142475, + "grad_norm": 2.85648250579834, + "learning_rate": 1e-06, + "loss": 0.3603, + "mean_token_accuracy": 0.8705649971961975, + "num_tokens": 435982257.0, + "step": 11228 + }, + { + "epoch": 1.428444218292838, + "grad_norm": 2.7011489868164062, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9010069966316223, + "num_tokens": 436017980.0, + "step": 11229 + }, + { + "epoch": 1.4285714285714286, + "grad_norm": 2.4864556789398193, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.890900731086731, + "num_tokens": 436060240.0, + "step": 11230 + }, + { + "epoch": 1.4286986388500191, + "grad_norm": 2.6063613891601562, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8800104856491089, + "num_tokens": 436101918.0, + "step": 11231 + }, + { + "epoch": 1.4288258491286097, + "grad_norm": 2.4647653102874756, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8970869183540344, + "num_tokens": 436141297.0, + "step": 11232 + }, + { + "epoch": 1.4289530594072, + "grad_norm": 2.5476884841918945, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9006851315498352, + "num_tokens": 436176309.0, + "step": 11233 + }, + { + "epoch": 1.4290802696857905, + "grad_norm": 2.663290023803711, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8927446603775024, + "num_tokens": 436208970.0, + "step": 11234 + }, + { + "epoch": 1.429207479964381, + "grad_norm": 2.524162530899048, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8980963230133057, + "num_tokens": 436247191.0, + "step": 11235 + }, + { + "epoch": 1.4293346902429716, + "grad_norm": 2.7755234241485596, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.901070237159729, + "num_tokens": 436286344.0, + "step": 11236 + }, + { + "epoch": 1.429461900521562, + "grad_norm": 2.8697104454040527, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8792955279350281, + "num_tokens": 436326126.0, + "step": 11237 + }, + { + "epoch": 1.4295891108001526, + "grad_norm": 2.7241733074188232, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8853920698165894, + "num_tokens": 436362617.0, + "step": 11238 + }, + { + "epoch": 1.4297163210787431, + "grad_norm": 2.797762155532837, + "learning_rate": 1e-06, + "loss": 0.3465, + "mean_token_accuracy": 0.8772473335266113, + "num_tokens": 436401732.0, + "step": 11239 + }, + { + "epoch": 1.4298435313573337, + "grad_norm": 2.4788482189178467, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8980388045310974, + "num_tokens": 436443554.0, + "step": 11240 + }, + { + "epoch": 1.4299707416359242, + "grad_norm": 2.6868531703948975, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8911102414131165, + "num_tokens": 436481818.0, + "step": 11241 + }, + { + "epoch": 1.4300979519145147, + "grad_norm": 2.4332473278045654, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8915250301361084, + "num_tokens": 436524999.0, + "step": 11242 + }, + { + "epoch": 1.4302251621931052, + "grad_norm": 2.5911691188812256, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8874939680099487, + "num_tokens": 436563400.0, + "step": 11243 + }, + { + "epoch": 1.4303523724716958, + "grad_norm": 2.4230356216430664, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8885015249252319, + "num_tokens": 436607310.0, + "step": 11244 + }, + { + "epoch": 1.4304795827502863, + "grad_norm": 2.618830680847168, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8954228162765503, + "num_tokens": 436643401.0, + "step": 11245 + }, + { + "epoch": 1.4306067930288768, + "grad_norm": 2.397369146347046, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8982199430465698, + "num_tokens": 436686119.0, + "step": 11246 + }, + { + "epoch": 1.4307340033074674, + "grad_norm": 2.637002944946289, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.887464165687561, + "num_tokens": 436722704.0, + "step": 11247 + }, + { + "epoch": 1.430861213586058, + "grad_norm": 2.7496302127838135, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8861838579177856, + "num_tokens": 436758842.0, + "step": 11248 + }, + { + "epoch": 1.4309884238646482, + "grad_norm": 2.5164167881011963, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8892055153846741, + "num_tokens": 436803625.0, + "step": 11249 + }, + { + "epoch": 1.4311156341432387, + "grad_norm": 2.833205461502075, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8843386173248291, + "num_tokens": 436839080.0, + "step": 11250 + }, + { + "epoch": 1.4312428444218293, + "grad_norm": 2.689070701599121, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8999466896057129, + "num_tokens": 436874798.0, + "step": 11251 + }, + { + "epoch": 1.4313700547004198, + "grad_norm": 2.5340516567230225, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8970666527748108, + "num_tokens": 436914315.0, + "step": 11252 + }, + { + "epoch": 1.4314972649790103, + "grad_norm": 2.87500262260437, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8767120838165283, + "num_tokens": 436957636.0, + "step": 11253 + }, + { + "epoch": 1.4316244752576008, + "grad_norm": 2.618544340133667, + "learning_rate": 1e-06, + "loss": 0.3471, + "mean_token_accuracy": 0.8785011768341064, + "num_tokens": 437002124.0, + "step": 11254 + }, + { + "epoch": 1.4317516855361914, + "grad_norm": 2.9294486045837402, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8745337128639221, + "num_tokens": 437036668.0, + "step": 11255 + }, + { + "epoch": 1.431878895814782, + "grad_norm": 2.5114235877990723, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.888597846031189, + "num_tokens": 437078777.0, + "step": 11256 + }, + { + "epoch": 1.4320061060933724, + "grad_norm": 2.7413253784179688, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8931440114974976, + "num_tokens": 437116744.0, + "step": 11257 + }, + { + "epoch": 1.4321333163719627, + "grad_norm": 2.508655548095703, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8953185677528381, + "num_tokens": 437157672.0, + "step": 11258 + }, + { + "epoch": 1.4322605266505533, + "grad_norm": 2.423011064529419, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8966382741928101, + "num_tokens": 437204153.0, + "step": 11259 + }, + { + "epoch": 1.4323877369291438, + "grad_norm": 2.7590432167053223, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8727213144302368, + "num_tokens": 437241827.0, + "step": 11260 + }, + { + "epoch": 1.4325149472077343, + "grad_norm": 2.608412981033325, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.886886715888977, + "num_tokens": 437282424.0, + "step": 11261 + }, + { + "epoch": 1.4326421574863248, + "grad_norm": 2.608485221862793, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8876522779464722, + "num_tokens": 437321692.0, + "step": 11262 + }, + { + "epoch": 1.4327693677649154, + "grad_norm": 2.4541707038879395, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.89666748046875, + "num_tokens": 437361306.0, + "step": 11263 + }, + { + "epoch": 1.432896578043506, + "grad_norm": 2.6211254596710205, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8934109807014465, + "num_tokens": 437400722.0, + "step": 11264 + }, + { + "epoch": 1.4330237883220964, + "grad_norm": 2.6472744941711426, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8921591639518738, + "num_tokens": 437438743.0, + "step": 11265 + }, + { + "epoch": 1.433150998600687, + "grad_norm": 2.8013081550598145, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8869931697845459, + "num_tokens": 437474148.0, + "step": 11266 + }, + { + "epoch": 1.4332782088792775, + "grad_norm": 2.8376049995422363, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8837757110595703, + "num_tokens": 437509941.0, + "step": 11267 + }, + { + "epoch": 1.433405419157868, + "grad_norm": 2.8264448642730713, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9021780490875244, + "num_tokens": 437540514.0, + "step": 11268 + }, + { + "epoch": 1.4335326294364585, + "grad_norm": 2.813793420791626, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8782446980476379, + "num_tokens": 437580394.0, + "step": 11269 + }, + { + "epoch": 1.433659839715049, + "grad_norm": 2.471280336380005, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8883652687072754, + "num_tokens": 437625354.0, + "step": 11270 + }, + { + "epoch": 1.4337870499936396, + "grad_norm": 2.872313976287842, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8987633585929871, + "num_tokens": 437658245.0, + "step": 11271 + }, + { + "epoch": 1.4339142602722301, + "grad_norm": 2.6055188179016113, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8814970850944519, + "num_tokens": 437699665.0, + "step": 11272 + }, + { + "epoch": 1.4340414705508204, + "grad_norm": 2.597654342651367, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8933299779891968, + "num_tokens": 437740949.0, + "step": 11273 + }, + { + "epoch": 1.434168680829411, + "grad_norm": 2.682681083679199, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8841335773468018, + "num_tokens": 437781746.0, + "step": 11274 + }, + { + "epoch": 1.4342958911080015, + "grad_norm": 2.7015058994293213, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8880218267440796, + "num_tokens": 437821047.0, + "step": 11275 + }, + { + "epoch": 1.434423101386592, + "grad_norm": 2.606171131134033, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8964552283287048, + "num_tokens": 437858218.0, + "step": 11276 + }, + { + "epoch": 1.4345503116651825, + "grad_norm": 2.928067684173584, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8877006769180298, + "num_tokens": 437891385.0, + "step": 11277 + }, + { + "epoch": 1.434677521943773, + "grad_norm": 2.6307482719421387, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.887474000453949, + "num_tokens": 437929825.0, + "step": 11278 + }, + { + "epoch": 1.4348047322223636, + "grad_norm": 2.573957681655884, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8902938365936279, + "num_tokens": 437972541.0, + "step": 11279 + }, + { + "epoch": 1.4349319425009541, + "grad_norm": 2.7863545417785645, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8897933959960938, + "num_tokens": 438007148.0, + "step": 11280 + }, + { + "epoch": 1.4350591527795447, + "grad_norm": 2.5359957218170166, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.894101619720459, + "num_tokens": 438046282.0, + "step": 11281 + }, + { + "epoch": 1.435186363058135, + "grad_norm": 2.566297769546509, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8869776129722595, + "num_tokens": 438086906.0, + "step": 11282 + }, + { + "epoch": 1.4353135733367255, + "grad_norm": 2.5078349113464355, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8991349935531616, + "num_tokens": 438124375.0, + "step": 11283 + }, + { + "epoch": 1.435440783615316, + "grad_norm": 2.731599807739258, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8830572962760925, + "num_tokens": 438164833.0, + "step": 11284 + }, + { + "epoch": 1.4355679938939065, + "grad_norm": 2.489856719970703, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8933315873146057, + "num_tokens": 438205875.0, + "step": 11285 + }, + { + "epoch": 1.435695204172497, + "grad_norm": 2.6467599868774414, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8917967677116394, + "num_tokens": 438243626.0, + "step": 11286 + }, + { + "epoch": 1.4358224144510876, + "grad_norm": 2.644102096557617, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9015246033668518, + "num_tokens": 438281420.0, + "step": 11287 + }, + { + "epoch": 1.4359496247296781, + "grad_norm": 2.6989591121673584, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8950309753417969, + "num_tokens": 438319349.0, + "step": 11288 + }, + { + "epoch": 1.4360768350082687, + "grad_norm": 2.368974208831787, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9113755226135254, + "num_tokens": 438362544.0, + "step": 11289 + }, + { + "epoch": 1.4362040452868592, + "grad_norm": 2.4660890102386475, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8951895236968994, + "num_tokens": 438405771.0, + "step": 11290 + }, + { + "epoch": 1.4363312555654497, + "grad_norm": 2.401210069656372, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.8970786929130554, + "num_tokens": 438446616.0, + "step": 11291 + }, + { + "epoch": 1.4364584658440402, + "grad_norm": 2.5833868980407715, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.894569993019104, + "num_tokens": 438487103.0, + "step": 11292 + }, + { + "epoch": 1.4365856761226308, + "grad_norm": 2.637401819229126, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8894063830375671, + "num_tokens": 438527451.0, + "step": 11293 + }, + { + "epoch": 1.4367128864012213, + "grad_norm": 2.7565619945526123, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8855788111686707, + "num_tokens": 438569514.0, + "step": 11294 + }, + { + "epoch": 1.4368400966798118, + "grad_norm": 2.6878631114959717, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.888361394405365, + "num_tokens": 438607447.0, + "step": 11295 + }, + { + "epoch": 1.4369673069584024, + "grad_norm": 2.443434000015259, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9021871089935303, + "num_tokens": 438646300.0, + "step": 11296 + }, + { + "epoch": 1.4370945172369929, + "grad_norm": 2.4430673122406006, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.8999927043914795, + "num_tokens": 438687902.0, + "step": 11297 + }, + { + "epoch": 1.4372217275155832, + "grad_norm": 2.4266369342803955, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8920798301696777, + "num_tokens": 438731965.0, + "step": 11298 + }, + { + "epoch": 1.4373489377941737, + "grad_norm": 2.515993356704712, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8992159962654114, + "num_tokens": 438773099.0, + "step": 11299 + }, + { + "epoch": 1.4374761480727642, + "grad_norm": 2.681309938430786, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8907822370529175, + "num_tokens": 438813170.0, + "step": 11300 + }, + { + "epoch": 1.4376033583513548, + "grad_norm": 2.4663565158843994, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8953503370285034, + "num_tokens": 438852817.0, + "step": 11301 + }, + { + "epoch": 1.4377305686299453, + "grad_norm": 2.8432629108428955, + "learning_rate": 1e-06, + "loss": 0.3386, + "mean_token_accuracy": 0.879749059677124, + "num_tokens": 438893732.0, + "step": 11302 + }, + { + "epoch": 1.4378577789085358, + "grad_norm": 2.8105204105377197, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8928306102752686, + "num_tokens": 438931371.0, + "step": 11303 + }, + { + "epoch": 1.4379849891871264, + "grad_norm": 2.61557936668396, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.88877934217453, + "num_tokens": 438971526.0, + "step": 11304 + }, + { + "epoch": 1.438112199465717, + "grad_norm": 2.7071175575256348, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8840159177780151, + "num_tokens": 439014109.0, + "step": 11305 + }, + { + "epoch": 1.4382394097443074, + "grad_norm": 2.6421546936035156, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8922012448310852, + "num_tokens": 439055351.0, + "step": 11306 + }, + { + "epoch": 1.4383666200228977, + "grad_norm": 2.868480920791626, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8716253042221069, + "num_tokens": 439094168.0, + "step": 11307 + }, + { + "epoch": 1.4384938303014883, + "grad_norm": 2.691100835800171, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8873969912528992, + "num_tokens": 439131307.0, + "step": 11308 + }, + { + "epoch": 1.4386210405800788, + "grad_norm": 2.389420747756958, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8867838382720947, + "num_tokens": 439177354.0, + "step": 11309 + }, + { + "epoch": 1.4387482508586693, + "grad_norm": 2.638850212097168, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8834019899368286, + "num_tokens": 439214596.0, + "step": 11310 + }, + { + "epoch": 1.4388754611372598, + "grad_norm": 2.5378670692443848, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8899645209312439, + "num_tokens": 439256055.0, + "step": 11311 + }, + { + "epoch": 1.4390026714158504, + "grad_norm": 2.9331893920898438, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8909488916397095, + "num_tokens": 439288315.0, + "step": 11312 + }, + { + "epoch": 1.439129881694441, + "grad_norm": 2.537405014038086, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9024766087532043, + "num_tokens": 439325602.0, + "step": 11313 + }, + { + "epoch": 1.4392570919730314, + "grad_norm": 2.641296625137329, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9005692005157471, + "num_tokens": 439362516.0, + "step": 11314 + }, + { + "epoch": 1.439384302251622, + "grad_norm": 2.5904018878936768, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8871986865997314, + "num_tokens": 439400150.0, + "step": 11315 + }, + { + "epoch": 1.4395115125302125, + "grad_norm": 2.444326400756836, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9095718264579773, + "num_tokens": 439434854.0, + "step": 11316 + }, + { + "epoch": 1.439638722808803, + "grad_norm": 2.687380075454712, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8919018507003784, + "num_tokens": 439475109.0, + "step": 11317 + }, + { + "epoch": 1.4397659330873935, + "grad_norm": 2.760685920715332, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8859401941299438, + "num_tokens": 439514612.0, + "step": 11318 + }, + { + "epoch": 1.439893143365984, + "grad_norm": 2.6715621948242188, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.892156720161438, + "num_tokens": 439552364.0, + "step": 11319 + }, + { + "epoch": 1.4400203536445746, + "grad_norm": 3.029466152191162, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8958288431167603, + "num_tokens": 439587840.0, + "step": 11320 + }, + { + "epoch": 1.4401475639231651, + "grad_norm": 2.6849770545959473, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.901664674282074, + "num_tokens": 439623304.0, + "step": 11321 + }, + { + "epoch": 1.4402747742017554, + "grad_norm": 2.7590279579162598, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8953160643577576, + "num_tokens": 439659814.0, + "step": 11322 + }, + { + "epoch": 1.440401984480346, + "grad_norm": 2.7967023849487305, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8793407082557678, + "num_tokens": 439695430.0, + "step": 11323 + }, + { + "epoch": 1.4405291947589365, + "grad_norm": 2.6886181831359863, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8926589488983154, + "num_tokens": 439736201.0, + "step": 11324 + }, + { + "epoch": 1.440656405037527, + "grad_norm": 2.5390031337738037, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9038451313972473, + "num_tokens": 439775833.0, + "step": 11325 + }, + { + "epoch": 1.4407836153161175, + "grad_norm": 2.7290425300598145, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8877745270729065, + "num_tokens": 439812968.0, + "step": 11326 + }, + { + "epoch": 1.440910825594708, + "grad_norm": 2.9045519828796387, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.87457674741745, + "num_tokens": 439850986.0, + "step": 11327 + }, + { + "epoch": 1.4410380358732986, + "grad_norm": 2.8216073513031006, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9052242636680603, + "num_tokens": 439882211.0, + "step": 11328 + }, + { + "epoch": 1.4411652461518891, + "grad_norm": 2.9470059871673584, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8924281001091003, + "num_tokens": 439914361.0, + "step": 11329 + }, + { + "epoch": 1.4412924564304797, + "grad_norm": 2.7238616943359375, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.8977483510971069, + "num_tokens": 439949481.0, + "step": 11330 + }, + { + "epoch": 1.44141966670907, + "grad_norm": 2.786431074142456, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.884749174118042, + "num_tokens": 439985468.0, + "step": 11331 + }, + { + "epoch": 1.4415468769876605, + "grad_norm": 2.696855306625366, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8937281370162964, + "num_tokens": 440022457.0, + "step": 11332 + }, + { + "epoch": 1.441674087266251, + "grad_norm": 2.7045116424560547, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8821993470191956, + "num_tokens": 440065300.0, + "step": 11333 + }, + { + "epoch": 1.4418012975448415, + "grad_norm": 2.385772466659546, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8889132142066956, + "num_tokens": 440113620.0, + "step": 11334 + }, + { + "epoch": 1.441928507823432, + "grad_norm": 2.620663642883301, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8913089036941528, + "num_tokens": 440157271.0, + "step": 11335 + }, + { + "epoch": 1.4420557181020226, + "grad_norm": 2.6769509315490723, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8939440846443176, + "num_tokens": 440193368.0, + "step": 11336 + }, + { + "epoch": 1.4421829283806131, + "grad_norm": 2.4572722911834717, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8938348293304443, + "num_tokens": 440234658.0, + "step": 11337 + }, + { + "epoch": 1.4423101386592037, + "grad_norm": 2.710785150527954, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.895294189453125, + "num_tokens": 440268138.0, + "step": 11338 + }, + { + "epoch": 1.4424373489377942, + "grad_norm": 2.7193174362182617, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8978129625320435, + "num_tokens": 440303280.0, + "step": 11339 + }, + { + "epoch": 1.4425645592163847, + "grad_norm": 2.8260765075683594, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9012959599494934, + "num_tokens": 440334482.0, + "step": 11340 + }, + { + "epoch": 1.4426917694949752, + "grad_norm": 3.263636589050293, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8917315006256104, + "num_tokens": 440373318.0, + "step": 11341 + }, + { + "epoch": 1.4428189797735658, + "grad_norm": 2.7622337341308594, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8915083408355713, + "num_tokens": 440411889.0, + "step": 11342 + }, + { + "epoch": 1.4429461900521563, + "grad_norm": 2.683912515640259, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8938158750534058, + "num_tokens": 440449827.0, + "step": 11343 + }, + { + "epoch": 1.4430734003307468, + "grad_norm": 2.9015660285949707, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8898357152938843, + "num_tokens": 440482336.0, + "step": 11344 + }, + { + "epoch": 1.4432006106093374, + "grad_norm": 2.7187933921813965, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8869354128837585, + "num_tokens": 440523315.0, + "step": 11345 + }, + { + "epoch": 1.4433278208879279, + "grad_norm": 2.5472326278686523, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8903993368148804, + "num_tokens": 440566581.0, + "step": 11346 + }, + { + "epoch": 1.4434550311665182, + "grad_norm": 2.5811853408813477, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8910847902297974, + "num_tokens": 440605497.0, + "step": 11347 + }, + { + "epoch": 1.4435822414451087, + "grad_norm": 2.5563175678253174, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8950753808021545, + "num_tokens": 440647595.0, + "step": 11348 + }, + { + "epoch": 1.4437094517236992, + "grad_norm": 2.9720304012298584, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8794045448303223, + "num_tokens": 440684362.0, + "step": 11349 + }, + { + "epoch": 1.4438366620022898, + "grad_norm": 2.571563959121704, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8866958618164062, + "num_tokens": 440728023.0, + "step": 11350 + }, + { + "epoch": 1.4439638722808803, + "grad_norm": 2.8354923725128174, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8781425952911377, + "num_tokens": 440768688.0, + "step": 11351 + }, + { + "epoch": 1.4440910825594708, + "grad_norm": 2.5705673694610596, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.893318772315979, + "num_tokens": 440809155.0, + "step": 11352 + }, + { + "epoch": 1.4442182928380614, + "grad_norm": 2.8336281776428223, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8938815593719482, + "num_tokens": 440844099.0, + "step": 11353 + }, + { + "epoch": 1.4443455031166519, + "grad_norm": 2.882023572921753, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.895422101020813, + "num_tokens": 440875001.0, + "step": 11354 + }, + { + "epoch": 1.4444727133952424, + "grad_norm": 2.3610734939575195, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.9001063108444214, + "num_tokens": 440917574.0, + "step": 11355 + }, + { + "epoch": 1.4445999236738327, + "grad_norm": 2.7163844108581543, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8924897909164429, + "num_tokens": 440953937.0, + "step": 11356 + }, + { + "epoch": 1.4447271339524232, + "grad_norm": 2.760761260986328, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.8976471424102783, + "num_tokens": 440987903.0, + "step": 11357 + }, + { + "epoch": 1.4448543442310138, + "grad_norm": 3.1230711936950684, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8904746174812317, + "num_tokens": 441019208.0, + "step": 11358 + }, + { + "epoch": 1.4449815545096043, + "grad_norm": 2.6672184467315674, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8817175626754761, + "num_tokens": 441061822.0, + "step": 11359 + }, + { + "epoch": 1.4451087647881948, + "grad_norm": 2.5391969680786133, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8872627019882202, + "num_tokens": 441103531.0, + "step": 11360 + }, + { + "epoch": 1.4452359750667854, + "grad_norm": 2.7722725868225098, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8779537677764893, + "num_tokens": 441143013.0, + "step": 11361 + }, + { + "epoch": 1.445363185345376, + "grad_norm": 2.562962532043457, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8920971155166626, + "num_tokens": 441182870.0, + "step": 11362 + }, + { + "epoch": 1.4454903956239664, + "grad_norm": 2.5434467792510986, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8874398469924927, + "num_tokens": 441223634.0, + "step": 11363 + }, + { + "epoch": 1.445617605902557, + "grad_norm": 2.5257692337036133, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8905091881752014, + "num_tokens": 441265305.0, + "step": 11364 + }, + { + "epoch": 1.4457448161811475, + "grad_norm": 2.404829502105713, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.886340320110321, + "num_tokens": 441310585.0, + "step": 11365 + }, + { + "epoch": 1.445872026459738, + "grad_norm": 2.7408602237701416, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8927589654922485, + "num_tokens": 441345072.0, + "step": 11366 + }, + { + "epoch": 1.4459992367383285, + "grad_norm": 2.506699800491333, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.892381489276886, + "num_tokens": 441385795.0, + "step": 11367 + }, + { + "epoch": 1.446126447016919, + "grad_norm": 2.590129852294922, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.902098536491394, + "num_tokens": 441420297.0, + "step": 11368 + }, + { + "epoch": 1.4462536572955096, + "grad_norm": 2.5659642219543457, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8924534916877747, + "num_tokens": 441458100.0, + "step": 11369 + }, + { + "epoch": 1.4463808675741001, + "grad_norm": 2.9871885776519775, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8888911008834839, + "num_tokens": 441489540.0, + "step": 11370 + }, + { + "epoch": 1.4465080778526904, + "grad_norm": 3.080737829208374, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.8995717167854309, + "num_tokens": 441516223.0, + "step": 11371 + }, + { + "epoch": 1.446635288131281, + "grad_norm": 2.8210642337799072, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8932132720947266, + "num_tokens": 441549419.0, + "step": 11372 + }, + { + "epoch": 1.4467624984098715, + "grad_norm": 2.643625497817993, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8974207639694214, + "num_tokens": 441587779.0, + "step": 11373 + }, + { + "epoch": 1.446889708688462, + "grad_norm": 2.9172914028167725, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8844616413116455, + "num_tokens": 441622790.0, + "step": 11374 + }, + { + "epoch": 1.4470169189670525, + "grad_norm": 2.7757441997528076, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8948911428451538, + "num_tokens": 441657941.0, + "step": 11375 + }, + { + "epoch": 1.447144129245643, + "grad_norm": 2.529109239578247, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9005133509635925, + "num_tokens": 441697285.0, + "step": 11376 + }, + { + "epoch": 1.4472713395242336, + "grad_norm": 2.8782601356506348, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8790290355682373, + "num_tokens": 441734493.0, + "step": 11377 + }, + { + "epoch": 1.4473985498028241, + "grad_norm": 2.578655481338501, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8888387680053711, + "num_tokens": 441776137.0, + "step": 11378 + }, + { + "epoch": 1.4475257600814146, + "grad_norm": 2.879387140274048, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8906694650650024, + "num_tokens": 441808396.0, + "step": 11379 + }, + { + "epoch": 1.447652970360005, + "grad_norm": 2.6407439708709717, + "learning_rate": 1e-06, + "loss": 0.2442, + "mean_token_accuracy": 0.9096869230270386, + "num_tokens": 441840112.0, + "step": 11380 + }, + { + "epoch": 1.4477801806385955, + "grad_norm": 2.7455244064331055, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8783272504806519, + "num_tokens": 441882468.0, + "step": 11381 + }, + { + "epoch": 1.447907390917186, + "grad_norm": 2.6461143493652344, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8991196155548096, + "num_tokens": 441918839.0, + "step": 11382 + }, + { + "epoch": 1.4480346011957765, + "grad_norm": 2.5656847953796387, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8923553228378296, + "num_tokens": 441961189.0, + "step": 11383 + }, + { + "epoch": 1.448161811474367, + "grad_norm": 2.600743293762207, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8966254591941833, + "num_tokens": 441998219.0, + "step": 11384 + }, + { + "epoch": 1.4482890217529576, + "grad_norm": 2.8112332820892334, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8770207762718201, + "num_tokens": 442035002.0, + "step": 11385 + }, + { + "epoch": 1.4484162320315481, + "grad_norm": 2.7100718021392822, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8759248852729797, + "num_tokens": 442073630.0, + "step": 11386 + }, + { + "epoch": 1.4485434423101387, + "grad_norm": 2.5940299034118652, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8911348581314087, + "num_tokens": 442111585.0, + "step": 11387 + }, + { + "epoch": 1.4486706525887292, + "grad_norm": 2.869258165359497, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8980580568313599, + "num_tokens": 442144738.0, + "step": 11388 + }, + { + "epoch": 1.4487978628673197, + "grad_norm": 2.4842238426208496, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.895416796207428, + "num_tokens": 442185253.0, + "step": 11389 + }, + { + "epoch": 1.4489250731459102, + "grad_norm": 2.549678087234497, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8981533050537109, + "num_tokens": 442223838.0, + "step": 11390 + }, + { + "epoch": 1.4490522834245008, + "grad_norm": 2.5128333568573, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8953647613525391, + "num_tokens": 442263496.0, + "step": 11391 + }, + { + "epoch": 1.4491794937030913, + "grad_norm": 2.6488845348358154, + "learning_rate": 1e-06, + "loss": 0.352, + "mean_token_accuracy": 0.877737820148468, + "num_tokens": 442307425.0, + "step": 11392 + }, + { + "epoch": 1.4493067039816818, + "grad_norm": 2.7315657138824463, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.880949854850769, + "num_tokens": 442349034.0, + "step": 11393 + }, + { + "epoch": 1.4494339142602723, + "grad_norm": 2.4534523487091064, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8915821313858032, + "num_tokens": 442391613.0, + "step": 11394 + }, + { + "epoch": 1.4495611245388629, + "grad_norm": 2.5634043216705322, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8980013132095337, + "num_tokens": 442431341.0, + "step": 11395 + }, + { + "epoch": 1.4496883348174532, + "grad_norm": 2.3799824714660645, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9038543105125427, + "num_tokens": 442474458.0, + "step": 11396 + }, + { + "epoch": 1.4498155450960437, + "grad_norm": 2.635190010070801, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8956855535507202, + "num_tokens": 442512568.0, + "step": 11397 + }, + { + "epoch": 1.4499427553746342, + "grad_norm": 2.8245725631713867, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8850771188735962, + "num_tokens": 442548456.0, + "step": 11398 + }, + { + "epoch": 1.4500699656532248, + "grad_norm": 2.983525276184082, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8815882205963135, + "num_tokens": 442584791.0, + "step": 11399 + }, + { + "epoch": 1.4501971759318153, + "grad_norm": 2.7117199897766113, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8914766907691956, + "num_tokens": 442625667.0, + "step": 11400 + }, + { + "epoch": 1.4503243862104058, + "grad_norm": 2.7030551433563232, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8914965391159058, + "num_tokens": 442660362.0, + "step": 11401 + }, + { + "epoch": 1.4504515964889964, + "grad_norm": 2.733762741088867, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8828980326652527, + "num_tokens": 442702739.0, + "step": 11402 + }, + { + "epoch": 1.4505788067675869, + "grad_norm": 2.872877359390259, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.896973192691803, + "num_tokens": 442735194.0, + "step": 11403 + }, + { + "epoch": 1.4507060170461774, + "grad_norm": 2.570129632949829, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8949713706970215, + "num_tokens": 442776960.0, + "step": 11404 + }, + { + "epoch": 1.4508332273247677, + "grad_norm": 2.6388919353485107, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9009389877319336, + "num_tokens": 442813825.0, + "step": 11405 + }, + { + "epoch": 1.4509604376033582, + "grad_norm": 2.6789445877075195, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8871070742607117, + "num_tokens": 442848886.0, + "step": 11406 + }, + { + "epoch": 1.4510876478819488, + "grad_norm": 2.629835605621338, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8956539034843445, + "num_tokens": 442887113.0, + "step": 11407 + }, + { + "epoch": 1.4512148581605393, + "grad_norm": 2.3887832164764404, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8895281553268433, + "num_tokens": 442933284.0, + "step": 11408 + }, + { + "epoch": 1.4513420684391298, + "grad_norm": 2.9419331550598145, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.881548285484314, + "num_tokens": 442966893.0, + "step": 11409 + }, + { + "epoch": 1.4514692787177204, + "grad_norm": 2.6875228881835938, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8938581943511963, + "num_tokens": 443002295.0, + "step": 11410 + }, + { + "epoch": 1.4515964889963109, + "grad_norm": 2.722710132598877, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.890048623085022, + "num_tokens": 443039117.0, + "step": 11411 + }, + { + "epoch": 1.4517236992749014, + "grad_norm": 2.4870429039001465, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8928901553153992, + "num_tokens": 443081662.0, + "step": 11412 + }, + { + "epoch": 1.451850909553492, + "grad_norm": 2.595174551010132, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8955751657485962, + "num_tokens": 443121962.0, + "step": 11413 + }, + { + "epoch": 1.4519781198320825, + "grad_norm": 2.6965293884277344, + "learning_rate": 1e-06, + "loss": 0.3536, + "mean_token_accuracy": 0.8761777281761169, + "num_tokens": 443163940.0, + "step": 11414 + }, + { + "epoch": 1.452105330110673, + "grad_norm": 2.9894368648529053, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8895598649978638, + "num_tokens": 443207257.0, + "step": 11415 + }, + { + "epoch": 1.4522325403892635, + "grad_norm": 2.6956050395965576, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8898496627807617, + "num_tokens": 443245294.0, + "step": 11416 + }, + { + "epoch": 1.452359750667854, + "grad_norm": 2.8951942920684814, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8833534717559814, + "num_tokens": 443278754.0, + "step": 11417 + }, + { + "epoch": 1.4524869609464446, + "grad_norm": 2.430356502532959, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8872685432434082, + "num_tokens": 443324017.0, + "step": 11418 + }, + { + "epoch": 1.4526141712250351, + "grad_norm": 2.853393793106079, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9034919738769531, + "num_tokens": 443354129.0, + "step": 11419 + }, + { + "epoch": 1.4527413815036254, + "grad_norm": 2.6387667655944824, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8892316818237305, + "num_tokens": 443393527.0, + "step": 11420 + }, + { + "epoch": 1.452868591782216, + "grad_norm": 2.640061378479004, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9029523134231567, + "num_tokens": 443432413.0, + "step": 11421 + }, + { + "epoch": 1.4529958020608065, + "grad_norm": 2.9794251918792725, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8857970237731934, + "num_tokens": 443463349.0, + "step": 11422 + }, + { + "epoch": 1.453123012339397, + "grad_norm": 2.7268877029418945, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8806511163711548, + "num_tokens": 443503912.0, + "step": 11423 + }, + { + "epoch": 1.4532502226179875, + "grad_norm": 2.710200309753418, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8867164850234985, + "num_tokens": 443545563.0, + "step": 11424 + }, + { + "epoch": 1.453377432896578, + "grad_norm": 2.472813606262207, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.8998162746429443, + "num_tokens": 443585610.0, + "step": 11425 + }, + { + "epoch": 1.4535046431751686, + "grad_norm": 2.683650493621826, + "learning_rate": 1e-06, + "loss": 0.356, + "mean_token_accuracy": 0.8743603825569153, + "num_tokens": 443627572.0, + "step": 11426 + }, + { + "epoch": 1.4536318534537591, + "grad_norm": 2.5216622352600098, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8965444564819336, + "num_tokens": 443667839.0, + "step": 11427 + }, + { + "epoch": 1.4537590637323496, + "grad_norm": 2.5325064659118652, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8834156394004822, + "num_tokens": 443707985.0, + "step": 11428 + }, + { + "epoch": 1.45388627401094, + "grad_norm": 2.493912696838379, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8999834060668945, + "num_tokens": 443750882.0, + "step": 11429 + }, + { + "epoch": 1.4540134842895305, + "grad_norm": 2.866459608078003, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8801089525222778, + "num_tokens": 443789485.0, + "step": 11430 + }, + { + "epoch": 1.454140694568121, + "grad_norm": 2.4620509147644043, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8968615531921387, + "num_tokens": 443829557.0, + "step": 11431 + }, + { + "epoch": 1.4542679048467115, + "grad_norm": 2.7599871158599854, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8891246318817139, + "num_tokens": 443862551.0, + "step": 11432 + }, + { + "epoch": 1.454395115125302, + "grad_norm": 2.723829507827759, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8876748085021973, + "num_tokens": 443896896.0, + "step": 11433 + }, + { + "epoch": 1.4545223254038926, + "grad_norm": 2.6425342559814453, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.896248996257782, + "num_tokens": 443934629.0, + "step": 11434 + }, + { + "epoch": 1.4546495356824831, + "grad_norm": 2.704118251800537, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8741748332977295, + "num_tokens": 443976331.0, + "step": 11435 + }, + { + "epoch": 1.4547767459610736, + "grad_norm": 2.786264419555664, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8691604137420654, + "num_tokens": 444014756.0, + "step": 11436 + }, + { + "epoch": 1.4549039562396642, + "grad_norm": 2.5545055866241455, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8927686214447021, + "num_tokens": 444052995.0, + "step": 11437 + }, + { + "epoch": 1.4550311665182547, + "grad_norm": 2.692261219024658, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8962439298629761, + "num_tokens": 444090296.0, + "step": 11438 + }, + { + "epoch": 1.4551583767968452, + "grad_norm": 2.5999603271484375, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8933408856391907, + "num_tokens": 444126642.0, + "step": 11439 + }, + { + "epoch": 1.4552855870754358, + "grad_norm": 2.662137508392334, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8900223970413208, + "num_tokens": 444168414.0, + "step": 11440 + }, + { + "epoch": 1.4554127973540263, + "grad_norm": 2.8321149349212646, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8871877789497375, + "num_tokens": 444206474.0, + "step": 11441 + }, + { + "epoch": 1.4555400076326168, + "grad_norm": 2.699232339859009, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9031929969787598, + "num_tokens": 444244392.0, + "step": 11442 + }, + { + "epoch": 1.4556672179112073, + "grad_norm": 2.7204885482788086, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8774427771568298, + "num_tokens": 444284753.0, + "step": 11443 + }, + { + "epoch": 1.4557944281897979, + "grad_norm": 2.8247575759887695, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8901681303977966, + "num_tokens": 444323087.0, + "step": 11444 + }, + { + "epoch": 1.4559216384683882, + "grad_norm": 2.7440831661224365, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8818188905715942, + "num_tokens": 444363056.0, + "step": 11445 + }, + { + "epoch": 1.4560488487469787, + "grad_norm": 2.6682491302490234, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8877785205841064, + "num_tokens": 444401994.0, + "step": 11446 + }, + { + "epoch": 1.4561760590255692, + "grad_norm": 2.541563034057617, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8906053304672241, + "num_tokens": 444443829.0, + "step": 11447 + }, + { + "epoch": 1.4563032693041598, + "grad_norm": 2.6875357627868652, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.891345739364624, + "num_tokens": 444481179.0, + "step": 11448 + }, + { + "epoch": 1.4564304795827503, + "grad_norm": 2.8192129135131836, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8875049948692322, + "num_tokens": 444516852.0, + "step": 11449 + }, + { + "epoch": 1.4565576898613408, + "grad_norm": 2.894547462463379, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8787415027618408, + "num_tokens": 444554641.0, + "step": 11450 + }, + { + "epoch": 1.4566849001399313, + "grad_norm": 2.6155264377593994, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8906314373016357, + "num_tokens": 444593370.0, + "step": 11451 + }, + { + "epoch": 1.4568121104185219, + "grad_norm": 2.795811891555786, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8862932324409485, + "num_tokens": 444630582.0, + "step": 11452 + }, + { + "epoch": 1.4569393206971124, + "grad_norm": 2.5808732509613037, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9012669324874878, + "num_tokens": 444669673.0, + "step": 11453 + }, + { + "epoch": 1.4570665309757027, + "grad_norm": 2.4806461334228516, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8966339826583862, + "num_tokens": 444713166.0, + "step": 11454 + }, + { + "epoch": 1.4571937412542932, + "grad_norm": 2.4542531967163086, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9020562171936035, + "num_tokens": 444753695.0, + "step": 11455 + }, + { + "epoch": 1.4573209515328838, + "grad_norm": 2.830505847930908, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.895750105381012, + "num_tokens": 444787455.0, + "step": 11456 + }, + { + "epoch": 1.4574481618114743, + "grad_norm": 2.5687079429626465, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8935344219207764, + "num_tokens": 444828309.0, + "step": 11457 + }, + { + "epoch": 1.4575753720900648, + "grad_norm": 2.446024179458618, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8970208168029785, + "num_tokens": 444870635.0, + "step": 11458 + }, + { + "epoch": 1.4577025823686554, + "grad_norm": 2.523218870162964, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8933180570602417, + "num_tokens": 444911720.0, + "step": 11459 + }, + { + "epoch": 1.4578297926472459, + "grad_norm": 2.473607301712036, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8883204460144043, + "num_tokens": 444957861.0, + "step": 11460 + }, + { + "epoch": 1.4579570029258364, + "grad_norm": 2.793215274810791, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8923951387405396, + "num_tokens": 444989316.0, + "step": 11461 + }, + { + "epoch": 1.458084213204427, + "grad_norm": 2.7444355487823486, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8905315399169922, + "num_tokens": 445024198.0, + "step": 11462 + }, + { + "epoch": 1.4582114234830175, + "grad_norm": 2.439070463180542, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8977950215339661, + "num_tokens": 445067902.0, + "step": 11463 + }, + { + "epoch": 1.458338633761608, + "grad_norm": 2.782888650894165, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.882485032081604, + "num_tokens": 445106069.0, + "step": 11464 + }, + { + "epoch": 1.4584658440401985, + "grad_norm": 3.0676300525665283, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8781982660293579, + "num_tokens": 445140426.0, + "step": 11465 + }, + { + "epoch": 1.458593054318789, + "grad_norm": 2.6210525035858154, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8816113471984863, + "num_tokens": 445185123.0, + "step": 11466 + }, + { + "epoch": 1.4587202645973796, + "grad_norm": 2.6339786052703857, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9010885953903198, + "num_tokens": 445222940.0, + "step": 11467 + }, + { + "epoch": 1.45884747487597, + "grad_norm": 2.6323816776275635, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8901548981666565, + "num_tokens": 445261603.0, + "step": 11468 + }, + { + "epoch": 1.4589746851545604, + "grad_norm": 2.585376024246216, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9106615781784058, + "num_tokens": 445295741.0, + "step": 11469 + }, + { + "epoch": 1.459101895433151, + "grad_norm": 4.806787490844727, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8959224224090576, + "num_tokens": 445329651.0, + "step": 11470 + }, + { + "epoch": 1.4592291057117415, + "grad_norm": 2.623004913330078, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9022583961486816, + "num_tokens": 445364897.0, + "step": 11471 + }, + { + "epoch": 1.459356315990332, + "grad_norm": 2.622098684310913, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8895545601844788, + "num_tokens": 445404518.0, + "step": 11472 + }, + { + "epoch": 1.4594835262689225, + "grad_norm": 2.7197725772857666, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.893622875213623, + "num_tokens": 445441748.0, + "step": 11473 + }, + { + "epoch": 1.459610736547513, + "grad_norm": 2.7791860103607178, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8917264938354492, + "num_tokens": 445479501.0, + "step": 11474 + }, + { + "epoch": 1.4597379468261036, + "grad_norm": 2.5692555904388428, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9083601236343384, + "num_tokens": 445516535.0, + "step": 11475 + }, + { + "epoch": 1.4598651571046941, + "grad_norm": 2.5994691848754883, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8934077024459839, + "num_tokens": 445554677.0, + "step": 11476 + }, + { + "epoch": 1.4599923673832846, + "grad_norm": 2.82423996925354, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8910662531852722, + "num_tokens": 445587083.0, + "step": 11477 + }, + { + "epoch": 1.460119577661875, + "grad_norm": 2.9907796382904053, + "learning_rate": 1e-06, + "loss": 0.3226, + "mean_token_accuracy": 0.8880612850189209, + "num_tokens": 445620480.0, + "step": 11478 + }, + { + "epoch": 1.4602467879404655, + "grad_norm": 2.5239787101745605, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8944576978683472, + "num_tokens": 445659748.0, + "step": 11479 + }, + { + "epoch": 1.460373998219056, + "grad_norm": 2.8420002460479736, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8899376392364502, + "num_tokens": 445694521.0, + "step": 11480 + }, + { + "epoch": 1.4605012084976465, + "grad_norm": 2.57112193107605, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8867285251617432, + "num_tokens": 445734025.0, + "step": 11481 + }, + { + "epoch": 1.460628418776237, + "grad_norm": 2.6543490886688232, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8946374654769897, + "num_tokens": 445771525.0, + "step": 11482 + }, + { + "epoch": 1.4607556290548276, + "grad_norm": 2.570368528366089, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.8956471681594849, + "num_tokens": 445810666.0, + "step": 11483 + }, + { + "epoch": 1.4608828393334181, + "grad_norm": 2.494797706604004, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9028759002685547, + "num_tokens": 445850287.0, + "step": 11484 + }, + { + "epoch": 1.4610100496120086, + "grad_norm": 2.582412004470825, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8841032981872559, + "num_tokens": 445888129.0, + "step": 11485 + }, + { + "epoch": 1.4611372598905992, + "grad_norm": 2.625030755996704, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8978341221809387, + "num_tokens": 445925801.0, + "step": 11486 + }, + { + "epoch": 1.4612644701691897, + "grad_norm": 2.8107528686523438, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8949295878410339, + "num_tokens": 445962082.0, + "step": 11487 + }, + { + "epoch": 1.4613916804477802, + "grad_norm": 2.565176010131836, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9032996892929077, + "num_tokens": 445996901.0, + "step": 11488 + }, + { + "epoch": 1.4615188907263708, + "grad_norm": 2.8329341411590576, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8785820007324219, + "num_tokens": 446035714.0, + "step": 11489 + }, + { + "epoch": 1.4616461010049613, + "grad_norm": 2.468033790588379, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.8948748707771301, + "num_tokens": 446078385.0, + "step": 11490 + }, + { + "epoch": 1.4617733112835518, + "grad_norm": 2.4045462608337402, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9003069400787354, + "num_tokens": 446118682.0, + "step": 11491 + }, + { + "epoch": 1.4619005215621423, + "grad_norm": 2.6439931392669678, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8937790393829346, + "num_tokens": 446155462.0, + "step": 11492 + }, + { + "epoch": 1.4620277318407329, + "grad_norm": 2.5489180088043213, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8888049721717834, + "num_tokens": 446197252.0, + "step": 11493 + }, + { + "epoch": 1.4621549421193232, + "grad_norm": 2.6826539039611816, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8832741379737854, + "num_tokens": 446238742.0, + "step": 11494 + }, + { + "epoch": 1.4622821523979137, + "grad_norm": 2.6744418144226074, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.888093113899231, + "num_tokens": 446277725.0, + "step": 11495 + }, + { + "epoch": 1.4624093626765042, + "grad_norm": 2.832094192504883, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8908196687698364, + "num_tokens": 446314175.0, + "step": 11496 + }, + { + "epoch": 1.4625365729550948, + "grad_norm": 2.6514313220977783, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8927159309387207, + "num_tokens": 446353501.0, + "step": 11497 + }, + { + "epoch": 1.4626637832336853, + "grad_norm": 2.5852537155151367, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8970251083374023, + "num_tokens": 446390453.0, + "step": 11498 + }, + { + "epoch": 1.4627909935122758, + "grad_norm": 2.587357521057129, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8903828859329224, + "num_tokens": 446433609.0, + "step": 11499 + }, + { + "epoch": 1.4629182037908663, + "grad_norm": 2.702291965484619, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8892883062362671, + "num_tokens": 446472568.0, + "step": 11500 + }, + { + "epoch": 1.4630454140694569, + "grad_norm": 2.9449000358581543, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8853166699409485, + "num_tokens": 446507887.0, + "step": 11501 + }, + { + "epoch": 1.4631726243480474, + "grad_norm": 2.8014183044433594, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8842913508415222, + "num_tokens": 446543280.0, + "step": 11502 + }, + { + "epoch": 1.4632998346266377, + "grad_norm": 2.4172351360321045, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9055005311965942, + "num_tokens": 446580655.0, + "step": 11503 + }, + { + "epoch": 1.4634270449052282, + "grad_norm": 2.5634043216705322, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8861929178237915, + "num_tokens": 446619962.0, + "step": 11504 + }, + { + "epoch": 1.4635542551838188, + "grad_norm": 2.7910614013671875, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8879265785217285, + "num_tokens": 446654729.0, + "step": 11505 + }, + { + "epoch": 1.4636814654624093, + "grad_norm": 2.755734920501709, + "learning_rate": 1e-06, + "loss": 0.3397, + "mean_token_accuracy": 0.8807333707809448, + "num_tokens": 446693977.0, + "step": 11506 + }, + { + "epoch": 1.4638086757409998, + "grad_norm": 2.702025890350342, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8858879804611206, + "num_tokens": 446736575.0, + "step": 11507 + }, + { + "epoch": 1.4639358860195903, + "grad_norm": 2.6417200565338135, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.905229926109314, + "num_tokens": 446772709.0, + "step": 11508 + }, + { + "epoch": 1.4640630962981809, + "grad_norm": 2.914577007293701, + "learning_rate": 1e-06, + "loss": 0.3618, + "mean_token_accuracy": 0.8756215572357178, + "num_tokens": 446812053.0, + "step": 11509 + }, + { + "epoch": 1.4641903065767714, + "grad_norm": 2.4604504108428955, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.905442476272583, + "num_tokens": 446848827.0, + "step": 11510 + }, + { + "epoch": 1.464317516855362, + "grad_norm": 2.5155868530273438, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8964458703994751, + "num_tokens": 446893843.0, + "step": 11511 + }, + { + "epoch": 1.4644447271339525, + "grad_norm": 2.509601354598999, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8996258974075317, + "num_tokens": 446936792.0, + "step": 11512 + }, + { + "epoch": 1.464571937412543, + "grad_norm": 2.7937417030334473, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8983535766601562, + "num_tokens": 446972823.0, + "step": 11513 + }, + { + "epoch": 1.4646991476911335, + "grad_norm": 2.695775270462036, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8881181478500366, + "num_tokens": 447011353.0, + "step": 11514 + }, + { + "epoch": 1.464826357969724, + "grad_norm": 2.792200803756714, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8857460021972656, + "num_tokens": 447050447.0, + "step": 11515 + }, + { + "epoch": 1.4649535682483146, + "grad_norm": 2.6040756702423096, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8913315534591675, + "num_tokens": 447089625.0, + "step": 11516 + }, + { + "epoch": 1.465080778526905, + "grad_norm": 2.7387208938598633, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8935930132865906, + "num_tokens": 447123777.0, + "step": 11517 + }, + { + "epoch": 1.4652079888054954, + "grad_norm": 2.6261255741119385, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8927022218704224, + "num_tokens": 447164761.0, + "step": 11518 + }, + { + "epoch": 1.465335199084086, + "grad_norm": 2.931011438369751, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8860306739807129, + "num_tokens": 447196712.0, + "step": 11519 + }, + { + "epoch": 1.4654624093626765, + "grad_norm": 2.8368477821350098, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8810369968414307, + "num_tokens": 447237482.0, + "step": 11520 + }, + { + "epoch": 1.465589619641267, + "grad_norm": 2.420236349105835, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.8977741599082947, + "num_tokens": 447279409.0, + "step": 11521 + }, + { + "epoch": 1.4657168299198575, + "grad_norm": 2.7397420406341553, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8795989155769348, + "num_tokens": 447318794.0, + "step": 11522 + }, + { + "epoch": 1.465844040198448, + "grad_norm": 3.0148866176605225, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8764955997467041, + "num_tokens": 447352647.0, + "step": 11523 + }, + { + "epoch": 1.4659712504770386, + "grad_norm": 2.6283023357391357, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.886185884475708, + "num_tokens": 447392936.0, + "step": 11524 + }, + { + "epoch": 1.466098460755629, + "grad_norm": 2.4050416946411133, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9035022854804993, + "num_tokens": 447433080.0, + "step": 11525 + }, + { + "epoch": 1.4662256710342196, + "grad_norm": 2.6923882961273193, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8896483182907104, + "num_tokens": 447472742.0, + "step": 11526 + }, + { + "epoch": 1.46635288131281, + "grad_norm": 2.7251174449920654, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8924809694290161, + "num_tokens": 447508598.0, + "step": 11527 + }, + { + "epoch": 1.4664800915914005, + "grad_norm": 2.685816526412964, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8810966610908508, + "num_tokens": 447549255.0, + "step": 11528 + }, + { + "epoch": 1.466607301869991, + "grad_norm": 2.4968807697296143, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8985718488693237, + "num_tokens": 447588402.0, + "step": 11529 + }, + { + "epoch": 1.4667345121485815, + "grad_norm": 2.5639522075653076, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8908456563949585, + "num_tokens": 447629923.0, + "step": 11530 + }, + { + "epoch": 1.466861722427172, + "grad_norm": 2.671631097793579, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8845721483230591, + "num_tokens": 447670235.0, + "step": 11531 + }, + { + "epoch": 1.4669889327057626, + "grad_norm": 2.5155739784240723, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9015328884124756, + "num_tokens": 447711123.0, + "step": 11532 + }, + { + "epoch": 1.467116142984353, + "grad_norm": 2.684100389480591, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.8978610038757324, + "num_tokens": 447745421.0, + "step": 11533 + }, + { + "epoch": 1.4672433532629436, + "grad_norm": 2.8754982948303223, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8866632580757141, + "num_tokens": 447781405.0, + "step": 11534 + }, + { + "epoch": 1.4673705635415342, + "grad_norm": 2.3860368728637695, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8918324708938599, + "num_tokens": 447825981.0, + "step": 11535 + }, + { + "epoch": 1.4674977738201247, + "grad_norm": 2.919914960861206, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8910766243934631, + "num_tokens": 447859662.0, + "step": 11536 + }, + { + "epoch": 1.4676249840987152, + "grad_norm": 2.8182661533355713, + "learning_rate": 1e-06, + "loss": 0.3437, + "mean_token_accuracy": 0.8752535581588745, + "num_tokens": 447897783.0, + "step": 11537 + }, + { + "epoch": 1.4677521943773058, + "grad_norm": 2.288346290588379, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8879128694534302, + "num_tokens": 447946160.0, + "step": 11538 + }, + { + "epoch": 1.4678794046558963, + "grad_norm": 3.0015130043029785, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8970638513565063, + "num_tokens": 447977067.0, + "step": 11539 + }, + { + "epoch": 1.4680066149344868, + "grad_norm": 2.5934972763061523, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.889473021030426, + "num_tokens": 448018637.0, + "step": 11540 + }, + { + "epoch": 1.4681338252130773, + "grad_norm": 2.550323963165283, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8972630500793457, + "num_tokens": 448056623.0, + "step": 11541 + }, + { + "epoch": 1.4682610354916679, + "grad_norm": 2.4369938373565674, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9092246890068054, + "num_tokens": 448093318.0, + "step": 11542 + }, + { + "epoch": 1.4683882457702582, + "grad_norm": 2.5194623470306396, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8934603929519653, + "num_tokens": 448135606.0, + "step": 11543 + }, + { + "epoch": 1.4685154560488487, + "grad_norm": 2.526775598526001, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.9003992080688477, + "num_tokens": 448176409.0, + "step": 11544 + }, + { + "epoch": 1.4686426663274392, + "grad_norm": 2.5431723594665527, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8826498985290527, + "num_tokens": 448220607.0, + "step": 11545 + }, + { + "epoch": 1.4687698766060298, + "grad_norm": 2.893740653991699, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8836172819137573, + "num_tokens": 448258081.0, + "step": 11546 + }, + { + "epoch": 1.4688970868846203, + "grad_norm": 2.5914440155029297, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8779296875, + "num_tokens": 448303208.0, + "step": 11547 + }, + { + "epoch": 1.4690242971632108, + "grad_norm": 2.5223500728607178, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9009964466094971, + "num_tokens": 448341655.0, + "step": 11548 + }, + { + "epoch": 1.4691515074418013, + "grad_norm": 2.61625599861145, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.895159900188446, + "num_tokens": 448383914.0, + "step": 11549 + }, + { + "epoch": 1.4692787177203919, + "grad_norm": 2.7366747856140137, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.89701247215271, + "num_tokens": 448418962.0, + "step": 11550 + }, + { + "epoch": 1.4694059279989824, + "grad_norm": 2.699831008911133, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8920477628707886, + "num_tokens": 448456310.0, + "step": 11551 + }, + { + "epoch": 1.4695331382775727, + "grad_norm": 2.6899940967559814, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.897743821144104, + "num_tokens": 448493965.0, + "step": 11552 + }, + { + "epoch": 1.4696603485561632, + "grad_norm": 2.600822925567627, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8875071406364441, + "num_tokens": 448537014.0, + "step": 11553 + }, + { + "epoch": 1.4697875588347538, + "grad_norm": 2.479419231414795, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8865905404090881, + "num_tokens": 448582919.0, + "step": 11554 + }, + { + "epoch": 1.4699147691133443, + "grad_norm": 2.7992165088653564, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8927007913589478, + "num_tokens": 448618060.0, + "step": 11555 + }, + { + "epoch": 1.4700419793919348, + "grad_norm": 3.2671780586242676, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8856464624404907, + "num_tokens": 448648256.0, + "step": 11556 + }, + { + "epoch": 1.4701691896705253, + "grad_norm": 2.852152109146118, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8859075903892517, + "num_tokens": 448680193.0, + "step": 11557 + }, + { + "epoch": 1.4702963999491159, + "grad_norm": 2.505160093307495, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8875959515571594, + "num_tokens": 448720773.0, + "step": 11558 + }, + { + "epoch": 1.4704236102277064, + "grad_norm": 2.5082828998565674, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8911793231964111, + "num_tokens": 448760381.0, + "step": 11559 + }, + { + "epoch": 1.470550820506297, + "grad_norm": 3.0037851333618164, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8786543011665344, + "num_tokens": 448798086.0, + "step": 11560 + }, + { + "epoch": 1.4706780307848875, + "grad_norm": 2.545480251312256, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8851571083068848, + "num_tokens": 448839247.0, + "step": 11561 + }, + { + "epoch": 1.470805241063478, + "grad_norm": 2.6317853927612305, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8945542573928833, + "num_tokens": 448878396.0, + "step": 11562 + }, + { + "epoch": 1.4709324513420685, + "grad_norm": 2.6907451152801514, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8991232514381409, + "num_tokens": 448915246.0, + "step": 11563 + }, + { + "epoch": 1.471059661620659, + "grad_norm": 2.6242802143096924, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.890627384185791, + "num_tokens": 448950902.0, + "step": 11564 + }, + { + "epoch": 1.4711868718992496, + "grad_norm": 2.49712872505188, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8874098062515259, + "num_tokens": 448996080.0, + "step": 11565 + }, + { + "epoch": 1.47131408217784, + "grad_norm": 2.6504623889923096, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8911496996879578, + "num_tokens": 449036318.0, + "step": 11566 + }, + { + "epoch": 1.4714412924564304, + "grad_norm": 2.4864728450775146, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8957087993621826, + "num_tokens": 449077365.0, + "step": 11567 + }, + { + "epoch": 1.471568502735021, + "grad_norm": 2.745692253112793, + "learning_rate": 1e-06, + "loss": 0.3329, + "mean_token_accuracy": 0.8853513598442078, + "num_tokens": 449118939.0, + "step": 11568 + }, + { + "epoch": 1.4716957130136115, + "grad_norm": 2.6576929092407227, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8958030939102173, + "num_tokens": 449161432.0, + "step": 11569 + }, + { + "epoch": 1.471822923292202, + "grad_norm": 2.5640041828155518, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8939417600631714, + "num_tokens": 449201063.0, + "step": 11570 + }, + { + "epoch": 1.4719501335707925, + "grad_norm": 2.82119083404541, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8829177021980286, + "num_tokens": 449239192.0, + "step": 11571 + }, + { + "epoch": 1.472077343849383, + "grad_norm": 2.391369581222534, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8920606970787048, + "num_tokens": 449284131.0, + "step": 11572 + }, + { + "epoch": 1.4722045541279736, + "grad_norm": 2.552734375, + "learning_rate": 1e-06, + "loss": 0.3539, + "mean_token_accuracy": 0.8729618191719055, + "num_tokens": 449329485.0, + "step": 11573 + }, + { + "epoch": 1.472331764406564, + "grad_norm": 2.5453569889068604, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8863762617111206, + "num_tokens": 449371669.0, + "step": 11574 + }, + { + "epoch": 1.4724589746851546, + "grad_norm": 2.728877544403076, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8915939927101135, + "num_tokens": 449408354.0, + "step": 11575 + }, + { + "epoch": 1.472586184963745, + "grad_norm": 2.536607265472412, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8826783299446106, + "num_tokens": 449449787.0, + "step": 11576 + }, + { + "epoch": 1.4727133952423355, + "grad_norm": 2.6187057495117188, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8829296827316284, + "num_tokens": 449491606.0, + "step": 11577 + }, + { + "epoch": 1.472840605520926, + "grad_norm": 2.5847671031951904, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8822897672653198, + "num_tokens": 449534880.0, + "step": 11578 + }, + { + "epoch": 1.4729678157995165, + "grad_norm": 2.3583734035491943, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9043295383453369, + "num_tokens": 449577631.0, + "step": 11579 + }, + { + "epoch": 1.473095026078107, + "grad_norm": 2.6285617351531982, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8876664638519287, + "num_tokens": 449620490.0, + "step": 11580 + }, + { + "epoch": 1.4732222363566976, + "grad_norm": 2.566713571548462, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9022119045257568, + "num_tokens": 449659019.0, + "step": 11581 + }, + { + "epoch": 1.473349446635288, + "grad_norm": 2.8398475646972656, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8936267495155334, + "num_tokens": 449690594.0, + "step": 11582 + }, + { + "epoch": 1.4734766569138786, + "grad_norm": 2.5784695148468018, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8891531229019165, + "num_tokens": 449730310.0, + "step": 11583 + }, + { + "epoch": 1.4736038671924692, + "grad_norm": 2.4562368392944336, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9043360948562622, + "num_tokens": 449772062.0, + "step": 11584 + }, + { + "epoch": 1.4737310774710597, + "grad_norm": 2.6086201667785645, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8909668922424316, + "num_tokens": 449816231.0, + "step": 11585 + }, + { + "epoch": 1.4738582877496502, + "grad_norm": 2.535165786743164, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8964182734489441, + "num_tokens": 449854406.0, + "step": 11586 + }, + { + "epoch": 1.4739854980282407, + "grad_norm": 3.085906982421875, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8993184566497803, + "num_tokens": 449884740.0, + "step": 11587 + }, + { + "epoch": 1.4741127083068313, + "grad_norm": 2.522866725921631, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8869908452033997, + "num_tokens": 449926171.0, + "step": 11588 + }, + { + "epoch": 1.4742399185854218, + "grad_norm": 2.79252028465271, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8801416158676147, + "num_tokens": 449966454.0, + "step": 11589 + }, + { + "epoch": 1.4743671288640123, + "grad_norm": 2.498758554458618, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8984283208847046, + "num_tokens": 450006256.0, + "step": 11590 + }, + { + "epoch": 1.4744943391426026, + "grad_norm": 2.805172920227051, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8888726830482483, + "num_tokens": 450038877.0, + "step": 11591 + }, + { + "epoch": 1.4746215494211932, + "grad_norm": 2.963414430618286, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8864449262619019, + "num_tokens": 450073477.0, + "step": 11592 + }, + { + "epoch": 1.4747487596997837, + "grad_norm": 2.3772976398468018, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.902276337146759, + "num_tokens": 450112209.0, + "step": 11593 + }, + { + "epoch": 1.4748759699783742, + "grad_norm": 2.7107367515563965, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8953585624694824, + "num_tokens": 450149412.0, + "step": 11594 + }, + { + "epoch": 1.4750031802569648, + "grad_norm": 2.51314115524292, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.894508957862854, + "num_tokens": 450189867.0, + "step": 11595 + }, + { + "epoch": 1.4751303905355553, + "grad_norm": 2.5730082988739014, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.9021578431129456, + "num_tokens": 450225920.0, + "step": 11596 + }, + { + "epoch": 1.4752576008141458, + "grad_norm": 2.599439859390259, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8889390230178833, + "num_tokens": 450267255.0, + "step": 11597 + }, + { + "epoch": 1.4753848110927363, + "grad_norm": 2.65122389793396, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8920174241065979, + "num_tokens": 450306213.0, + "step": 11598 + }, + { + "epoch": 1.4755120213713269, + "grad_norm": 2.568044662475586, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8864559531211853, + "num_tokens": 450346314.0, + "step": 11599 + }, + { + "epoch": 1.4756392316499174, + "grad_norm": 2.659882068634033, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8823028802871704, + "num_tokens": 450388719.0, + "step": 11600 + }, + { + "epoch": 1.4757664419285077, + "grad_norm": 2.621169328689575, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8901867866516113, + "num_tokens": 450428836.0, + "step": 11601 + }, + { + "epoch": 1.4758936522070982, + "grad_norm": 2.5446372032165527, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8966476917266846, + "num_tokens": 450471926.0, + "step": 11602 + }, + { + "epoch": 1.4760208624856888, + "grad_norm": 2.518800735473633, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8916621208190918, + "num_tokens": 450509538.0, + "step": 11603 + }, + { + "epoch": 1.4761480727642793, + "grad_norm": 2.4845080375671387, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8991662263870239, + "num_tokens": 450552790.0, + "step": 11604 + }, + { + "epoch": 1.4762752830428698, + "grad_norm": 2.5046229362487793, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8953208923339844, + "num_tokens": 450594994.0, + "step": 11605 + }, + { + "epoch": 1.4764024933214603, + "grad_norm": 2.490964889526367, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8881367444992065, + "num_tokens": 450637781.0, + "step": 11606 + }, + { + "epoch": 1.4765297036000509, + "grad_norm": 2.5223605632781982, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8955225944519043, + "num_tokens": 450677649.0, + "step": 11607 + }, + { + "epoch": 1.4766569138786414, + "grad_norm": 2.5051803588867188, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8925580382347107, + "num_tokens": 450720039.0, + "step": 11608 + }, + { + "epoch": 1.476784124157232, + "grad_norm": 2.6747498512268066, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8895699381828308, + "num_tokens": 450757619.0, + "step": 11609 + }, + { + "epoch": 1.4769113344358225, + "grad_norm": 2.851269245147705, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8952858448028564, + "num_tokens": 450790680.0, + "step": 11610 + }, + { + "epoch": 1.477038544714413, + "grad_norm": 2.699904441833496, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8928917050361633, + "num_tokens": 450828116.0, + "step": 11611 + }, + { + "epoch": 1.4771657549930035, + "grad_norm": 2.6347668170928955, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8927882313728333, + "num_tokens": 450865746.0, + "step": 11612 + }, + { + "epoch": 1.477292965271594, + "grad_norm": 2.5934624671936035, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8769538998603821, + "num_tokens": 450909899.0, + "step": 11613 + }, + { + "epoch": 1.4774201755501846, + "grad_norm": 2.487696409225464, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9041787385940552, + "num_tokens": 450948539.0, + "step": 11614 + }, + { + "epoch": 1.477547385828775, + "grad_norm": 2.9089608192443848, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8921523094177246, + "num_tokens": 450981360.0, + "step": 11615 + }, + { + "epoch": 1.4776745961073654, + "grad_norm": 2.289095401763916, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.892646074295044, + "num_tokens": 451025407.0, + "step": 11616 + }, + { + "epoch": 1.477801806385956, + "grad_norm": 2.6063921451568604, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8843428492546082, + "num_tokens": 451070387.0, + "step": 11617 + }, + { + "epoch": 1.4779290166645465, + "grad_norm": 3.0069167613983154, + "learning_rate": 1e-06, + "loss": 0.3674, + "mean_token_accuracy": 0.8747729063034058, + "num_tokens": 451105973.0, + "step": 11618 + }, + { + "epoch": 1.478056226943137, + "grad_norm": 2.908667802810669, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8920494914054871, + "num_tokens": 451139143.0, + "step": 11619 + }, + { + "epoch": 1.4781834372217275, + "grad_norm": 2.847341537475586, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8914409279823303, + "num_tokens": 451170884.0, + "step": 11620 + }, + { + "epoch": 1.478310647500318, + "grad_norm": 2.49534273147583, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8913364410400391, + "num_tokens": 451213682.0, + "step": 11621 + }, + { + "epoch": 1.4784378577789086, + "grad_norm": 2.5500869750976562, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.902296245098114, + "num_tokens": 451251643.0, + "step": 11622 + }, + { + "epoch": 1.478565068057499, + "grad_norm": 2.516641139984131, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8998396396636963, + "num_tokens": 451293955.0, + "step": 11623 + }, + { + "epoch": 1.4786922783360896, + "grad_norm": 2.4350364208221436, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8946294188499451, + "num_tokens": 451335274.0, + "step": 11624 + }, + { + "epoch": 1.47881948861468, + "grad_norm": 2.6337966918945312, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.882125735282898, + "num_tokens": 451378097.0, + "step": 11625 + }, + { + "epoch": 1.4789466988932705, + "grad_norm": 2.6102426052093506, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8839998245239258, + "num_tokens": 451419908.0, + "step": 11626 + }, + { + "epoch": 1.479073909171861, + "grad_norm": 2.542308807373047, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8872778415679932, + "num_tokens": 451459920.0, + "step": 11627 + }, + { + "epoch": 1.4792011194504515, + "grad_norm": 2.899454116821289, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8853292465209961, + "num_tokens": 451494391.0, + "step": 11628 + }, + { + "epoch": 1.479328329729042, + "grad_norm": 2.7026138305664062, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.89743971824646, + "num_tokens": 451529165.0, + "step": 11629 + }, + { + "epoch": 1.4794555400076326, + "grad_norm": 2.8326103687286377, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8946402072906494, + "num_tokens": 451562885.0, + "step": 11630 + }, + { + "epoch": 1.479582750286223, + "grad_norm": 2.589552402496338, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8999909162521362, + "num_tokens": 451600570.0, + "step": 11631 + }, + { + "epoch": 1.4797099605648136, + "grad_norm": 2.813434362411499, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.884056806564331, + "num_tokens": 451637291.0, + "step": 11632 + }, + { + "epoch": 1.4798371708434042, + "grad_norm": 2.6065590381622314, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8970386981964111, + "num_tokens": 451674887.0, + "step": 11633 + }, + { + "epoch": 1.4799643811219947, + "grad_norm": 2.4873242378234863, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.889103889465332, + "num_tokens": 451718804.0, + "step": 11634 + }, + { + "epoch": 1.4800915914005852, + "grad_norm": 2.3653647899627686, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.899688720703125, + "num_tokens": 451761670.0, + "step": 11635 + }, + { + "epoch": 1.4802188016791757, + "grad_norm": 2.9279873371124268, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8781167268753052, + "num_tokens": 451797895.0, + "step": 11636 + }, + { + "epoch": 1.4803460119577663, + "grad_norm": 2.5253283977508545, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9030143022537231, + "num_tokens": 451836015.0, + "step": 11637 + }, + { + "epoch": 1.4804732222363568, + "grad_norm": 2.557541847229004, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9006519913673401, + "num_tokens": 451872117.0, + "step": 11638 + }, + { + "epoch": 1.4806004325149473, + "grad_norm": 2.5677123069763184, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8866332769393921, + "num_tokens": 451912136.0, + "step": 11639 + }, + { + "epoch": 1.4807276427935376, + "grad_norm": 2.659897565841675, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8957468271255493, + "num_tokens": 451947752.0, + "step": 11640 + }, + { + "epoch": 1.4808548530721282, + "grad_norm": 2.6340227127075195, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.893073320388794, + "num_tokens": 451984923.0, + "step": 11641 + }, + { + "epoch": 1.4809820633507187, + "grad_norm": 2.641373634338379, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8861459493637085, + "num_tokens": 452024387.0, + "step": 11642 + }, + { + "epoch": 1.4811092736293092, + "grad_norm": 2.596245527267456, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8878127336502075, + "num_tokens": 452063010.0, + "step": 11643 + }, + { + "epoch": 1.4812364839078997, + "grad_norm": 2.7179958820343018, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8944973945617676, + "num_tokens": 452098129.0, + "step": 11644 + }, + { + "epoch": 1.4813636941864903, + "grad_norm": 2.5064914226531982, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9092628359794617, + "num_tokens": 452133505.0, + "step": 11645 + }, + { + "epoch": 1.4814909044650808, + "grad_norm": 2.7374744415283203, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8769029974937439, + "num_tokens": 452173593.0, + "step": 11646 + }, + { + "epoch": 1.4816181147436713, + "grad_norm": 2.532072067260742, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8915501832962036, + "num_tokens": 452211279.0, + "step": 11647 + }, + { + "epoch": 1.4817453250222619, + "grad_norm": 3.1593704223632812, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8735918402671814, + "num_tokens": 452241469.0, + "step": 11648 + }, + { + "epoch": 1.4818725353008524, + "grad_norm": 2.818173885345459, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9055324792861938, + "num_tokens": 452274012.0, + "step": 11649 + }, + { + "epoch": 1.4819997455794427, + "grad_norm": 2.6633729934692383, + "learning_rate": 1e-06, + "loss": 0.3534, + "mean_token_accuracy": 0.873640775680542, + "num_tokens": 452320441.0, + "step": 11650 + }, + { + "epoch": 1.4821269558580332, + "grad_norm": 2.482668161392212, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8836169242858887, + "num_tokens": 452366120.0, + "step": 11651 + }, + { + "epoch": 1.4822541661366238, + "grad_norm": 2.69707989692688, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8932226896286011, + "num_tokens": 452402079.0, + "step": 11652 + }, + { + "epoch": 1.4823813764152143, + "grad_norm": 2.869746446609497, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8959407806396484, + "num_tokens": 452433706.0, + "step": 11653 + }, + { + "epoch": 1.4825085866938048, + "grad_norm": 2.787973403930664, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.884684681892395, + "num_tokens": 452469760.0, + "step": 11654 + }, + { + "epoch": 1.4826357969723953, + "grad_norm": 2.8746285438537598, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8927996158599854, + "num_tokens": 452504854.0, + "step": 11655 + }, + { + "epoch": 1.4827630072509859, + "grad_norm": 2.7658655643463135, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8900876641273499, + "num_tokens": 452539582.0, + "step": 11656 + }, + { + "epoch": 1.4828902175295764, + "grad_norm": 2.542616844177246, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8929645419120789, + "num_tokens": 452581338.0, + "step": 11657 + }, + { + "epoch": 1.483017427808167, + "grad_norm": 2.5693199634552, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8945672512054443, + "num_tokens": 452622323.0, + "step": 11658 + }, + { + "epoch": 1.4831446380867574, + "grad_norm": 2.5176374912261963, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9040859341621399, + "num_tokens": 452659505.0, + "step": 11659 + }, + { + "epoch": 1.483271848365348, + "grad_norm": 2.8790886402130127, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8935636878013611, + "num_tokens": 452692568.0, + "step": 11660 + }, + { + "epoch": 1.4833990586439385, + "grad_norm": 2.7279698848724365, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8848966360092163, + "num_tokens": 452731098.0, + "step": 11661 + }, + { + "epoch": 1.483526268922529, + "grad_norm": 2.6761863231658936, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8847165703773499, + "num_tokens": 452770648.0, + "step": 11662 + }, + { + "epoch": 1.4836534792011196, + "grad_norm": 2.8310728073120117, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9004019498825073, + "num_tokens": 452802909.0, + "step": 11663 + }, + { + "epoch": 1.48378068947971, + "grad_norm": 2.823521137237549, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8860282897949219, + "num_tokens": 452840270.0, + "step": 11664 + }, + { + "epoch": 1.4839078997583004, + "grad_norm": 2.597085952758789, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8929829597473145, + "num_tokens": 452880785.0, + "step": 11665 + }, + { + "epoch": 1.484035110036891, + "grad_norm": 2.7898385524749756, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8696945905685425, + "num_tokens": 452922160.0, + "step": 11666 + }, + { + "epoch": 1.4841623203154815, + "grad_norm": 2.99958872795105, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8787927627563477, + "num_tokens": 452954480.0, + "step": 11667 + }, + { + "epoch": 1.484289530594072, + "grad_norm": 2.5379526615142822, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8866329789161682, + "num_tokens": 452995693.0, + "step": 11668 + }, + { + "epoch": 1.4844167408726625, + "grad_norm": 2.7829556465148926, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8884071111679077, + "num_tokens": 453029707.0, + "step": 11669 + }, + { + "epoch": 1.484543951151253, + "grad_norm": 3.0311050415039062, + "learning_rate": 1e-06, + "loss": 0.3511, + "mean_token_accuracy": 0.877008855342865, + "num_tokens": 453063379.0, + "step": 11670 + }, + { + "epoch": 1.4846711614298436, + "grad_norm": 2.4568209648132324, + "learning_rate": 1e-06, + "loss": 0.3556, + "mean_token_accuracy": 0.8730084896087646, + "num_tokens": 453110004.0, + "step": 11671 + }, + { + "epoch": 1.484798371708434, + "grad_norm": 3.087529182434082, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8767812848091125, + "num_tokens": 453142716.0, + "step": 11672 + }, + { + "epoch": 1.4849255819870246, + "grad_norm": 2.6854655742645264, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8852715492248535, + "num_tokens": 453180634.0, + "step": 11673 + }, + { + "epoch": 1.485052792265615, + "grad_norm": 2.4414637088775635, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8923123478889465, + "num_tokens": 453220311.0, + "step": 11674 + }, + { + "epoch": 1.4851800025442055, + "grad_norm": 2.3634140491485596, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8915636539459229, + "num_tokens": 453263162.0, + "step": 11675 + }, + { + "epoch": 1.485307212822796, + "grad_norm": 2.6131274700164795, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8998926877975464, + "num_tokens": 453302305.0, + "step": 11676 + }, + { + "epoch": 1.4854344231013865, + "grad_norm": 2.73225736618042, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8887240886688232, + "num_tokens": 453341063.0, + "step": 11677 + }, + { + "epoch": 1.485561633379977, + "grad_norm": 2.749677896499634, + "learning_rate": 1e-06, + "loss": 0.3402, + "mean_token_accuracy": 0.8794794082641602, + "num_tokens": 453378934.0, + "step": 11678 + }, + { + "epoch": 1.4856888436585676, + "grad_norm": 2.595053195953369, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8959155678749084, + "num_tokens": 453414586.0, + "step": 11679 + }, + { + "epoch": 1.485816053937158, + "grad_norm": 2.435673236846924, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.89409339427948, + "num_tokens": 453456309.0, + "step": 11680 + }, + { + "epoch": 1.4859432642157486, + "grad_norm": 2.506648540496826, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9088345766067505, + "num_tokens": 453494213.0, + "step": 11681 + }, + { + "epoch": 1.4860704744943392, + "grad_norm": 2.3843064308166504, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8880056738853455, + "num_tokens": 453540040.0, + "step": 11682 + }, + { + "epoch": 1.4861976847729297, + "grad_norm": 2.5650360584259033, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9018887281417847, + "num_tokens": 453579900.0, + "step": 11683 + }, + { + "epoch": 1.4863248950515202, + "grad_norm": 2.57207989692688, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8941402435302734, + "num_tokens": 453616058.0, + "step": 11684 + }, + { + "epoch": 1.4864521053301107, + "grad_norm": 2.484950304031372, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8983228206634521, + "num_tokens": 453655319.0, + "step": 11685 + }, + { + "epoch": 1.4865793156087013, + "grad_norm": 2.553313970565796, + "learning_rate": 1e-06, + "loss": 0.3391, + "mean_token_accuracy": 0.8793348670005798, + "num_tokens": 453699100.0, + "step": 11686 + }, + { + "epoch": 1.4867065258872918, + "grad_norm": 2.8512794971466064, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8745324015617371, + "num_tokens": 453732926.0, + "step": 11687 + }, + { + "epoch": 1.4868337361658823, + "grad_norm": 2.5193095207214355, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8958162665367126, + "num_tokens": 453772542.0, + "step": 11688 + }, + { + "epoch": 1.4869609464444726, + "grad_norm": 2.861081600189209, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8908365964889526, + "num_tokens": 453805955.0, + "step": 11689 + }, + { + "epoch": 1.4870881567230632, + "grad_norm": 2.816737174987793, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8762616515159607, + "num_tokens": 453841423.0, + "step": 11690 + }, + { + "epoch": 1.4872153670016537, + "grad_norm": 2.5643208026885986, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8860247135162354, + "num_tokens": 453883644.0, + "step": 11691 + }, + { + "epoch": 1.4873425772802442, + "grad_norm": 2.2862894535064697, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9146815538406372, + "num_tokens": 453925326.0, + "step": 11692 + }, + { + "epoch": 1.4874697875588347, + "grad_norm": 2.800236701965332, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8896985054016113, + "num_tokens": 453962164.0, + "step": 11693 + }, + { + "epoch": 1.4875969978374253, + "grad_norm": 2.5057132244110107, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9001283645629883, + "num_tokens": 454001264.0, + "step": 11694 + }, + { + "epoch": 1.4877242081160158, + "grad_norm": 2.5389766693115234, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8916219472885132, + "num_tokens": 454040755.0, + "step": 11695 + }, + { + "epoch": 1.4878514183946063, + "grad_norm": 2.7880711555480957, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8922479152679443, + "num_tokens": 454074853.0, + "step": 11696 + }, + { + "epoch": 1.4879786286731969, + "grad_norm": 2.451097011566162, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8917784094810486, + "num_tokens": 454118939.0, + "step": 11697 + }, + { + "epoch": 1.4881058389517874, + "grad_norm": 2.682187080383301, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9003769159317017, + "num_tokens": 454155154.0, + "step": 11698 + }, + { + "epoch": 1.4882330492303777, + "grad_norm": 2.623215913772583, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.878328800201416, + "num_tokens": 454197732.0, + "step": 11699 + }, + { + "epoch": 1.4883602595089682, + "grad_norm": 2.830688714981079, + "learning_rate": 1e-06, + "loss": 0.3551, + "mean_token_accuracy": 0.8717957139015198, + "num_tokens": 454237000.0, + "step": 11700 + }, + { + "epoch": 1.4884874697875587, + "grad_norm": 2.7069430351257324, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8860560655593872, + "num_tokens": 454274325.0, + "step": 11701 + }, + { + "epoch": 1.4886146800661493, + "grad_norm": 2.874924421310425, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8770885467529297, + "num_tokens": 454310601.0, + "step": 11702 + }, + { + "epoch": 1.4887418903447398, + "grad_norm": 2.543224811553955, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8906199932098389, + "num_tokens": 454351198.0, + "step": 11703 + }, + { + "epoch": 1.4888691006233303, + "grad_norm": 3.0120065212249756, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8863680958747864, + "num_tokens": 454383092.0, + "step": 11704 + }, + { + "epoch": 1.4889963109019209, + "grad_norm": 2.7294368743896484, + "learning_rate": 1e-06, + "loss": 0.3576, + "mean_token_accuracy": 0.8710421919822693, + "num_tokens": 454427611.0, + "step": 11705 + }, + { + "epoch": 1.4891235211805114, + "grad_norm": 2.5574910640716553, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8967424631118774, + "num_tokens": 454468388.0, + "step": 11706 + }, + { + "epoch": 1.489250731459102, + "grad_norm": 2.531174898147583, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9023945331573486, + "num_tokens": 454505664.0, + "step": 11707 + }, + { + "epoch": 1.4893779417376924, + "grad_norm": 2.5528509616851807, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9035040140151978, + "num_tokens": 454540471.0, + "step": 11708 + }, + { + "epoch": 1.489505152016283, + "grad_norm": 2.645538091659546, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8939254879951477, + "num_tokens": 454580402.0, + "step": 11709 + }, + { + "epoch": 1.4896323622948735, + "grad_norm": 2.619255304336548, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8941651582717896, + "num_tokens": 454615626.0, + "step": 11710 + }, + { + "epoch": 1.489759572573464, + "grad_norm": 2.573168992996216, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8974018692970276, + "num_tokens": 454654864.0, + "step": 11711 + }, + { + "epoch": 1.4898867828520546, + "grad_norm": 2.5559747219085693, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8880224227905273, + "num_tokens": 454697477.0, + "step": 11712 + }, + { + "epoch": 1.490013993130645, + "grad_norm": 2.7119622230529785, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9053347110748291, + "num_tokens": 454730305.0, + "step": 11713 + }, + { + "epoch": 1.4901412034092354, + "grad_norm": 2.436187744140625, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8902959227561951, + "num_tokens": 454774374.0, + "step": 11714 + }, + { + "epoch": 1.490268413687826, + "grad_norm": 2.7623019218444824, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8878322839736938, + "num_tokens": 454808542.0, + "step": 11715 + }, + { + "epoch": 1.4903956239664164, + "grad_norm": 2.5518250465393066, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8823118209838867, + "num_tokens": 454853334.0, + "step": 11716 + }, + { + "epoch": 1.490522834245007, + "grad_norm": 2.5717732906341553, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8820589780807495, + "num_tokens": 454895348.0, + "step": 11717 + }, + { + "epoch": 1.4906500445235975, + "grad_norm": 2.7344565391540527, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9002910852432251, + "num_tokens": 454927422.0, + "step": 11718 + }, + { + "epoch": 1.490777254802188, + "grad_norm": 2.8836605548858643, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8839985728263855, + "num_tokens": 454963436.0, + "step": 11719 + }, + { + "epoch": 1.4909044650807786, + "grad_norm": 2.6352968215942383, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8847521543502808, + "num_tokens": 455004389.0, + "step": 11720 + }, + { + "epoch": 1.491031675359369, + "grad_norm": 2.805626630783081, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8855670690536499, + "num_tokens": 455039669.0, + "step": 11721 + }, + { + "epoch": 1.4911588856379596, + "grad_norm": 2.8521435260772705, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8930670022964478, + "num_tokens": 455075517.0, + "step": 11722 + }, + { + "epoch": 1.49128609591655, + "grad_norm": 2.462528944015503, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8905307054519653, + "num_tokens": 455117153.0, + "step": 11723 + }, + { + "epoch": 1.4914133061951405, + "grad_norm": 2.625063419342041, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8952999114990234, + "num_tokens": 455153196.0, + "step": 11724 + }, + { + "epoch": 1.491540516473731, + "grad_norm": 2.447934865951538, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.899867594242096, + "num_tokens": 455192775.0, + "step": 11725 + }, + { + "epoch": 1.4916677267523215, + "grad_norm": 2.8417396545410156, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9011470675468445, + "num_tokens": 455225641.0, + "step": 11726 + }, + { + "epoch": 1.491794937030912, + "grad_norm": 2.58258318901062, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8861007690429688, + "num_tokens": 455267523.0, + "step": 11727 + }, + { + "epoch": 1.4919221473095026, + "grad_norm": 2.4799396991729736, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8888628482818604, + "num_tokens": 455311039.0, + "step": 11728 + }, + { + "epoch": 1.492049357588093, + "grad_norm": 2.5811173915863037, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9005172848701477, + "num_tokens": 455345400.0, + "step": 11729 + }, + { + "epoch": 1.4921765678666836, + "grad_norm": 3.050107717514038, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8805388808250427, + "num_tokens": 455377804.0, + "step": 11730 + }, + { + "epoch": 1.4923037781452742, + "grad_norm": 2.3973236083984375, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.899095892906189, + "num_tokens": 455421959.0, + "step": 11731 + }, + { + "epoch": 1.4924309884238647, + "grad_norm": 2.8630151748657227, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.900947093963623, + "num_tokens": 455452429.0, + "step": 11732 + }, + { + "epoch": 1.4925581987024552, + "grad_norm": 2.6267237663269043, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8875047564506531, + "num_tokens": 455492698.0, + "step": 11733 + }, + { + "epoch": 1.4926854089810457, + "grad_norm": 2.439840078353882, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9015541076660156, + "num_tokens": 455531037.0, + "step": 11734 + }, + { + "epoch": 1.4928126192596363, + "grad_norm": 2.87363600730896, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8805021643638611, + "num_tokens": 455569017.0, + "step": 11735 + }, + { + "epoch": 1.4929398295382268, + "grad_norm": 2.6463019847869873, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8869665861129761, + "num_tokens": 455608979.0, + "step": 11736 + }, + { + "epoch": 1.4930670398168173, + "grad_norm": 2.326815128326416, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8999673128128052, + "num_tokens": 455653559.0, + "step": 11737 + }, + { + "epoch": 1.4931942500954076, + "grad_norm": 2.603860855102539, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8945261240005493, + "num_tokens": 455689362.0, + "step": 11738 + }, + { + "epoch": 1.4933214603739982, + "grad_norm": 2.454298496246338, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.896086573600769, + "num_tokens": 455730482.0, + "step": 11739 + }, + { + "epoch": 1.4934486706525887, + "grad_norm": 2.7382559776306152, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8931350708007812, + "num_tokens": 455767491.0, + "step": 11740 + }, + { + "epoch": 1.4935758809311792, + "grad_norm": 2.4282171726226807, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9065027236938477, + "num_tokens": 455806413.0, + "step": 11741 + }, + { + "epoch": 1.4937030912097697, + "grad_norm": 2.562551259994507, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8903288841247559, + "num_tokens": 455849125.0, + "step": 11742 + }, + { + "epoch": 1.4938303014883603, + "grad_norm": 2.7845866680145264, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8819828629493713, + "num_tokens": 455887955.0, + "step": 11743 + }, + { + "epoch": 1.4939575117669508, + "grad_norm": 2.7572498321533203, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8990790843963623, + "num_tokens": 455922114.0, + "step": 11744 + }, + { + "epoch": 1.4940847220455413, + "grad_norm": 2.504453182220459, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.898637056350708, + "num_tokens": 455963047.0, + "step": 11745 + }, + { + "epoch": 1.4942119323241319, + "grad_norm": 2.5188686847686768, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8843299150466919, + "num_tokens": 456009725.0, + "step": 11746 + }, + { + "epoch": 1.4943391426027224, + "grad_norm": 2.80907940864563, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.886176347732544, + "num_tokens": 456047011.0, + "step": 11747 + }, + { + "epoch": 1.4944663528813127, + "grad_norm": 2.8162925243377686, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8837295770645142, + "num_tokens": 456083762.0, + "step": 11748 + }, + { + "epoch": 1.4945935631599032, + "grad_norm": 2.704636335372925, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8761207461357117, + "num_tokens": 456121645.0, + "step": 11749 + }, + { + "epoch": 1.4947207734384937, + "grad_norm": 2.668180227279663, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8892869353294373, + "num_tokens": 456159329.0, + "step": 11750 + }, + { + "epoch": 1.4948479837170843, + "grad_norm": 2.5865864753723145, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8860767483711243, + "num_tokens": 456202634.0, + "step": 11751 + }, + { + "epoch": 1.4949751939956748, + "grad_norm": 2.7197396755218506, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.88913494348526, + "num_tokens": 456242005.0, + "step": 11752 + }, + { + "epoch": 1.4951024042742653, + "grad_norm": 2.6240265369415283, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8944470286369324, + "num_tokens": 456278535.0, + "step": 11753 + }, + { + "epoch": 1.4952296145528559, + "grad_norm": 2.8542158603668213, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8911240696907043, + "num_tokens": 456313253.0, + "step": 11754 + }, + { + "epoch": 1.4953568248314464, + "grad_norm": 2.6038026809692383, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.895641565322876, + "num_tokens": 456351626.0, + "step": 11755 + }, + { + "epoch": 1.495484035110037, + "grad_norm": 2.5056865215301514, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8893164396286011, + "num_tokens": 456394801.0, + "step": 11756 + }, + { + "epoch": 1.4956112453886274, + "grad_norm": 3.014143705368042, + "learning_rate": 1e-06, + "loss": 0.3787, + "mean_token_accuracy": 0.8634395599365234, + "num_tokens": 456429621.0, + "step": 11757 + }, + { + "epoch": 1.495738455667218, + "grad_norm": 2.708771228790283, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.89002925157547, + "num_tokens": 456466226.0, + "step": 11758 + }, + { + "epoch": 1.4958656659458085, + "grad_norm": 2.5828709602355957, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8899281024932861, + "num_tokens": 456507222.0, + "step": 11759 + }, + { + "epoch": 1.495992876224399, + "grad_norm": 2.98075795173645, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.882066011428833, + "num_tokens": 456538420.0, + "step": 11760 + }, + { + "epoch": 1.4961200865029896, + "grad_norm": 2.428166151046753, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8969582319259644, + "num_tokens": 456581043.0, + "step": 11761 + }, + { + "epoch": 1.49624729678158, + "grad_norm": 2.517667293548584, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8890650272369385, + "num_tokens": 456623328.0, + "step": 11762 + }, + { + "epoch": 1.4963745070601704, + "grad_norm": 2.513852119445801, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9009412527084351, + "num_tokens": 456662582.0, + "step": 11763 + }, + { + "epoch": 1.496501717338761, + "grad_norm": 2.5411462783813477, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9008656740188599, + "num_tokens": 456704471.0, + "step": 11764 + }, + { + "epoch": 1.4966289276173514, + "grad_norm": 2.7616171836853027, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8751921653747559, + "num_tokens": 456742881.0, + "step": 11765 + }, + { + "epoch": 1.496756137895942, + "grad_norm": 2.556511163711548, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8939063549041748, + "num_tokens": 456785140.0, + "step": 11766 + }, + { + "epoch": 1.4968833481745325, + "grad_norm": 2.55208158493042, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8930174112319946, + "num_tokens": 456827258.0, + "step": 11767 + }, + { + "epoch": 1.497010558453123, + "grad_norm": 2.847062587738037, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8739643096923828, + "num_tokens": 456866985.0, + "step": 11768 + }, + { + "epoch": 1.4971377687317136, + "grad_norm": 2.83414626121521, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.891657829284668, + "num_tokens": 456903164.0, + "step": 11769 + }, + { + "epoch": 1.497264979010304, + "grad_norm": 2.6084465980529785, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8963989019393921, + "num_tokens": 456943334.0, + "step": 11770 + }, + { + "epoch": 1.4973921892888946, + "grad_norm": 2.5471701622009277, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8881574869155884, + "num_tokens": 456984824.0, + "step": 11771 + }, + { + "epoch": 1.497519399567485, + "grad_norm": 2.8600575923919678, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8882378935813904, + "num_tokens": 457019417.0, + "step": 11772 + }, + { + "epoch": 1.4976466098460754, + "grad_norm": 2.5816457271575928, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.8999820947647095, + "num_tokens": 457055931.0, + "step": 11773 + }, + { + "epoch": 1.497773820124666, + "grad_norm": 2.648223400115967, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8913158178329468, + "num_tokens": 457096881.0, + "step": 11774 + }, + { + "epoch": 1.4979010304032565, + "grad_norm": 2.7002997398376465, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8970531821250916, + "num_tokens": 457134934.0, + "step": 11775 + }, + { + "epoch": 1.498028240681847, + "grad_norm": 2.7832283973693848, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8733269572257996, + "num_tokens": 457177086.0, + "step": 11776 + }, + { + "epoch": 1.4981554509604376, + "grad_norm": 2.4743664264678955, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8990944623947144, + "num_tokens": 457219285.0, + "step": 11777 + }, + { + "epoch": 1.498282661239028, + "grad_norm": 2.708775281906128, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8829302191734314, + "num_tokens": 457257262.0, + "step": 11778 + }, + { + "epoch": 1.4984098715176186, + "grad_norm": 2.7151708602905273, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8878318071365356, + "num_tokens": 457295500.0, + "step": 11779 + }, + { + "epoch": 1.4985370817962091, + "grad_norm": 2.472525119781494, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8978794813156128, + "num_tokens": 457335792.0, + "step": 11780 + }, + { + "epoch": 1.4986642920747997, + "grad_norm": 2.842068910598755, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8869825005531311, + "num_tokens": 457371805.0, + "step": 11781 + }, + { + "epoch": 1.4987915023533902, + "grad_norm": 2.520418167114258, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8993687629699707, + "num_tokens": 457413672.0, + "step": 11782 + }, + { + "epoch": 1.4989187126319807, + "grad_norm": 2.8264265060424805, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8835420608520508, + "num_tokens": 457448753.0, + "step": 11783 + }, + { + "epoch": 1.4990459229105713, + "grad_norm": 2.416619300842285, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8850865364074707, + "num_tokens": 457498350.0, + "step": 11784 + }, + { + "epoch": 1.4991731331891618, + "grad_norm": 2.564150810241699, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8946729898452759, + "num_tokens": 457539850.0, + "step": 11785 + }, + { + "epoch": 1.4993003434677523, + "grad_norm": 2.731081247329712, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8830534815788269, + "num_tokens": 457582338.0, + "step": 11786 + }, + { + "epoch": 1.4994275537463426, + "grad_norm": 2.653564453125, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8843687772750854, + "num_tokens": 457624196.0, + "step": 11787 + }, + { + "epoch": 1.4995547640249332, + "grad_norm": 2.5466842651367188, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8839408159255981, + "num_tokens": 457669388.0, + "step": 11788 + }, + { + "epoch": 1.4996819743035237, + "grad_norm": 2.913402557373047, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8933215141296387, + "num_tokens": 457699546.0, + "step": 11789 + }, + { + "epoch": 1.4998091845821142, + "grad_norm": 2.76654052734375, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8916869759559631, + "num_tokens": 457734787.0, + "step": 11790 + }, + { + "epoch": 1.4999363948607047, + "grad_norm": 2.6273539066314697, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8988751769065857, + "num_tokens": 457770765.0, + "step": 11791 + }, + { + "epoch": 1.5000636051392953, + "grad_norm": 2.8838138580322266, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.882111668586731, + "num_tokens": 457808475.0, + "step": 11792 + }, + { + "epoch": 1.5001908154178858, + "grad_norm": 2.6332242488861084, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8923338651657104, + "num_tokens": 457847926.0, + "step": 11793 + }, + { + "epoch": 1.5003180256964763, + "grad_norm": 2.593100070953369, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8878135681152344, + "num_tokens": 457887419.0, + "step": 11794 + }, + { + "epoch": 1.5004452359750666, + "grad_norm": 2.7095022201538086, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8975380063056946, + "num_tokens": 457922298.0, + "step": 11795 + }, + { + "epoch": 1.5005724462536572, + "grad_norm": 2.7497730255126953, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8778613805770874, + "num_tokens": 457965951.0, + "step": 11796 + }, + { + "epoch": 1.5006996565322477, + "grad_norm": 2.7915773391723633, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8796378970146179, + "num_tokens": 458004348.0, + "step": 11797 + }, + { + "epoch": 1.5008268668108382, + "grad_norm": 2.5267186164855957, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.9009513258934021, + "num_tokens": 458041859.0, + "step": 11798 + }, + { + "epoch": 1.5009540770894287, + "grad_norm": 2.655630588531494, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8816020488739014, + "num_tokens": 458085799.0, + "step": 11799 + }, + { + "epoch": 1.5010812873680193, + "grad_norm": 2.4042301177978516, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.8962879776954651, + "num_tokens": 458129205.0, + "step": 11800 + }, + { + "epoch": 1.5012084976466098, + "grad_norm": 2.514432907104492, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8907281756401062, + "num_tokens": 458170286.0, + "step": 11801 + }, + { + "epoch": 1.5013357079252003, + "grad_norm": 2.7041852474212646, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8901058435440063, + "num_tokens": 458203828.0, + "step": 11802 + }, + { + "epoch": 1.5014629182037909, + "grad_norm": 2.4256479740142822, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8953655958175659, + "num_tokens": 458244530.0, + "step": 11803 + }, + { + "epoch": 1.5015901284823814, + "grad_norm": 2.721289873123169, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8836127519607544, + "num_tokens": 458282831.0, + "step": 11804 + }, + { + "epoch": 1.501717338760972, + "grad_norm": 2.493655204772949, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8924188613891602, + "num_tokens": 458327392.0, + "step": 11805 + }, + { + "epoch": 1.5018445490395624, + "grad_norm": 2.3651862144470215, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.901801347732544, + "num_tokens": 458366964.0, + "step": 11806 + }, + { + "epoch": 1.501971759318153, + "grad_norm": 2.4559199810028076, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8976268172264099, + "num_tokens": 458406111.0, + "step": 11807 + }, + { + "epoch": 1.5020989695967435, + "grad_norm": 2.627194881439209, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.886296272277832, + "num_tokens": 458447587.0, + "step": 11808 + }, + { + "epoch": 1.502226179875334, + "grad_norm": 2.870783805847168, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8744357228279114, + "num_tokens": 458482643.0, + "step": 11809 + }, + { + "epoch": 1.5023533901539246, + "grad_norm": 2.8606467247009277, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8910713195800781, + "num_tokens": 458515290.0, + "step": 11810 + }, + { + "epoch": 1.502480600432515, + "grad_norm": 2.5845983028411865, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8964022397994995, + "num_tokens": 458556550.0, + "step": 11811 + }, + { + "epoch": 1.5026078107111056, + "grad_norm": 2.66094970703125, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8907015323638916, + "num_tokens": 458599490.0, + "step": 11812 + }, + { + "epoch": 1.502735020989696, + "grad_norm": 2.6680307388305664, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8796275854110718, + "num_tokens": 458644467.0, + "step": 11813 + }, + { + "epoch": 1.5028622312682864, + "grad_norm": 2.9456849098205566, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8814584612846375, + "num_tokens": 458677893.0, + "step": 11814 + }, + { + "epoch": 1.502989441546877, + "grad_norm": 2.5388598442077637, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8916997313499451, + "num_tokens": 458721075.0, + "step": 11815 + }, + { + "epoch": 1.5031166518254675, + "grad_norm": 2.6099462509155273, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8993184566497803, + "num_tokens": 458757146.0, + "step": 11816 + }, + { + "epoch": 1.503243862104058, + "grad_norm": 2.640286445617676, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8896671533584595, + "num_tokens": 458796001.0, + "step": 11817 + }, + { + "epoch": 1.5033710723826486, + "grad_norm": 2.5492103099823, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.894849419593811, + "num_tokens": 458832573.0, + "step": 11818 + }, + { + "epoch": 1.503498282661239, + "grad_norm": 2.722191095352173, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8921622633934021, + "num_tokens": 458867989.0, + "step": 11819 + }, + { + "epoch": 1.5036254929398294, + "grad_norm": 2.6519291400909424, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9005176424980164, + "num_tokens": 458904830.0, + "step": 11820 + }, + { + "epoch": 1.50375270321842, + "grad_norm": 2.6472110748291016, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8833686113357544, + "num_tokens": 458945832.0, + "step": 11821 + }, + { + "epoch": 1.5038799134970104, + "grad_norm": 2.7639384269714355, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8806160092353821, + "num_tokens": 458988590.0, + "step": 11822 + }, + { + "epoch": 1.504007123775601, + "grad_norm": 2.859701633453369, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8787550330162048, + "num_tokens": 459025034.0, + "step": 11823 + }, + { + "epoch": 1.5041343340541915, + "grad_norm": 2.7087948322296143, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8866649270057678, + "num_tokens": 459063581.0, + "step": 11824 + }, + { + "epoch": 1.504261544332782, + "grad_norm": 2.621032953262329, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8973579406738281, + "num_tokens": 459099167.0, + "step": 11825 + }, + { + "epoch": 1.5043887546113726, + "grad_norm": 2.5297086238861084, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8957979083061218, + "num_tokens": 459139560.0, + "step": 11826 + }, + { + "epoch": 1.504515964889963, + "grad_norm": 2.8818295001983643, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9048014879226685, + "num_tokens": 459167412.0, + "step": 11827 + }, + { + "epoch": 1.5046431751685536, + "grad_norm": 2.5230395793914795, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8953679203987122, + "num_tokens": 459205074.0, + "step": 11828 + }, + { + "epoch": 1.5047703854471441, + "grad_norm": 2.6019539833068848, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8880852460861206, + "num_tokens": 459246112.0, + "step": 11829 + }, + { + "epoch": 1.5048975957257347, + "grad_norm": 2.5326249599456787, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9043263792991638, + "num_tokens": 459281675.0, + "step": 11830 + }, + { + "epoch": 1.5050248060043252, + "grad_norm": 2.7295217514038086, + "learning_rate": 1e-06, + "loss": 0.3498, + "mean_token_accuracy": 0.8770316243171692, + "num_tokens": 459322416.0, + "step": 11831 + }, + { + "epoch": 1.5051520162829157, + "grad_norm": 2.927692413330078, + "learning_rate": 1e-06, + "loss": 0.3559, + "mean_token_accuracy": 0.8755072355270386, + "num_tokens": 459359625.0, + "step": 11832 + }, + { + "epoch": 1.5052792265615063, + "grad_norm": 2.6479854583740234, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8912963271141052, + "num_tokens": 459395708.0, + "step": 11833 + }, + { + "epoch": 1.5054064368400968, + "grad_norm": 2.8769357204437256, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8838275671005249, + "num_tokens": 459431095.0, + "step": 11834 + }, + { + "epoch": 1.5055336471186873, + "grad_norm": 2.852834939956665, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8903998136520386, + "num_tokens": 459461941.0, + "step": 11835 + }, + { + "epoch": 1.5056608573972778, + "grad_norm": 2.568326234817505, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8967572450637817, + "num_tokens": 459498757.0, + "step": 11836 + }, + { + "epoch": 1.5057880676758684, + "grad_norm": 2.728790283203125, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.899788498878479, + "num_tokens": 459533148.0, + "step": 11837 + }, + { + "epoch": 1.5059152779544587, + "grad_norm": 2.848963975906372, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9028058052062988, + "num_tokens": 459562827.0, + "step": 11838 + }, + { + "epoch": 1.5060424882330492, + "grad_norm": 2.6061620712280273, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9026836156845093, + "num_tokens": 459597971.0, + "step": 11839 + }, + { + "epoch": 1.5061696985116397, + "grad_norm": 2.497587203979492, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8920866847038269, + "num_tokens": 459636181.0, + "step": 11840 + }, + { + "epoch": 1.5062969087902303, + "grad_norm": 2.491359233856201, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8866406679153442, + "num_tokens": 459678603.0, + "step": 11841 + }, + { + "epoch": 1.5064241190688208, + "grad_norm": 2.6695010662078857, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8800573348999023, + "num_tokens": 459719836.0, + "step": 11842 + }, + { + "epoch": 1.5065513293474113, + "grad_norm": 2.532471179962158, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8822861909866333, + "num_tokens": 459764445.0, + "step": 11843 + }, + { + "epoch": 1.5066785396260016, + "grad_norm": 2.715329170227051, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8929327726364136, + "num_tokens": 459801107.0, + "step": 11844 + }, + { + "epoch": 1.5068057499045922, + "grad_norm": 2.53558611869812, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8911337852478027, + "num_tokens": 459842515.0, + "step": 11845 + }, + { + "epoch": 1.5069329601831827, + "grad_norm": 2.7065651416778564, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8964012861251831, + "num_tokens": 459877100.0, + "step": 11846 + }, + { + "epoch": 1.5070601704617732, + "grad_norm": 2.814976453781128, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8815959692001343, + "num_tokens": 459912912.0, + "step": 11847 + }, + { + "epoch": 1.5071873807403637, + "grad_norm": 2.618717670440674, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.896865725517273, + "num_tokens": 459950282.0, + "step": 11848 + }, + { + "epoch": 1.5073145910189543, + "grad_norm": 3.0235514640808105, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9006156921386719, + "num_tokens": 459978592.0, + "step": 11849 + }, + { + "epoch": 1.5074418012975448, + "grad_norm": 2.292520523071289, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.8995229005813599, + "num_tokens": 460021360.0, + "step": 11850 + }, + { + "epoch": 1.5075690115761353, + "grad_norm": 2.710268020629883, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8871653079986572, + "num_tokens": 460060686.0, + "step": 11851 + }, + { + "epoch": 1.5076962218547258, + "grad_norm": 2.840430974960327, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8779997229576111, + "num_tokens": 460100696.0, + "step": 11852 + }, + { + "epoch": 1.5078234321333164, + "grad_norm": 2.628288984298706, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9041444063186646, + "num_tokens": 460135998.0, + "step": 11853 + }, + { + "epoch": 1.507950642411907, + "grad_norm": 2.5694892406463623, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8875104784965515, + "num_tokens": 460176982.0, + "step": 11854 + }, + { + "epoch": 1.5080778526904974, + "grad_norm": 2.63360595703125, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.88083815574646, + "num_tokens": 460217299.0, + "step": 11855 + }, + { + "epoch": 1.508205062969088, + "grad_norm": 2.885408401489258, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.869361400604248, + "num_tokens": 460256684.0, + "step": 11856 + }, + { + "epoch": 1.5083322732476785, + "grad_norm": 2.5282130241394043, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.886452317237854, + "num_tokens": 460294479.0, + "step": 11857 + }, + { + "epoch": 1.508459483526269, + "grad_norm": 2.7083280086517334, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8800190091133118, + "num_tokens": 460334955.0, + "step": 11858 + }, + { + "epoch": 1.5085866938048595, + "grad_norm": 2.528498888015747, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8851838111877441, + "num_tokens": 460375959.0, + "step": 11859 + }, + { + "epoch": 1.50871390408345, + "grad_norm": 2.9005846977233887, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.88068687915802, + "num_tokens": 460409063.0, + "step": 11860 + }, + { + "epoch": 1.5088411143620406, + "grad_norm": 2.5574610233306885, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8909848928451538, + "num_tokens": 460449540.0, + "step": 11861 + }, + { + "epoch": 1.508968324640631, + "grad_norm": 2.6077239513397217, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8735451698303223, + "num_tokens": 460493182.0, + "step": 11862 + }, + { + "epoch": 1.5090955349192214, + "grad_norm": 2.911043643951416, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.893631100654602, + "num_tokens": 460527350.0, + "step": 11863 + }, + { + "epoch": 1.509222745197812, + "grad_norm": 2.760908842086792, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8832838535308838, + "num_tokens": 460565908.0, + "step": 11864 + }, + { + "epoch": 1.5093499554764025, + "grad_norm": 2.557013511657715, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9041702151298523, + "num_tokens": 460601055.0, + "step": 11865 + }, + { + "epoch": 1.509477165754993, + "grad_norm": 2.5448789596557617, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8854019045829773, + "num_tokens": 460644586.0, + "step": 11866 + }, + { + "epoch": 1.5096043760335836, + "grad_norm": 2.5982227325439453, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8925220370292664, + "num_tokens": 460681118.0, + "step": 11867 + }, + { + "epoch": 1.509731586312174, + "grad_norm": 2.7312307357788086, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.881547212600708, + "num_tokens": 460720582.0, + "step": 11868 + }, + { + "epoch": 1.5098587965907644, + "grad_norm": 2.520674467086792, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8963212370872498, + "num_tokens": 460764384.0, + "step": 11869 + }, + { + "epoch": 1.509986006869355, + "grad_norm": 2.495250940322876, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8994894027709961, + "num_tokens": 460804958.0, + "step": 11870 + }, + { + "epoch": 1.5101132171479454, + "grad_norm": 2.674420118331909, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.892457127571106, + "num_tokens": 460843039.0, + "step": 11871 + }, + { + "epoch": 1.510240427426536, + "grad_norm": 2.6494476795196533, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.901828944683075, + "num_tokens": 460875044.0, + "step": 11872 + }, + { + "epoch": 1.5103676377051265, + "grad_norm": 2.5385348796844482, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8864809274673462, + "num_tokens": 460912970.0, + "step": 11873 + }, + { + "epoch": 1.510494847983717, + "grad_norm": 2.5234055519104004, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8947511911392212, + "num_tokens": 460952956.0, + "step": 11874 + }, + { + "epoch": 1.5106220582623076, + "grad_norm": 2.5243263244628906, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.898350715637207, + "num_tokens": 460992639.0, + "step": 11875 + }, + { + "epoch": 1.510749268540898, + "grad_norm": 2.7178168296813965, + "learning_rate": 1e-06, + "loss": 0.3343, + "mean_token_accuracy": 0.8795273303985596, + "num_tokens": 461031008.0, + "step": 11876 + }, + { + "epoch": 1.5108764788194886, + "grad_norm": 2.6344172954559326, + "learning_rate": 1e-06, + "loss": 0.3443, + "mean_token_accuracy": 0.8784995675086975, + "num_tokens": 461076069.0, + "step": 11877 + }, + { + "epoch": 1.5110036890980791, + "grad_norm": 2.574366569519043, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8907672166824341, + "num_tokens": 461112447.0, + "step": 11878 + }, + { + "epoch": 1.5111308993766697, + "grad_norm": 2.411379098892212, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9028476476669312, + "num_tokens": 461155952.0, + "step": 11879 + }, + { + "epoch": 1.5112581096552602, + "grad_norm": 2.6618449687957764, + "learning_rate": 1e-06, + "loss": 0.3494, + "mean_token_accuracy": 0.8750184774398804, + "num_tokens": 461197556.0, + "step": 11880 + }, + { + "epoch": 1.5113853199338507, + "grad_norm": 2.5627901554107666, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8943679332733154, + "num_tokens": 461238080.0, + "step": 11881 + }, + { + "epoch": 1.5115125302124413, + "grad_norm": 2.7388017177581787, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8737422823905945, + "num_tokens": 461280380.0, + "step": 11882 + }, + { + "epoch": 1.5116397404910318, + "grad_norm": 2.869882345199585, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8907204866409302, + "num_tokens": 461315547.0, + "step": 11883 + }, + { + "epoch": 1.5117669507696223, + "grad_norm": 2.753612518310547, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8856382369995117, + "num_tokens": 461354563.0, + "step": 11884 + }, + { + "epoch": 1.5118941610482128, + "grad_norm": 2.871011972427368, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8780555725097656, + "num_tokens": 461388784.0, + "step": 11885 + }, + { + "epoch": 1.5120213713268034, + "grad_norm": 2.524444103240967, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9050006866455078, + "num_tokens": 461425623.0, + "step": 11886 + }, + { + "epoch": 1.5121485816053937, + "grad_norm": 2.5797624588012695, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8906153440475464, + "num_tokens": 461465753.0, + "step": 11887 + }, + { + "epoch": 1.5122757918839842, + "grad_norm": 2.692603588104248, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.896089494228363, + "num_tokens": 461504216.0, + "step": 11888 + }, + { + "epoch": 1.5124030021625747, + "grad_norm": 2.543091058731079, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9034054279327393, + "num_tokens": 461541771.0, + "step": 11889 + }, + { + "epoch": 1.5125302124411653, + "grad_norm": 2.727847099304199, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8913960456848145, + "num_tokens": 461584043.0, + "step": 11890 + }, + { + "epoch": 1.5126574227197558, + "grad_norm": 2.3774006366729736, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8982585072517395, + "num_tokens": 461625197.0, + "step": 11891 + }, + { + "epoch": 1.5127846329983463, + "grad_norm": 2.3789868354797363, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8912250399589539, + "num_tokens": 461670354.0, + "step": 11892 + }, + { + "epoch": 1.5129118432769366, + "grad_norm": 2.7036032676696777, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8946524858474731, + "num_tokens": 461706760.0, + "step": 11893 + }, + { + "epoch": 1.5130390535555271, + "grad_norm": 2.630091667175293, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.892630934715271, + "num_tokens": 461747793.0, + "step": 11894 + }, + { + "epoch": 1.5131662638341177, + "grad_norm": 2.8105216026306152, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8898651599884033, + "num_tokens": 461784429.0, + "step": 11895 + }, + { + "epoch": 1.5132934741127082, + "grad_norm": 2.671743869781494, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8831667304039001, + "num_tokens": 461822997.0, + "step": 11896 + }, + { + "epoch": 1.5134206843912987, + "grad_norm": 2.7181851863861084, + "learning_rate": 1e-06, + "loss": 0.325, + "mean_token_accuracy": 0.8818237781524658, + "num_tokens": 461862331.0, + "step": 11897 + }, + { + "epoch": 1.5135478946698893, + "grad_norm": 2.5603532791137695, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8918746709823608, + "num_tokens": 461899904.0, + "step": 11898 + }, + { + "epoch": 1.5136751049484798, + "grad_norm": 2.8304355144500732, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.900759756565094, + "num_tokens": 461931833.0, + "step": 11899 + }, + { + "epoch": 1.5138023152270703, + "grad_norm": 2.7358789443969727, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8790934085845947, + "num_tokens": 461972202.0, + "step": 11900 + }, + { + "epoch": 1.5139295255056608, + "grad_norm": 4.2300920486450195, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8729997873306274, + "num_tokens": 462010282.0, + "step": 11901 + }, + { + "epoch": 1.5140567357842514, + "grad_norm": 2.726583242416382, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8858586549758911, + "num_tokens": 462048292.0, + "step": 11902 + }, + { + "epoch": 1.514183946062842, + "grad_norm": 2.580984115600586, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8827379941940308, + "num_tokens": 462093772.0, + "step": 11903 + }, + { + "epoch": 1.5143111563414324, + "grad_norm": 2.6164116859436035, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8804054856300354, + "num_tokens": 462137654.0, + "step": 11904 + }, + { + "epoch": 1.514438366620023, + "grad_norm": 2.6592085361480713, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8971152901649475, + "num_tokens": 462171365.0, + "step": 11905 + }, + { + "epoch": 1.5145655768986135, + "grad_norm": 2.310441732406616, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9054717421531677, + "num_tokens": 462212501.0, + "step": 11906 + }, + { + "epoch": 1.514692787177204, + "grad_norm": 2.751375913619995, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8791790008544922, + "num_tokens": 462252940.0, + "step": 11907 + }, + { + "epoch": 1.5148199974557945, + "grad_norm": 3.074345588684082, + "learning_rate": 1e-06, + "loss": 0.3748, + "mean_token_accuracy": 0.86761873960495, + "num_tokens": 462287837.0, + "step": 11908 + }, + { + "epoch": 1.514947207734385, + "grad_norm": 2.7780873775482178, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9011703729629517, + "num_tokens": 462319453.0, + "step": 11909 + }, + { + "epoch": 1.5150744180129756, + "grad_norm": 2.751291036605835, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9021041393280029, + "num_tokens": 462359451.0, + "step": 11910 + }, + { + "epoch": 1.515201628291566, + "grad_norm": 2.760660409927368, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8870612382888794, + "num_tokens": 462396161.0, + "step": 11911 + }, + { + "epoch": 1.5153288385701564, + "grad_norm": 2.4401445388793945, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8962564468383789, + "num_tokens": 462438455.0, + "step": 11912 + }, + { + "epoch": 1.515456048848747, + "grad_norm": 2.503546714782715, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8950284719467163, + "num_tokens": 462475985.0, + "step": 11913 + }, + { + "epoch": 1.5155832591273375, + "grad_norm": 2.5057218074798584, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8997397422790527, + "num_tokens": 462516480.0, + "step": 11914 + }, + { + "epoch": 1.515710469405928, + "grad_norm": 2.5615901947021484, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8867711424827576, + "num_tokens": 462557105.0, + "step": 11915 + }, + { + "epoch": 1.5158376796845185, + "grad_norm": 2.59840726852417, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8954405784606934, + "num_tokens": 462594926.0, + "step": 11916 + }, + { + "epoch": 1.515964889963109, + "grad_norm": 2.6331305503845215, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8979775905609131, + "num_tokens": 462632377.0, + "step": 11917 + }, + { + "epoch": 1.5160921002416994, + "grad_norm": 2.510956048965454, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8943811058998108, + "num_tokens": 462676111.0, + "step": 11918 + }, + { + "epoch": 1.51621931052029, + "grad_norm": 2.6790859699249268, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8788175582885742, + "num_tokens": 462716609.0, + "step": 11919 + }, + { + "epoch": 1.5163465207988804, + "grad_norm": 2.652707576751709, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8847906589508057, + "num_tokens": 462754257.0, + "step": 11920 + }, + { + "epoch": 1.516473731077471, + "grad_norm": 2.6399736404418945, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8857320547103882, + "num_tokens": 462791952.0, + "step": 11921 + }, + { + "epoch": 1.5166009413560615, + "grad_norm": 2.648577928543091, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8891929388046265, + "num_tokens": 462832687.0, + "step": 11922 + }, + { + "epoch": 1.516728151634652, + "grad_norm": 2.844844341278076, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8966600298881531, + "num_tokens": 462867779.0, + "step": 11923 + }, + { + "epoch": 1.5168553619132426, + "grad_norm": 2.646078109741211, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8820568323135376, + "num_tokens": 462909234.0, + "step": 11924 + }, + { + "epoch": 1.516982572191833, + "grad_norm": 2.6548032760620117, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8976966142654419, + "num_tokens": 462943365.0, + "step": 11925 + }, + { + "epoch": 1.5171097824704236, + "grad_norm": 2.8564624786376953, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.891661524772644, + "num_tokens": 462979387.0, + "step": 11926 + }, + { + "epoch": 1.5172369927490141, + "grad_norm": 2.6159632205963135, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.894618034362793, + "num_tokens": 463019408.0, + "step": 11927 + }, + { + "epoch": 1.5173642030276047, + "grad_norm": 2.623603343963623, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8947815895080566, + "num_tokens": 463055781.0, + "step": 11928 + }, + { + "epoch": 1.5174914133061952, + "grad_norm": 2.7863738536834717, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8980445861816406, + "num_tokens": 463091236.0, + "step": 11929 + }, + { + "epoch": 1.5176186235847857, + "grad_norm": 2.4167141914367676, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8911818265914917, + "num_tokens": 463135623.0, + "step": 11930 + }, + { + "epoch": 1.5177458338633762, + "grad_norm": 2.61777400970459, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8897915482521057, + "num_tokens": 463176550.0, + "step": 11931 + }, + { + "epoch": 1.5178730441419668, + "grad_norm": 2.4951281547546387, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8994917869567871, + "num_tokens": 463216693.0, + "step": 11932 + }, + { + "epoch": 1.5180002544205573, + "grad_norm": 2.3932113647460938, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.90655517578125, + "num_tokens": 463254623.0, + "step": 11933 + }, + { + "epoch": 1.5181274646991478, + "grad_norm": 2.423452377319336, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8940262794494629, + "num_tokens": 463299177.0, + "step": 11934 + }, + { + "epoch": 1.5182546749777384, + "grad_norm": 2.610664129257202, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8946684002876282, + "num_tokens": 463336327.0, + "step": 11935 + }, + { + "epoch": 1.5183818852563287, + "grad_norm": 2.600095272064209, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8912991285324097, + "num_tokens": 463374268.0, + "step": 11936 + }, + { + "epoch": 1.5185090955349192, + "grad_norm": 2.424161195755005, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.894844114780426, + "num_tokens": 463416488.0, + "step": 11937 + }, + { + "epoch": 1.5186363058135097, + "grad_norm": 2.7677199840545654, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8916428089141846, + "num_tokens": 463454044.0, + "step": 11938 + }, + { + "epoch": 1.5187635160921003, + "grad_norm": 2.715754270553589, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8929277062416077, + "num_tokens": 463489795.0, + "step": 11939 + }, + { + "epoch": 1.5188907263706908, + "grad_norm": 2.5744245052337646, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9079465270042419, + "num_tokens": 463525895.0, + "step": 11940 + }, + { + "epoch": 1.5190179366492813, + "grad_norm": 2.685447931289673, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8897847533226013, + "num_tokens": 463569414.0, + "step": 11941 + }, + { + "epoch": 1.5191451469278716, + "grad_norm": 2.370777130126953, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8965993523597717, + "num_tokens": 463613392.0, + "step": 11942 + }, + { + "epoch": 1.5192723572064621, + "grad_norm": 2.6586902141571045, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.88665771484375, + "num_tokens": 463653061.0, + "step": 11943 + }, + { + "epoch": 1.5193995674850527, + "grad_norm": 2.3083043098449707, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.88929283618927, + "num_tokens": 463701138.0, + "step": 11944 + }, + { + "epoch": 1.5195267777636432, + "grad_norm": 3.164447546005249, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8887614011764526, + "num_tokens": 463730321.0, + "step": 11945 + }, + { + "epoch": 1.5196539880422337, + "grad_norm": 2.743563413619995, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8864319324493408, + "num_tokens": 463769601.0, + "step": 11946 + }, + { + "epoch": 1.5197811983208243, + "grad_norm": 2.432741403579712, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8992542624473572, + "num_tokens": 463808557.0, + "step": 11947 + }, + { + "epoch": 1.5199084085994148, + "grad_norm": 2.446706533432007, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8925076723098755, + "num_tokens": 463852347.0, + "step": 11948 + }, + { + "epoch": 1.5200356188780053, + "grad_norm": 2.830538511276245, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8837172985076904, + "num_tokens": 463889661.0, + "step": 11949 + }, + { + "epoch": 1.5201628291565958, + "grad_norm": 2.8176796436309814, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8913671970367432, + "num_tokens": 463925737.0, + "step": 11950 + }, + { + "epoch": 1.5202900394351864, + "grad_norm": 2.614927291870117, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8859174847602844, + "num_tokens": 463965563.0, + "step": 11951 + }, + { + "epoch": 1.520417249713777, + "grad_norm": 2.649052619934082, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8869123458862305, + "num_tokens": 464005841.0, + "step": 11952 + }, + { + "epoch": 1.5205444599923674, + "grad_norm": 2.412775754928589, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8851574063301086, + "num_tokens": 464052405.0, + "step": 11953 + }, + { + "epoch": 1.520671670270958, + "grad_norm": 2.780601739883423, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8942176103591919, + "num_tokens": 464090080.0, + "step": 11954 + }, + { + "epoch": 1.5207988805495485, + "grad_norm": 2.4273014068603516, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8829220533370972, + "num_tokens": 464133794.0, + "step": 11955 + }, + { + "epoch": 1.520926090828139, + "grad_norm": 2.5253095626831055, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8957791328430176, + "num_tokens": 464172889.0, + "step": 11956 + }, + { + "epoch": 1.5210533011067295, + "grad_norm": 2.4910969734191895, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9083814024925232, + "num_tokens": 464210583.0, + "step": 11957 + }, + { + "epoch": 1.52118051138532, + "grad_norm": 2.477428913116455, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9005104899406433, + "num_tokens": 464246076.0, + "step": 11958 + }, + { + "epoch": 1.5213077216639106, + "grad_norm": 2.6858675479888916, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8943302631378174, + "num_tokens": 464286254.0, + "step": 11959 + }, + { + "epoch": 1.521434931942501, + "grad_norm": 2.636000633239746, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8985695838928223, + "num_tokens": 464326539.0, + "step": 11960 + }, + { + "epoch": 1.5215621422210914, + "grad_norm": 2.788673162460327, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8828043937683105, + "num_tokens": 464363540.0, + "step": 11961 + }, + { + "epoch": 1.521689352499682, + "grad_norm": 2.680284023284912, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8904098272323608, + "num_tokens": 464400748.0, + "step": 11962 + }, + { + "epoch": 1.5218165627782725, + "grad_norm": 2.505405902862549, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9005466103553772, + "num_tokens": 464440554.0, + "step": 11963 + }, + { + "epoch": 1.521943773056863, + "grad_norm": 2.5477685928344727, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8864341974258423, + "num_tokens": 464481670.0, + "step": 11964 + }, + { + "epoch": 1.5220709833354535, + "grad_norm": 2.6372251510620117, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8828286528587341, + "num_tokens": 464528338.0, + "step": 11965 + }, + { + "epoch": 1.5221981936140438, + "grad_norm": 3.0132970809936523, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8915271759033203, + "num_tokens": 464559026.0, + "step": 11966 + }, + { + "epoch": 1.5223254038926344, + "grad_norm": 2.6281626224517822, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.885087788105011, + "num_tokens": 464599450.0, + "step": 11967 + }, + { + "epoch": 1.522452614171225, + "grad_norm": 2.8619277477264404, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.897588849067688, + "num_tokens": 464631980.0, + "step": 11968 + }, + { + "epoch": 1.5225798244498154, + "grad_norm": 2.556094169616699, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8887732028961182, + "num_tokens": 464676316.0, + "step": 11969 + }, + { + "epoch": 1.522707034728406, + "grad_norm": 2.6629574298858643, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8978333473205566, + "num_tokens": 464715373.0, + "step": 11970 + }, + { + "epoch": 1.5228342450069965, + "grad_norm": 2.621645927429199, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8915808796882629, + "num_tokens": 464755951.0, + "step": 11971 + }, + { + "epoch": 1.522961455285587, + "grad_norm": 2.770153760910034, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.898219108581543, + "num_tokens": 464789059.0, + "step": 11972 + }, + { + "epoch": 1.5230886655641775, + "grad_norm": 2.6462881565093994, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.885387659072876, + "num_tokens": 464829326.0, + "step": 11973 + }, + { + "epoch": 1.523215875842768, + "grad_norm": 2.8522071838378906, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8887743949890137, + "num_tokens": 464869431.0, + "step": 11974 + }, + { + "epoch": 1.5233430861213586, + "grad_norm": 2.6403987407684326, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8937827944755554, + "num_tokens": 464905947.0, + "step": 11975 + }, + { + "epoch": 1.5234702963999491, + "grad_norm": 2.8239059448242188, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8913775086402893, + "num_tokens": 464937458.0, + "step": 11976 + }, + { + "epoch": 1.5235975066785397, + "grad_norm": 2.6995043754577637, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8778485655784607, + "num_tokens": 464982010.0, + "step": 11977 + }, + { + "epoch": 1.5237247169571302, + "grad_norm": 2.6144163608551025, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8965683579444885, + "num_tokens": 465022742.0, + "step": 11978 + }, + { + "epoch": 1.5238519272357207, + "grad_norm": 2.6975059509277344, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8931082487106323, + "num_tokens": 465062321.0, + "step": 11979 + }, + { + "epoch": 1.5239791375143112, + "grad_norm": 2.972796678543091, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8923575282096863, + "num_tokens": 465094336.0, + "step": 11980 + }, + { + "epoch": 1.5241063477929018, + "grad_norm": 2.6927640438079834, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8810280561447144, + "num_tokens": 465138807.0, + "step": 11981 + }, + { + "epoch": 1.5242335580714923, + "grad_norm": 2.6997992992401123, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8899153470993042, + "num_tokens": 465175073.0, + "step": 11982 + }, + { + "epoch": 1.5243607683500828, + "grad_norm": 2.6146881580352783, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8890820145606995, + "num_tokens": 465214749.0, + "step": 11983 + }, + { + "epoch": 1.5244879786286734, + "grad_norm": 2.779714584350586, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8814065456390381, + "num_tokens": 465253965.0, + "step": 11984 + }, + { + "epoch": 1.5246151889072637, + "grad_norm": 2.6693191528320312, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8948816061019897, + "num_tokens": 465291495.0, + "step": 11985 + }, + { + "epoch": 1.5247423991858542, + "grad_norm": 2.5806281566619873, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.888611912727356, + "num_tokens": 465333718.0, + "step": 11986 + }, + { + "epoch": 1.5248696094644447, + "grad_norm": 2.4411561489105225, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8995184898376465, + "num_tokens": 465374246.0, + "step": 11987 + }, + { + "epoch": 1.5249968197430352, + "grad_norm": 2.8363773822784424, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.900282621383667, + "num_tokens": 465408059.0, + "step": 11988 + }, + { + "epoch": 1.5251240300216258, + "grad_norm": 2.7453620433807373, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8820211887359619, + "num_tokens": 465447596.0, + "step": 11989 + }, + { + "epoch": 1.5252512403002163, + "grad_norm": 2.7363147735595703, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8966601490974426, + "num_tokens": 465483130.0, + "step": 11990 + }, + { + "epoch": 1.5253784505788066, + "grad_norm": 2.451481342315674, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8983911275863647, + "num_tokens": 465524192.0, + "step": 11991 + }, + { + "epoch": 1.5255056608573971, + "grad_norm": 2.7316231727600098, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.881950855255127, + "num_tokens": 465561343.0, + "step": 11992 + }, + { + "epoch": 1.5256328711359877, + "grad_norm": 2.9359967708587646, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8804978132247925, + "num_tokens": 465598020.0, + "step": 11993 + }, + { + "epoch": 1.5257600814145782, + "grad_norm": 2.7437963485717773, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8903295993804932, + "num_tokens": 465632454.0, + "step": 11994 + }, + { + "epoch": 1.5258872916931687, + "grad_norm": 2.445491313934326, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8911333084106445, + "num_tokens": 465677406.0, + "step": 11995 + }, + { + "epoch": 1.5260145019717593, + "grad_norm": 2.504439353942871, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8956031799316406, + "num_tokens": 465720384.0, + "step": 11996 + }, + { + "epoch": 1.5261417122503498, + "grad_norm": 2.9096555709838867, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8974542021751404, + "num_tokens": 465751286.0, + "step": 11997 + }, + { + "epoch": 1.5262689225289403, + "grad_norm": 2.5947201251983643, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.8972090482711792, + "num_tokens": 465788451.0, + "step": 11998 + }, + { + "epoch": 1.5263961328075308, + "grad_norm": 2.617119550704956, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8919051885604858, + "num_tokens": 465825412.0, + "step": 11999 + }, + { + "epoch": 1.5265233430861214, + "grad_norm": 2.4268686771392822, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8980915546417236, + "num_tokens": 465866806.0, + "step": 12000 + }, + { + "epoch": 1.526650553364712, + "grad_norm": 2.8069684505462646, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8910117149353027, + "num_tokens": 465903978.0, + "step": 12001 + }, + { + "epoch": 1.5267777636433024, + "grad_norm": 2.5131804943084717, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8918540477752686, + "num_tokens": 465949390.0, + "step": 12002 + }, + { + "epoch": 1.526904973921893, + "grad_norm": 2.637420415878296, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9007011651992798, + "num_tokens": 465987620.0, + "step": 12003 + }, + { + "epoch": 1.5270321842004835, + "grad_norm": 2.611912250518799, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8955050110816956, + "num_tokens": 466026776.0, + "step": 12004 + }, + { + "epoch": 1.527159394479074, + "grad_norm": 2.6416592597961426, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8972489237785339, + "num_tokens": 466063685.0, + "step": 12005 + }, + { + "epoch": 1.5272866047576645, + "grad_norm": 2.8016409873962402, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8778386116027832, + "num_tokens": 466100249.0, + "step": 12006 + }, + { + "epoch": 1.527413815036255, + "grad_norm": 2.7841007709503174, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8810704946517944, + "num_tokens": 466138943.0, + "step": 12007 + }, + { + "epoch": 1.5275410253148456, + "grad_norm": 2.769890069961548, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8931070566177368, + "num_tokens": 466171251.0, + "step": 12008 + }, + { + "epoch": 1.527668235593436, + "grad_norm": 2.618995189666748, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8911874294281006, + "num_tokens": 466213210.0, + "step": 12009 + }, + { + "epoch": 1.5277954458720264, + "grad_norm": 2.65867018699646, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8898926973342896, + "num_tokens": 466251731.0, + "step": 12010 + }, + { + "epoch": 1.527922656150617, + "grad_norm": 2.574801445007324, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8943921327590942, + "num_tokens": 466290649.0, + "step": 12011 + }, + { + "epoch": 1.5280498664292075, + "grad_norm": 2.640366554260254, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9027008414268494, + "num_tokens": 466329897.0, + "step": 12012 + }, + { + "epoch": 1.528177076707798, + "grad_norm": 2.8375844955444336, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9003456830978394, + "num_tokens": 466360225.0, + "step": 12013 + }, + { + "epoch": 1.5283042869863885, + "grad_norm": 2.676276206970215, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9048737287521362, + "num_tokens": 466395644.0, + "step": 12014 + }, + { + "epoch": 1.5284314972649788, + "grad_norm": 2.55871844291687, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8911648392677307, + "num_tokens": 466440808.0, + "step": 12015 + }, + { + "epoch": 1.5285587075435694, + "grad_norm": 2.6528067588806152, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8926655054092407, + "num_tokens": 466479869.0, + "step": 12016 + }, + { + "epoch": 1.52868591782216, + "grad_norm": 2.879340887069702, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8905816078186035, + "num_tokens": 466512506.0, + "step": 12017 + }, + { + "epoch": 1.5288131281007504, + "grad_norm": 2.5384700298309326, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8946831226348877, + "num_tokens": 466551948.0, + "step": 12018 + }, + { + "epoch": 1.528940338379341, + "grad_norm": 2.5370028018951416, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.894767701625824, + "num_tokens": 466591326.0, + "step": 12019 + }, + { + "epoch": 1.5290675486579315, + "grad_norm": 2.6879918575286865, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8848693370819092, + "num_tokens": 466628064.0, + "step": 12020 + }, + { + "epoch": 1.529194758936522, + "grad_norm": 2.673452138900757, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8907246589660645, + "num_tokens": 466666644.0, + "step": 12021 + }, + { + "epoch": 1.5293219692151125, + "grad_norm": 2.63978910446167, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8918322324752808, + "num_tokens": 466705745.0, + "step": 12022 + }, + { + "epoch": 1.529449179493703, + "grad_norm": 2.9284462928771973, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8904911875724792, + "num_tokens": 466742038.0, + "step": 12023 + }, + { + "epoch": 1.5295763897722936, + "grad_norm": 2.3592755794525146, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8947994112968445, + "num_tokens": 466789568.0, + "step": 12024 + }, + { + "epoch": 1.5297036000508841, + "grad_norm": 2.5216526985168457, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8875396847724915, + "num_tokens": 466831597.0, + "step": 12025 + }, + { + "epoch": 1.5298308103294747, + "grad_norm": 2.4965479373931885, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9024837017059326, + "num_tokens": 466876391.0, + "step": 12026 + }, + { + "epoch": 1.5299580206080652, + "grad_norm": 2.4913434982299805, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9090648889541626, + "num_tokens": 466910930.0, + "step": 12027 + }, + { + "epoch": 1.5300852308866557, + "grad_norm": 2.385617971420288, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8941534161567688, + "num_tokens": 466954252.0, + "step": 12028 + }, + { + "epoch": 1.5302124411652462, + "grad_norm": 2.627413511276245, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8988111019134521, + "num_tokens": 466989916.0, + "step": 12029 + }, + { + "epoch": 1.5303396514438368, + "grad_norm": 2.5520365238189697, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8976655006408691, + "num_tokens": 467031269.0, + "step": 12030 + }, + { + "epoch": 1.5304668617224273, + "grad_norm": 2.6822214126586914, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9004030823707581, + "num_tokens": 467065253.0, + "step": 12031 + }, + { + "epoch": 1.5305940720010178, + "grad_norm": 2.8298227787017822, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8829102516174316, + "num_tokens": 467100667.0, + "step": 12032 + }, + { + "epoch": 1.5307212822796084, + "grad_norm": 2.64270281791687, + "learning_rate": 1e-06, + "loss": 0.3588, + "mean_token_accuracy": 0.8770554065704346, + "num_tokens": 467143043.0, + "step": 12033 + }, + { + "epoch": 1.5308484925581987, + "grad_norm": 2.4938466548919678, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8891767263412476, + "num_tokens": 467182370.0, + "step": 12034 + }, + { + "epoch": 1.5309757028367892, + "grad_norm": 2.626107692718506, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8916240930557251, + "num_tokens": 467220714.0, + "step": 12035 + }, + { + "epoch": 1.5311029131153797, + "grad_norm": 2.7755260467529297, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8837494850158691, + "num_tokens": 467259413.0, + "step": 12036 + }, + { + "epoch": 1.5312301233939702, + "grad_norm": 2.6704330444335938, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8931764364242554, + "num_tokens": 467301184.0, + "step": 12037 + }, + { + "epoch": 1.5313573336725608, + "grad_norm": 2.7451024055480957, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8831029534339905, + "num_tokens": 467340037.0, + "step": 12038 + }, + { + "epoch": 1.5314845439511513, + "grad_norm": 2.4716403484344482, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8899691104888916, + "num_tokens": 467383115.0, + "step": 12039 + }, + { + "epoch": 1.5316117542297416, + "grad_norm": 2.333908796310425, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9052871465682983, + "num_tokens": 467425759.0, + "step": 12040 + }, + { + "epoch": 1.5317389645083321, + "grad_norm": 2.587747812271118, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8856596946716309, + "num_tokens": 467467379.0, + "step": 12041 + }, + { + "epoch": 1.5318661747869227, + "grad_norm": 2.738874912261963, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8955614566802979, + "num_tokens": 467503816.0, + "step": 12042 + }, + { + "epoch": 1.5319933850655132, + "grad_norm": 2.744290828704834, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8772662281990051, + "num_tokens": 467543828.0, + "step": 12043 + }, + { + "epoch": 1.5321205953441037, + "grad_norm": 3.279881000518799, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8836692571640015, + "num_tokens": 467573786.0, + "step": 12044 + }, + { + "epoch": 1.5322478056226942, + "grad_norm": 2.565070152282715, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8972411155700684, + "num_tokens": 467611568.0, + "step": 12045 + }, + { + "epoch": 1.5323750159012848, + "grad_norm": 3.0569229125976562, + "learning_rate": 1e-06, + "loss": 0.3673, + "mean_token_accuracy": 0.8737083077430725, + "num_tokens": 467646252.0, + "step": 12046 + }, + { + "epoch": 1.5325022261798753, + "grad_norm": 2.7504613399505615, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8863334059715271, + "num_tokens": 467681999.0, + "step": 12047 + }, + { + "epoch": 1.5326294364584658, + "grad_norm": 2.514594316482544, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8906431794166565, + "num_tokens": 467721200.0, + "step": 12048 + }, + { + "epoch": 1.5327566467370564, + "grad_norm": 2.5389885902404785, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8948248624801636, + "num_tokens": 467762328.0, + "step": 12049 + }, + { + "epoch": 1.532883857015647, + "grad_norm": 2.5920495986938477, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8773649334907532, + "num_tokens": 467803879.0, + "step": 12050 + }, + { + "epoch": 1.5330110672942374, + "grad_norm": 2.695843458175659, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8809632658958435, + "num_tokens": 467844214.0, + "step": 12051 + }, + { + "epoch": 1.533138277572828, + "grad_norm": 2.5091350078582764, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8880475759506226, + "num_tokens": 467886390.0, + "step": 12052 + }, + { + "epoch": 1.5332654878514185, + "grad_norm": 2.660274028778076, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8839176893234253, + "num_tokens": 467925613.0, + "step": 12053 + }, + { + "epoch": 1.533392698130009, + "grad_norm": 3.2183690071105957, + "learning_rate": 1e-06, + "loss": 0.3458, + "mean_token_accuracy": 0.8782495260238647, + "num_tokens": 467954955.0, + "step": 12054 + }, + { + "epoch": 1.5335199084085995, + "grad_norm": 2.7511534690856934, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.8828242421150208, + "num_tokens": 467992346.0, + "step": 12055 + }, + { + "epoch": 1.53364711868719, + "grad_norm": 2.567936897277832, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8866517543792725, + "num_tokens": 468034696.0, + "step": 12056 + }, + { + "epoch": 1.5337743289657806, + "grad_norm": 2.390822172164917, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9088075160980225, + "num_tokens": 468075738.0, + "step": 12057 + }, + { + "epoch": 1.533901539244371, + "grad_norm": 2.6733646392822266, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8882590532302856, + "num_tokens": 468117577.0, + "step": 12058 + }, + { + "epoch": 1.5340287495229614, + "grad_norm": 2.7643697261810303, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8823455572128296, + "num_tokens": 468156754.0, + "step": 12059 + }, + { + "epoch": 1.534155959801552, + "grad_norm": 2.641350507736206, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8844525814056396, + "num_tokens": 468197316.0, + "step": 12060 + }, + { + "epoch": 1.5342831700801425, + "grad_norm": 2.689635992050171, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.8986657857894897, + "num_tokens": 468231161.0, + "step": 12061 + }, + { + "epoch": 1.534410380358733, + "grad_norm": 2.369141101837158, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9044660329818726, + "num_tokens": 468271205.0, + "step": 12062 + }, + { + "epoch": 1.5345375906373235, + "grad_norm": 2.5546658039093018, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8957924246788025, + "num_tokens": 468310435.0, + "step": 12063 + }, + { + "epoch": 1.5346648009159138, + "grad_norm": 2.7139289379119873, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.8955918550491333, + "num_tokens": 468344873.0, + "step": 12064 + }, + { + "epoch": 1.5347920111945044, + "grad_norm": 2.6698429584503174, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8913732767105103, + "num_tokens": 468384266.0, + "step": 12065 + }, + { + "epoch": 1.534919221473095, + "grad_norm": 2.8654191493988037, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9002068042755127, + "num_tokens": 468413758.0, + "step": 12066 + }, + { + "epoch": 1.5350464317516854, + "grad_norm": 2.695713996887207, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8983180522918701, + "num_tokens": 468448012.0, + "step": 12067 + }, + { + "epoch": 1.535173642030276, + "grad_norm": 2.6184017658233643, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8821120262145996, + "num_tokens": 468491449.0, + "step": 12068 + }, + { + "epoch": 1.5353008523088665, + "grad_norm": 2.425023078918457, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9078490138053894, + "num_tokens": 468529689.0, + "step": 12069 + }, + { + "epoch": 1.535428062587457, + "grad_norm": 2.519965648651123, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9004286527633667, + "num_tokens": 468566546.0, + "step": 12070 + }, + { + "epoch": 1.5355552728660475, + "grad_norm": 2.682069778442383, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8848426342010498, + "num_tokens": 468603866.0, + "step": 12071 + }, + { + "epoch": 1.535682483144638, + "grad_norm": 2.692065954208374, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8873403072357178, + "num_tokens": 468642076.0, + "step": 12072 + }, + { + "epoch": 1.5358096934232286, + "grad_norm": 2.7834043502807617, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.880865216255188, + "num_tokens": 468681738.0, + "step": 12073 + }, + { + "epoch": 1.5359369037018191, + "grad_norm": 2.5457675457000732, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8961369395256042, + "num_tokens": 468724039.0, + "step": 12074 + }, + { + "epoch": 1.5360641139804097, + "grad_norm": 2.5906715393066406, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8940734267234802, + "num_tokens": 468763947.0, + "step": 12075 + }, + { + "epoch": 1.5361913242590002, + "grad_norm": 2.382575035095215, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8984221816062927, + "num_tokens": 468807132.0, + "step": 12076 + }, + { + "epoch": 1.5363185345375907, + "grad_norm": 2.6228883266448975, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9024688005447388, + "num_tokens": 468844713.0, + "step": 12077 + }, + { + "epoch": 1.5364457448161812, + "grad_norm": 2.4162509441375732, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8961639404296875, + "num_tokens": 468886270.0, + "step": 12078 + }, + { + "epoch": 1.5365729550947718, + "grad_norm": 2.9624414443969727, + "learning_rate": 1e-06, + "loss": 0.3359, + "mean_token_accuracy": 0.8820739388465881, + "num_tokens": 468925482.0, + "step": 12079 + }, + { + "epoch": 1.5367001653733623, + "grad_norm": 2.764589309692383, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8846111297607422, + "num_tokens": 468967934.0, + "step": 12080 + }, + { + "epoch": 1.5368273756519528, + "grad_norm": 2.949390411376953, + "learning_rate": 1e-06, + "loss": 0.3492, + "mean_token_accuracy": 0.8790711164474487, + "num_tokens": 469008144.0, + "step": 12081 + }, + { + "epoch": 1.5369545859305433, + "grad_norm": 3.074965476989746, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8793659210205078, + "num_tokens": 469040978.0, + "step": 12082 + }, + { + "epoch": 1.5370817962091337, + "grad_norm": 2.904306650161743, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8866001963615417, + "num_tokens": 469077268.0, + "step": 12083 + }, + { + "epoch": 1.5372090064877242, + "grad_norm": 2.827183246612549, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8832006454467773, + "num_tokens": 469113974.0, + "step": 12084 + }, + { + "epoch": 1.5373362167663147, + "grad_norm": 2.4997003078460693, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8872094750404358, + "num_tokens": 469156832.0, + "step": 12085 + }, + { + "epoch": 1.5374634270449052, + "grad_norm": 2.5641002655029297, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8948968648910522, + "num_tokens": 469198271.0, + "step": 12086 + }, + { + "epoch": 1.5375906373234958, + "grad_norm": 2.5458383560180664, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8872005343437195, + "num_tokens": 469239178.0, + "step": 12087 + }, + { + "epoch": 1.5377178476020863, + "grad_norm": 2.803398847579956, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.891310453414917, + "num_tokens": 469273920.0, + "step": 12088 + }, + { + "epoch": 1.5378450578806766, + "grad_norm": 2.448134660720825, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8950417041778564, + "num_tokens": 469319295.0, + "step": 12089 + }, + { + "epoch": 1.5379722681592671, + "grad_norm": 2.817142963409424, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8884586095809937, + "num_tokens": 469353685.0, + "step": 12090 + }, + { + "epoch": 1.5380994784378577, + "grad_norm": 2.9365592002868652, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8877450227737427, + "num_tokens": 469385791.0, + "step": 12091 + }, + { + "epoch": 1.5382266887164482, + "grad_norm": 2.7872838973999023, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8930162191390991, + "num_tokens": 469418941.0, + "step": 12092 + }, + { + "epoch": 1.5383538989950387, + "grad_norm": 2.760004758834839, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8884280920028687, + "num_tokens": 469455069.0, + "step": 12093 + }, + { + "epoch": 1.5384811092736292, + "grad_norm": 2.3548641204833984, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9046280384063721, + "num_tokens": 469497475.0, + "step": 12094 + }, + { + "epoch": 1.5386083195522198, + "grad_norm": 2.561586380004883, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.89316725730896, + "num_tokens": 469540237.0, + "step": 12095 + }, + { + "epoch": 1.5387355298308103, + "grad_norm": 2.8168909549713135, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8833592534065247, + "num_tokens": 469577393.0, + "step": 12096 + }, + { + "epoch": 1.5388627401094008, + "grad_norm": 2.6425414085388184, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.902596652507782, + "num_tokens": 469612770.0, + "step": 12097 + }, + { + "epoch": 1.5389899503879914, + "grad_norm": 2.828922748565674, + "learning_rate": 1e-06, + "loss": 0.358, + "mean_token_accuracy": 0.8722100853919983, + "num_tokens": 469651062.0, + "step": 12098 + }, + { + "epoch": 1.5391171606665819, + "grad_norm": 2.7686142921447754, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.8732150793075562, + "num_tokens": 469688455.0, + "step": 12099 + }, + { + "epoch": 1.5392443709451724, + "grad_norm": 2.479222297668457, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9026470184326172, + "num_tokens": 469728730.0, + "step": 12100 + }, + { + "epoch": 1.539371581223763, + "grad_norm": 2.9073448181152344, + "learning_rate": 1e-06, + "loss": 0.3654, + "mean_token_accuracy": 0.8682147264480591, + "num_tokens": 469767040.0, + "step": 12101 + }, + { + "epoch": 1.5394987915023535, + "grad_norm": 3.4947731494903564, + "learning_rate": 1e-06, + "loss": 0.3351, + "mean_token_accuracy": 0.8805687427520752, + "num_tokens": 469804521.0, + "step": 12102 + }, + { + "epoch": 1.539626001780944, + "grad_norm": 2.6066300868988037, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.893221378326416, + "num_tokens": 469844776.0, + "step": 12103 + }, + { + "epoch": 1.5397532120595345, + "grad_norm": 2.782224655151367, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8866977095603943, + "num_tokens": 469881543.0, + "step": 12104 + }, + { + "epoch": 1.539880422338125, + "grad_norm": 2.831573247909546, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8903786540031433, + "num_tokens": 469915389.0, + "step": 12105 + }, + { + "epoch": 1.5400076326167156, + "grad_norm": 2.511023998260498, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8850448131561279, + "num_tokens": 469961155.0, + "step": 12106 + }, + { + "epoch": 1.5401348428953059, + "grad_norm": 2.9063727855682373, + "learning_rate": 1e-06, + "loss": 0.364, + "mean_token_accuracy": 0.8698115348815918, + "num_tokens": 470001277.0, + "step": 12107 + }, + { + "epoch": 1.5402620531738964, + "grad_norm": 2.897916793823242, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.890060544013977, + "num_tokens": 470039797.0, + "step": 12108 + }, + { + "epoch": 1.540389263452487, + "grad_norm": 2.995112180709839, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.891474187374115, + "num_tokens": 470070940.0, + "step": 12109 + }, + { + "epoch": 1.5405164737310775, + "grad_norm": 2.8082435131073, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.889407753944397, + "num_tokens": 470104626.0, + "step": 12110 + }, + { + "epoch": 1.540643684009668, + "grad_norm": 2.6719794273376465, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8882930278778076, + "num_tokens": 470141064.0, + "step": 12111 + }, + { + "epoch": 1.5407708942882585, + "grad_norm": 2.612272262573242, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8961275815963745, + "num_tokens": 470177329.0, + "step": 12112 + }, + { + "epoch": 1.5408981045668488, + "grad_norm": 2.64041805267334, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8873218894004822, + "num_tokens": 470215680.0, + "step": 12113 + }, + { + "epoch": 1.5410253148454394, + "grad_norm": 2.692105770111084, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8916715383529663, + "num_tokens": 470252832.0, + "step": 12114 + }, + { + "epoch": 1.54115252512403, + "grad_norm": 2.768662691116333, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8937941193580627, + "num_tokens": 470289212.0, + "step": 12115 + }, + { + "epoch": 1.5412797354026204, + "grad_norm": 2.648709535598755, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8864379525184631, + "num_tokens": 470330328.0, + "step": 12116 + }, + { + "epoch": 1.541406945681211, + "grad_norm": 2.950674295425415, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8992563486099243, + "num_tokens": 470367700.0, + "step": 12117 + }, + { + "epoch": 1.5415341559598015, + "grad_norm": 2.75687837600708, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8971372246742249, + "num_tokens": 470406369.0, + "step": 12118 + }, + { + "epoch": 1.541661366238392, + "grad_norm": 2.5348687171936035, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.893918514251709, + "num_tokens": 470446466.0, + "step": 12119 + }, + { + "epoch": 1.5417885765169825, + "grad_norm": 2.3147499561309814, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9108046293258667, + "num_tokens": 470486251.0, + "step": 12120 + }, + { + "epoch": 1.541915786795573, + "grad_norm": 2.608193874359131, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.8879666924476624, + "num_tokens": 470526472.0, + "step": 12121 + }, + { + "epoch": 1.5420429970741636, + "grad_norm": 2.4362430572509766, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.909654974937439, + "num_tokens": 470561463.0, + "step": 12122 + }, + { + "epoch": 1.5421702073527541, + "grad_norm": 2.5952906608581543, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8809715509414673, + "num_tokens": 470603312.0, + "step": 12123 + }, + { + "epoch": 1.5422974176313446, + "grad_norm": 2.67573618888855, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8905847072601318, + "num_tokens": 470643020.0, + "step": 12124 + }, + { + "epoch": 1.5424246279099352, + "grad_norm": 2.541163444519043, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.8978015184402466, + "num_tokens": 470679320.0, + "step": 12125 + }, + { + "epoch": 1.5425518381885257, + "grad_norm": 2.5491912364959717, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8911305665969849, + "num_tokens": 470720780.0, + "step": 12126 + }, + { + "epoch": 1.5426790484671162, + "grad_norm": 2.741816520690918, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8989343047142029, + "num_tokens": 470754814.0, + "step": 12127 + }, + { + "epoch": 1.5428062587457068, + "grad_norm": 2.7822768688201904, + "learning_rate": 1e-06, + "loss": 0.3463, + "mean_token_accuracy": 0.8742846250534058, + "num_tokens": 470794524.0, + "step": 12128 + }, + { + "epoch": 1.5429334690242973, + "grad_norm": 2.443667411804199, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8844643831253052, + "num_tokens": 470837745.0, + "step": 12129 + }, + { + "epoch": 1.5430606793028878, + "grad_norm": 2.7044448852539062, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8988361358642578, + "num_tokens": 470872012.0, + "step": 12130 + }, + { + "epoch": 1.5431878895814783, + "grad_norm": 2.475658416748047, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8929935097694397, + "num_tokens": 470916278.0, + "step": 12131 + }, + { + "epoch": 1.5433150998600687, + "grad_norm": 2.795769214630127, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.892302393913269, + "num_tokens": 470953699.0, + "step": 12132 + }, + { + "epoch": 1.5434423101386592, + "grad_norm": 2.6497578620910645, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8810553550720215, + "num_tokens": 470992746.0, + "step": 12133 + }, + { + "epoch": 1.5435695204172497, + "grad_norm": 2.8519179821014404, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8828847408294678, + "num_tokens": 471029853.0, + "step": 12134 + }, + { + "epoch": 1.5436967306958402, + "grad_norm": 3.0792720317840576, + "learning_rate": 1e-06, + "loss": 0.336, + "mean_token_accuracy": 0.8812770843505859, + "num_tokens": 471061860.0, + "step": 12135 + }, + { + "epoch": 1.5438239409744308, + "grad_norm": 2.8976099491119385, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8885859251022339, + "num_tokens": 471097004.0, + "step": 12136 + }, + { + "epoch": 1.5439511512530213, + "grad_norm": 2.519810914993286, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.901289701461792, + "num_tokens": 471137914.0, + "step": 12137 + }, + { + "epoch": 1.5440783615316116, + "grad_norm": 2.550818681716919, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8879754543304443, + "num_tokens": 471176774.0, + "step": 12138 + }, + { + "epoch": 1.5442055718102021, + "grad_norm": 2.6600162982940674, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8866935968399048, + "num_tokens": 471216169.0, + "step": 12139 + }, + { + "epoch": 1.5443327820887927, + "grad_norm": 2.415332078933716, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8979425430297852, + "num_tokens": 471260414.0, + "step": 12140 + }, + { + "epoch": 1.5444599923673832, + "grad_norm": 2.454286813735962, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9003756046295166, + "num_tokens": 471299516.0, + "step": 12141 + }, + { + "epoch": 1.5445872026459737, + "grad_norm": 2.5884523391723633, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8912318348884583, + "num_tokens": 471337297.0, + "step": 12142 + }, + { + "epoch": 1.5447144129245642, + "grad_norm": 2.792649269104004, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8945767879486084, + "num_tokens": 471369600.0, + "step": 12143 + }, + { + "epoch": 1.5448416232031548, + "grad_norm": 2.7472221851348877, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8852008581161499, + "num_tokens": 471409866.0, + "step": 12144 + }, + { + "epoch": 1.5449688334817453, + "grad_norm": 2.8016915321350098, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8957599401473999, + "num_tokens": 471441161.0, + "step": 12145 + }, + { + "epoch": 1.5450960437603358, + "grad_norm": 2.5616443157196045, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9006746411323547, + "num_tokens": 471477637.0, + "step": 12146 + }, + { + "epoch": 1.5452232540389264, + "grad_norm": 2.657081365585327, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8878252506256104, + "num_tokens": 471517769.0, + "step": 12147 + }, + { + "epoch": 1.5453504643175169, + "grad_norm": 2.5732076168060303, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8976714611053467, + "num_tokens": 471554478.0, + "step": 12148 + }, + { + "epoch": 1.5454776745961074, + "grad_norm": 2.662295341491699, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8799279928207397, + "num_tokens": 471595330.0, + "step": 12149 + }, + { + "epoch": 1.545604884874698, + "grad_norm": 2.894604206085205, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8719690442085266, + "num_tokens": 471634609.0, + "step": 12150 + }, + { + "epoch": 1.5457320951532885, + "grad_norm": 2.487433433532715, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8972065448760986, + "num_tokens": 471673401.0, + "step": 12151 + }, + { + "epoch": 1.545859305431879, + "grad_norm": 2.640737533569336, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8933117389678955, + "num_tokens": 471709730.0, + "step": 12152 + }, + { + "epoch": 1.5459865157104695, + "grad_norm": 2.550325632095337, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8951715230941772, + "num_tokens": 471748251.0, + "step": 12153 + }, + { + "epoch": 1.54611372598906, + "grad_norm": 2.6631388664245605, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8861603736877441, + "num_tokens": 471789741.0, + "step": 12154 + }, + { + "epoch": 1.5462409362676506, + "grad_norm": 2.6573214530944824, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8761847615242004, + "num_tokens": 471830396.0, + "step": 12155 + }, + { + "epoch": 1.5463681465462409, + "grad_norm": 2.6305277347564697, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8891603946685791, + "num_tokens": 471870942.0, + "step": 12156 + }, + { + "epoch": 1.5464953568248314, + "grad_norm": 2.436129331588745, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9050132036209106, + "num_tokens": 471908036.0, + "step": 12157 + }, + { + "epoch": 1.546622567103422, + "grad_norm": 2.6838066577911377, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8885695934295654, + "num_tokens": 471947754.0, + "step": 12158 + }, + { + "epoch": 1.5467497773820125, + "grad_norm": 2.525979995727539, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8895375728607178, + "num_tokens": 471987247.0, + "step": 12159 + }, + { + "epoch": 1.546876987660603, + "grad_norm": 2.397078275680542, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8948651552200317, + "num_tokens": 472029202.0, + "step": 12160 + }, + { + "epoch": 1.5470041979391935, + "grad_norm": 2.732952356338501, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8942170143127441, + "num_tokens": 472065721.0, + "step": 12161 + }, + { + "epoch": 1.5471314082177838, + "grad_norm": 2.526369333267212, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.892918586730957, + "num_tokens": 472104163.0, + "step": 12162 + }, + { + "epoch": 1.5472586184963744, + "grad_norm": 2.896421194076538, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8870565295219421, + "num_tokens": 472140979.0, + "step": 12163 + }, + { + "epoch": 1.5473858287749649, + "grad_norm": 2.6697323322296143, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.896933913230896, + "num_tokens": 472176949.0, + "step": 12164 + }, + { + "epoch": 1.5475130390535554, + "grad_norm": 2.971404552459717, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8771553039550781, + "num_tokens": 472213619.0, + "step": 12165 + }, + { + "epoch": 1.547640249332146, + "grad_norm": 2.7919087409973145, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8769921660423279, + "num_tokens": 472253961.0, + "step": 12166 + }, + { + "epoch": 1.5477674596107365, + "grad_norm": 2.592247486114502, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8892539739608765, + "num_tokens": 472295121.0, + "step": 12167 + }, + { + "epoch": 1.547894669889327, + "grad_norm": 2.779634952545166, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9053624868392944, + "num_tokens": 472324933.0, + "step": 12168 + }, + { + "epoch": 1.5480218801679175, + "grad_norm": 2.500925064086914, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9014379978179932, + "num_tokens": 472362561.0, + "step": 12169 + }, + { + "epoch": 1.548149090446508, + "grad_norm": 2.4692776203155518, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.887321949005127, + "num_tokens": 472408767.0, + "step": 12170 + }, + { + "epoch": 1.5482763007250986, + "grad_norm": 2.737823009490967, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8815562725067139, + "num_tokens": 472446888.0, + "step": 12171 + }, + { + "epoch": 1.5484035110036891, + "grad_norm": 2.4482622146606445, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.906072735786438, + "num_tokens": 472487198.0, + "step": 12172 + }, + { + "epoch": 1.5485307212822796, + "grad_norm": 2.4759414196014404, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8944684267044067, + "num_tokens": 472528648.0, + "step": 12173 + }, + { + "epoch": 1.5486579315608702, + "grad_norm": 2.7040650844573975, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8897020816802979, + "num_tokens": 472564412.0, + "step": 12174 + }, + { + "epoch": 1.5487851418394607, + "grad_norm": 2.4681756496429443, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8943973779678345, + "num_tokens": 472605545.0, + "step": 12175 + }, + { + "epoch": 1.5489123521180512, + "grad_norm": 2.731966733932495, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8874874711036682, + "num_tokens": 472641129.0, + "step": 12176 + }, + { + "epoch": 1.5490395623966418, + "grad_norm": 2.637218475341797, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8836138844490051, + "num_tokens": 472683130.0, + "step": 12177 + }, + { + "epoch": 1.5491667726752323, + "grad_norm": 2.724907636642456, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8836996555328369, + "num_tokens": 472722707.0, + "step": 12178 + }, + { + "epoch": 1.5492939829538228, + "grad_norm": 2.6759912967681885, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8850890398025513, + "num_tokens": 472763853.0, + "step": 12179 + }, + { + "epoch": 1.5494211932324133, + "grad_norm": 2.3484437465667725, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8926429748535156, + "num_tokens": 472811646.0, + "step": 12180 + }, + { + "epoch": 1.5495484035110036, + "grad_norm": 2.7086894512176514, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8831357359886169, + "num_tokens": 472851455.0, + "step": 12181 + }, + { + "epoch": 1.5496756137895942, + "grad_norm": 2.562401056289673, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8943083882331848, + "num_tokens": 472890307.0, + "step": 12182 + }, + { + "epoch": 1.5498028240681847, + "grad_norm": 3.0533993244171143, + "learning_rate": 1e-06, + "loss": 0.3502, + "mean_token_accuracy": 0.8725314140319824, + "num_tokens": 472923970.0, + "step": 12183 + }, + { + "epoch": 1.5499300343467752, + "grad_norm": 2.457473039627075, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8870723247528076, + "num_tokens": 472972176.0, + "step": 12184 + }, + { + "epoch": 1.5500572446253658, + "grad_norm": 2.5354390144348145, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8966361284255981, + "num_tokens": 473010374.0, + "step": 12185 + }, + { + "epoch": 1.5501844549039563, + "grad_norm": 2.6002814769744873, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.90376877784729, + "num_tokens": 473045402.0, + "step": 12186 + }, + { + "epoch": 1.5503116651825466, + "grad_norm": 2.9642913341522217, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8844383358955383, + "num_tokens": 473078723.0, + "step": 12187 + }, + { + "epoch": 1.5504388754611371, + "grad_norm": 2.2877180576324463, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8968066573143005, + "num_tokens": 473122590.0, + "step": 12188 + }, + { + "epoch": 1.5505660857397277, + "grad_norm": 2.531266689300537, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8817888498306274, + "num_tokens": 473166875.0, + "step": 12189 + }, + { + "epoch": 1.5506932960183182, + "grad_norm": 2.7979233264923096, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8829537630081177, + "num_tokens": 473206217.0, + "step": 12190 + }, + { + "epoch": 1.5508205062969087, + "grad_norm": 2.5263354778289795, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8861924409866333, + "num_tokens": 473249618.0, + "step": 12191 + }, + { + "epoch": 1.5509477165754992, + "grad_norm": 2.7131075859069824, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.898077130317688, + "num_tokens": 473284089.0, + "step": 12192 + }, + { + "epoch": 1.5510749268540898, + "grad_norm": 2.3066155910491943, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8965182304382324, + "num_tokens": 473330258.0, + "step": 12193 + }, + { + "epoch": 1.5512021371326803, + "grad_norm": 2.5681042671203613, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8924147486686707, + "num_tokens": 473370066.0, + "step": 12194 + }, + { + "epoch": 1.5513293474112708, + "grad_norm": 2.6801321506500244, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8925410509109497, + "num_tokens": 473408896.0, + "step": 12195 + }, + { + "epoch": 1.5514565576898613, + "grad_norm": 2.6095502376556396, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8751408457756042, + "num_tokens": 473453398.0, + "step": 12196 + }, + { + "epoch": 1.5515837679684519, + "grad_norm": 2.6662890911102295, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9044378995895386, + "num_tokens": 473489573.0, + "step": 12197 + }, + { + "epoch": 1.5517109782470424, + "grad_norm": 2.5174031257629395, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.9013773202896118, + "num_tokens": 473530549.0, + "step": 12198 + }, + { + "epoch": 1.551838188525633, + "grad_norm": 2.763380527496338, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.876753568649292, + "num_tokens": 473570021.0, + "step": 12199 + }, + { + "epoch": 1.5519653988042235, + "grad_norm": 2.922574758529663, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8796663284301758, + "num_tokens": 473606754.0, + "step": 12200 + }, + { + "epoch": 1.552092609082814, + "grad_norm": 2.6108736991882324, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8810191750526428, + "num_tokens": 473650350.0, + "step": 12201 + }, + { + "epoch": 1.5522198193614045, + "grad_norm": 2.6059820652008057, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.895850419998169, + "num_tokens": 473686841.0, + "step": 12202 + }, + { + "epoch": 1.552347029639995, + "grad_norm": 2.8786909580230713, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8927969336509705, + "num_tokens": 473716679.0, + "step": 12203 + }, + { + "epoch": 1.5524742399185856, + "grad_norm": 2.550004482269287, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8906368613243103, + "num_tokens": 473755837.0, + "step": 12204 + }, + { + "epoch": 1.5526014501971759, + "grad_norm": 2.7059378623962402, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8894349336624146, + "num_tokens": 473792068.0, + "step": 12205 + }, + { + "epoch": 1.5527286604757664, + "grad_norm": 2.6504547595977783, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8856921195983887, + "num_tokens": 473831749.0, + "step": 12206 + }, + { + "epoch": 1.552855870754357, + "grad_norm": 2.57283091545105, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8881639242172241, + "num_tokens": 473872626.0, + "step": 12207 + }, + { + "epoch": 1.5529830810329475, + "grad_norm": 2.4225125312805176, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9016052484512329, + "num_tokens": 473916597.0, + "step": 12208 + }, + { + "epoch": 1.553110291311538, + "grad_norm": 2.4611406326293945, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8913665413856506, + "num_tokens": 473958404.0, + "step": 12209 + }, + { + "epoch": 1.5532375015901285, + "grad_norm": 2.5943868160247803, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8925179243087769, + "num_tokens": 474001689.0, + "step": 12210 + }, + { + "epoch": 1.5533647118687188, + "grad_norm": 2.9145920276641846, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8877081871032715, + "num_tokens": 474033501.0, + "step": 12211 + }, + { + "epoch": 1.5534919221473094, + "grad_norm": 2.992948293685913, + "learning_rate": 1e-06, + "loss": 0.3621, + "mean_token_accuracy": 0.8735500574111938, + "num_tokens": 474068553.0, + "step": 12212 + }, + { + "epoch": 1.5536191324258999, + "grad_norm": 2.6138174533843994, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8905173540115356, + "num_tokens": 474107267.0, + "step": 12213 + }, + { + "epoch": 1.5537463427044904, + "grad_norm": 2.755594491958618, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8813959956169128, + "num_tokens": 474146166.0, + "step": 12214 + }, + { + "epoch": 1.553873552983081, + "grad_norm": 2.4006168842315674, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9053707122802734, + "num_tokens": 474184446.0, + "step": 12215 + }, + { + "epoch": 1.5540007632616715, + "grad_norm": 2.9184908866882324, + "learning_rate": 1e-06, + "loss": 0.357, + "mean_token_accuracy": 0.8796297311782837, + "num_tokens": 474223309.0, + "step": 12216 + }, + { + "epoch": 1.554127973540262, + "grad_norm": 2.5329785346984863, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8950860500335693, + "num_tokens": 474261788.0, + "step": 12217 + }, + { + "epoch": 1.5542551838188525, + "grad_norm": 2.7773802280426025, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8784118890762329, + "num_tokens": 474298324.0, + "step": 12218 + }, + { + "epoch": 1.554382394097443, + "grad_norm": 2.3720405101776123, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8957267999649048, + "num_tokens": 474343445.0, + "step": 12219 + }, + { + "epoch": 1.5545096043760336, + "grad_norm": 2.5143728256225586, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8966084718704224, + "num_tokens": 474382697.0, + "step": 12220 + }, + { + "epoch": 1.554636814654624, + "grad_norm": 2.4765188694000244, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8896316289901733, + "num_tokens": 474430484.0, + "step": 12221 + }, + { + "epoch": 1.5547640249332146, + "grad_norm": 2.661334276199341, + "learning_rate": 1e-06, + "loss": 0.3461, + "mean_token_accuracy": 0.8779662847518921, + "num_tokens": 474474125.0, + "step": 12222 + }, + { + "epoch": 1.5548912352118052, + "grad_norm": 2.4322421550750732, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8907971978187561, + "num_tokens": 474516447.0, + "step": 12223 + }, + { + "epoch": 1.5550184454903957, + "grad_norm": 2.6836540699005127, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8854125738143921, + "num_tokens": 474555410.0, + "step": 12224 + }, + { + "epoch": 1.5551456557689862, + "grad_norm": 2.8493175506591797, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8910767436027527, + "num_tokens": 474589174.0, + "step": 12225 + }, + { + "epoch": 1.5552728660475768, + "grad_norm": 2.502373456954956, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8962031602859497, + "num_tokens": 474628299.0, + "step": 12226 + }, + { + "epoch": 1.5554000763261673, + "grad_norm": 2.5143420696258545, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8923593163490295, + "num_tokens": 474669372.0, + "step": 12227 + }, + { + "epoch": 1.5555272866047578, + "grad_norm": 2.6295197010040283, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8882575035095215, + "num_tokens": 474708419.0, + "step": 12228 + }, + { + "epoch": 1.5556544968833483, + "grad_norm": 2.487994909286499, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8978492021560669, + "num_tokens": 474747672.0, + "step": 12229 + }, + { + "epoch": 1.5557817071619386, + "grad_norm": 2.6779563426971436, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8993258476257324, + "num_tokens": 474786178.0, + "step": 12230 + }, + { + "epoch": 1.5559089174405292, + "grad_norm": 2.6409387588500977, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9036690592765808, + "num_tokens": 474823495.0, + "step": 12231 + }, + { + "epoch": 1.5560361277191197, + "grad_norm": 2.559187889099121, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8967516422271729, + "num_tokens": 474861009.0, + "step": 12232 + }, + { + "epoch": 1.5561633379977102, + "grad_norm": 3.0783166885375977, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8888065218925476, + "num_tokens": 474892598.0, + "step": 12233 + }, + { + "epoch": 1.5562905482763008, + "grad_norm": 2.9934825897216797, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8959177732467651, + "num_tokens": 474932730.0, + "step": 12234 + }, + { + "epoch": 1.5564177585548913, + "grad_norm": 2.6697542667388916, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8941607475280762, + "num_tokens": 474971330.0, + "step": 12235 + }, + { + "epoch": 1.5565449688334816, + "grad_norm": 2.644857406616211, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8867310285568237, + "num_tokens": 475011678.0, + "step": 12236 + }, + { + "epoch": 1.5566721791120721, + "grad_norm": 2.3758976459503174, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.893162727355957, + "num_tokens": 475052796.0, + "step": 12237 + }, + { + "epoch": 1.5567993893906626, + "grad_norm": 2.5891263484954834, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8986929059028625, + "num_tokens": 475091747.0, + "step": 12238 + }, + { + "epoch": 1.5569265996692532, + "grad_norm": 2.822145938873291, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8801941871643066, + "num_tokens": 475126459.0, + "step": 12239 + }, + { + "epoch": 1.5570538099478437, + "grad_norm": 2.5971531867980957, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8991572856903076, + "num_tokens": 475164815.0, + "step": 12240 + }, + { + "epoch": 1.5571810202264342, + "grad_norm": 2.786583185195923, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.889404296875, + "num_tokens": 475202745.0, + "step": 12241 + }, + { + "epoch": 1.5573082305050248, + "grad_norm": 2.423116683959961, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8873547911643982, + "num_tokens": 475248707.0, + "step": 12242 + }, + { + "epoch": 1.5574354407836153, + "grad_norm": 2.5342087745666504, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8853273987770081, + "num_tokens": 475292679.0, + "step": 12243 + }, + { + "epoch": 1.5575626510622058, + "grad_norm": 3.06461238861084, + "learning_rate": 1e-06, + "loss": 0.347, + "mean_token_accuracy": 0.880655825138092, + "num_tokens": 475328156.0, + "step": 12244 + }, + { + "epoch": 1.5576898613407963, + "grad_norm": 2.4689571857452393, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8987023830413818, + "num_tokens": 475366991.0, + "step": 12245 + }, + { + "epoch": 1.5578170716193869, + "grad_norm": 2.354377508163452, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9079007506370544, + "num_tokens": 475411818.0, + "step": 12246 + }, + { + "epoch": 1.5579442818979774, + "grad_norm": 2.555668830871582, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8910671472549438, + "num_tokens": 475452991.0, + "step": 12247 + }, + { + "epoch": 1.558071492176568, + "grad_norm": 2.468348979949951, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8946771621704102, + "num_tokens": 475494984.0, + "step": 12248 + }, + { + "epoch": 1.5581987024551585, + "grad_norm": 2.5154314041137695, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8927854895591736, + "num_tokens": 475532937.0, + "step": 12249 + }, + { + "epoch": 1.558325912733749, + "grad_norm": 2.701791763305664, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8878483772277832, + "num_tokens": 475570014.0, + "step": 12250 + }, + { + "epoch": 1.5584531230123395, + "grad_norm": 2.652883768081665, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8896989822387695, + "num_tokens": 475608945.0, + "step": 12251 + }, + { + "epoch": 1.55858033329093, + "grad_norm": 2.55975604057312, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8843162059783936, + "num_tokens": 475652956.0, + "step": 12252 + }, + { + "epoch": 1.5587075435695206, + "grad_norm": 2.511538028717041, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8930217027664185, + "num_tokens": 475696678.0, + "step": 12253 + }, + { + "epoch": 1.5588347538481109, + "grad_norm": 2.595181465148926, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8907788991928101, + "num_tokens": 475735440.0, + "step": 12254 + }, + { + "epoch": 1.5589619641267014, + "grad_norm": 2.617161512374878, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8990177512168884, + "num_tokens": 475771703.0, + "step": 12255 + }, + { + "epoch": 1.559089174405292, + "grad_norm": 2.7382171154022217, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8836410045623779, + "num_tokens": 475808565.0, + "step": 12256 + }, + { + "epoch": 1.5592163846838825, + "grad_norm": 2.7830514907836914, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9024745225906372, + "num_tokens": 475838593.0, + "step": 12257 + }, + { + "epoch": 1.559343594962473, + "grad_norm": 2.6065526008605957, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8911005258560181, + "num_tokens": 475878076.0, + "step": 12258 + }, + { + "epoch": 1.5594708052410635, + "grad_norm": 2.8500943183898926, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8865013718605042, + "num_tokens": 475915714.0, + "step": 12259 + }, + { + "epoch": 1.5595980155196538, + "grad_norm": 2.652637243270874, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8901174664497375, + "num_tokens": 475956596.0, + "step": 12260 + }, + { + "epoch": 1.5597252257982444, + "grad_norm": 2.6639413833618164, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8942194581031799, + "num_tokens": 475994528.0, + "step": 12261 + }, + { + "epoch": 1.5598524360768349, + "grad_norm": 2.8112831115722656, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8741933107376099, + "num_tokens": 476033847.0, + "step": 12262 + }, + { + "epoch": 1.5599796463554254, + "grad_norm": 2.5299432277679443, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.895339846611023, + "num_tokens": 476075297.0, + "step": 12263 + }, + { + "epoch": 1.560106856634016, + "grad_norm": 2.696808338165283, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8957653641700745, + "num_tokens": 476109839.0, + "step": 12264 + }, + { + "epoch": 1.5602340669126065, + "grad_norm": 2.6051125526428223, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8901330828666687, + "num_tokens": 476149606.0, + "step": 12265 + }, + { + "epoch": 1.560361277191197, + "grad_norm": 2.8781802654266357, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8993735313415527, + "num_tokens": 476179699.0, + "step": 12266 + }, + { + "epoch": 1.5604884874697875, + "grad_norm": 2.502098798751831, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8849384784698486, + "num_tokens": 476227995.0, + "step": 12267 + }, + { + "epoch": 1.560615697748378, + "grad_norm": 2.665865898132324, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8880020380020142, + "num_tokens": 476263982.0, + "step": 12268 + }, + { + "epoch": 1.5607429080269686, + "grad_norm": 2.844797372817993, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8885382413864136, + "num_tokens": 476297540.0, + "step": 12269 + }, + { + "epoch": 1.560870118305559, + "grad_norm": 2.7750821113586426, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8954078555107117, + "num_tokens": 476334408.0, + "step": 12270 + }, + { + "epoch": 1.5609973285841496, + "grad_norm": 2.680314064025879, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8950154781341553, + "num_tokens": 476372665.0, + "step": 12271 + }, + { + "epoch": 1.5611245388627402, + "grad_norm": 2.842217206954956, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8808953762054443, + "num_tokens": 476413400.0, + "step": 12272 + }, + { + "epoch": 1.5612517491413307, + "grad_norm": 2.6184258460998535, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9022480249404907, + "num_tokens": 476447547.0, + "step": 12273 + }, + { + "epoch": 1.5613789594199212, + "grad_norm": 2.7237281799316406, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8926283121109009, + "num_tokens": 476481800.0, + "step": 12274 + }, + { + "epoch": 1.5615061696985117, + "grad_norm": 2.586263418197632, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8934928178787231, + "num_tokens": 476524034.0, + "step": 12275 + }, + { + "epoch": 1.5616333799771023, + "grad_norm": 2.854653835296631, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8812711238861084, + "num_tokens": 476557827.0, + "step": 12276 + }, + { + "epoch": 1.5617605902556928, + "grad_norm": 2.560743808746338, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8909550905227661, + "num_tokens": 476597512.0, + "step": 12277 + }, + { + "epoch": 1.5618878005342833, + "grad_norm": 2.7448768615722656, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.8801094889640808, + "num_tokens": 476634694.0, + "step": 12278 + }, + { + "epoch": 1.5620150108128736, + "grad_norm": 3.091667890548706, + "learning_rate": 1e-06, + "loss": 0.3749, + "mean_token_accuracy": 0.8669224977493286, + "num_tokens": 476668225.0, + "step": 12279 + }, + { + "epoch": 1.5621422210914642, + "grad_norm": 2.2653188705444336, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8978571891784668, + "num_tokens": 476714403.0, + "step": 12280 + }, + { + "epoch": 1.5622694313700547, + "grad_norm": 2.8023760318756104, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8892375230789185, + "num_tokens": 476747223.0, + "step": 12281 + }, + { + "epoch": 1.5623966416486452, + "grad_norm": 2.587301254272461, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.894305944442749, + "num_tokens": 476788326.0, + "step": 12282 + }, + { + "epoch": 1.5625238519272358, + "grad_norm": 2.4460294246673584, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8916523456573486, + "num_tokens": 476829578.0, + "step": 12283 + }, + { + "epoch": 1.5626510622058263, + "grad_norm": 2.8431131839752197, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8831645846366882, + "num_tokens": 476867706.0, + "step": 12284 + }, + { + "epoch": 1.5627782724844166, + "grad_norm": 2.738727331161499, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8843671679496765, + "num_tokens": 476904965.0, + "step": 12285 + }, + { + "epoch": 1.5629054827630071, + "grad_norm": 2.6471774578094482, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8803476691246033, + "num_tokens": 476949155.0, + "step": 12286 + }, + { + "epoch": 1.5630326930415976, + "grad_norm": 2.6822307109832764, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8906841278076172, + "num_tokens": 476989539.0, + "step": 12287 + }, + { + "epoch": 1.5631599033201882, + "grad_norm": 2.6548588275909424, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8849290013313293, + "num_tokens": 477030817.0, + "step": 12288 + }, + { + "epoch": 1.5632871135987787, + "grad_norm": 2.646446704864502, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8845653533935547, + "num_tokens": 477070903.0, + "step": 12289 + }, + { + "epoch": 1.5634143238773692, + "grad_norm": 2.4075498580932617, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8876369595527649, + "num_tokens": 477115208.0, + "step": 12290 + }, + { + "epoch": 1.5635415341559598, + "grad_norm": 2.5773520469665527, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.8975017070770264, + "num_tokens": 477151489.0, + "step": 12291 + }, + { + "epoch": 1.5636687444345503, + "grad_norm": 2.4336812496185303, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9029819369316101, + "num_tokens": 477191244.0, + "step": 12292 + }, + { + "epoch": 1.5637959547131408, + "grad_norm": 2.8218212127685547, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.891599714756012, + "num_tokens": 477226325.0, + "step": 12293 + }, + { + "epoch": 1.5639231649917313, + "grad_norm": 2.844581127166748, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8809961676597595, + "num_tokens": 477263061.0, + "step": 12294 + }, + { + "epoch": 1.5640503752703219, + "grad_norm": 2.6412429809570312, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8984090089797974, + "num_tokens": 477297617.0, + "step": 12295 + }, + { + "epoch": 1.5641775855489124, + "grad_norm": 2.514045000076294, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.897013783454895, + "num_tokens": 477338166.0, + "step": 12296 + }, + { + "epoch": 1.564304795827503, + "grad_norm": 2.718092441558838, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9003332853317261, + "num_tokens": 477373785.0, + "step": 12297 + }, + { + "epoch": 1.5644320061060935, + "grad_norm": 2.835752248764038, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8808627128601074, + "num_tokens": 477410418.0, + "step": 12298 + }, + { + "epoch": 1.564559216384684, + "grad_norm": 2.559382438659668, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8935979008674622, + "num_tokens": 477449793.0, + "step": 12299 + }, + { + "epoch": 1.5646864266632745, + "grad_norm": 2.6029977798461914, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8890730142593384, + "num_tokens": 477492533.0, + "step": 12300 + }, + { + "epoch": 1.564813636941865, + "grad_norm": 2.8778491020202637, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8864700794219971, + "num_tokens": 477528739.0, + "step": 12301 + }, + { + "epoch": 1.5649408472204556, + "grad_norm": 2.6713953018188477, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8826048374176025, + "num_tokens": 477570173.0, + "step": 12302 + }, + { + "epoch": 1.5650680574990459, + "grad_norm": 2.962578296661377, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8869767189025879, + "num_tokens": 477604503.0, + "step": 12303 + }, + { + "epoch": 1.5651952677776364, + "grad_norm": 2.6264185905456543, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8941993713378906, + "num_tokens": 477640344.0, + "step": 12304 + }, + { + "epoch": 1.565322478056227, + "grad_norm": 2.588289260864258, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8922816514968872, + "num_tokens": 477677300.0, + "step": 12305 + }, + { + "epoch": 1.5654496883348175, + "grad_norm": 2.612976551055908, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8803844451904297, + "num_tokens": 477715887.0, + "step": 12306 + }, + { + "epoch": 1.565576898613408, + "grad_norm": 2.280146598815918, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9017572402954102, + "num_tokens": 477758892.0, + "step": 12307 + }, + { + "epoch": 1.5657041088919985, + "grad_norm": 2.5449471473693848, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8970255851745605, + "num_tokens": 477796929.0, + "step": 12308 + }, + { + "epoch": 1.5658313191705888, + "grad_norm": 2.6451447010040283, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8902692794799805, + "num_tokens": 477835763.0, + "step": 12309 + }, + { + "epoch": 1.5659585294491793, + "grad_norm": 2.618961811065674, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9027235507965088, + "num_tokens": 477867683.0, + "step": 12310 + }, + { + "epoch": 1.5660857397277699, + "grad_norm": 2.6915550231933594, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8948391675949097, + "num_tokens": 477900971.0, + "step": 12311 + }, + { + "epoch": 1.5662129500063604, + "grad_norm": 2.549765110015869, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8893585205078125, + "num_tokens": 477941341.0, + "step": 12312 + }, + { + "epoch": 1.566340160284951, + "grad_norm": 2.718017816543579, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8899862766265869, + "num_tokens": 477980554.0, + "step": 12313 + }, + { + "epoch": 1.5664673705635415, + "grad_norm": 2.8261959552764893, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8965056538581848, + "num_tokens": 478015732.0, + "step": 12314 + }, + { + "epoch": 1.566594580842132, + "grad_norm": 2.6611344814300537, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8928165435791016, + "num_tokens": 478055771.0, + "step": 12315 + }, + { + "epoch": 1.5667217911207225, + "grad_norm": 2.406118869781494, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8965084552764893, + "num_tokens": 478097335.0, + "step": 12316 + }, + { + "epoch": 1.566849001399313, + "grad_norm": 2.6834867000579834, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8914444446563721, + "num_tokens": 478134170.0, + "step": 12317 + }, + { + "epoch": 1.5669762116779036, + "grad_norm": 2.5512328147888184, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8822049498558044, + "num_tokens": 478180136.0, + "step": 12318 + }, + { + "epoch": 1.567103421956494, + "grad_norm": 2.9032390117645264, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8838200569152832, + "num_tokens": 478214194.0, + "step": 12319 + }, + { + "epoch": 1.5672306322350846, + "grad_norm": 2.89321231842041, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8881040811538696, + "num_tokens": 478248609.0, + "step": 12320 + }, + { + "epoch": 1.5673578425136752, + "grad_norm": 2.7057559490203857, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8858464360237122, + "num_tokens": 478287059.0, + "step": 12321 + }, + { + "epoch": 1.5674850527922657, + "grad_norm": 2.324134588241577, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8979887962341309, + "num_tokens": 478332424.0, + "step": 12322 + }, + { + "epoch": 1.5676122630708562, + "grad_norm": 2.5222649574279785, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8914764523506165, + "num_tokens": 478370707.0, + "step": 12323 + }, + { + "epoch": 1.5677394733494467, + "grad_norm": 2.5671000480651855, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8910220861434937, + "num_tokens": 478414203.0, + "step": 12324 + }, + { + "epoch": 1.5678666836280373, + "grad_norm": 2.7759335041046143, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8861008882522583, + "num_tokens": 478455298.0, + "step": 12325 + }, + { + "epoch": 1.5679938939066278, + "grad_norm": 2.523322582244873, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8948260545730591, + "num_tokens": 478494321.0, + "step": 12326 + }, + { + "epoch": 1.5681211041852183, + "grad_norm": 2.505779981613159, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8997099995613098, + "num_tokens": 478534940.0, + "step": 12327 + }, + { + "epoch": 1.5682483144638086, + "grad_norm": 2.8418588638305664, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.890294075012207, + "num_tokens": 478571830.0, + "step": 12328 + }, + { + "epoch": 1.5683755247423992, + "grad_norm": 2.5671873092651367, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8993191719055176, + "num_tokens": 478606877.0, + "step": 12329 + }, + { + "epoch": 1.5685027350209897, + "grad_norm": 2.3376898765563965, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.892825722694397, + "num_tokens": 478652939.0, + "step": 12330 + }, + { + "epoch": 1.5686299452995802, + "grad_norm": 2.7648441791534424, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8910239934921265, + "num_tokens": 478689243.0, + "step": 12331 + }, + { + "epoch": 1.5687571555781707, + "grad_norm": 2.6866977214813232, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8944714069366455, + "num_tokens": 478724014.0, + "step": 12332 + }, + { + "epoch": 1.5688843658567613, + "grad_norm": 2.8970768451690674, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8866164684295654, + "num_tokens": 478760448.0, + "step": 12333 + }, + { + "epoch": 1.5690115761353516, + "grad_norm": 2.9590506553649902, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8873019814491272, + "num_tokens": 478793596.0, + "step": 12334 + }, + { + "epoch": 1.569138786413942, + "grad_norm": 2.5839710235595703, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9001268148422241, + "num_tokens": 478830668.0, + "step": 12335 + }, + { + "epoch": 1.5692659966925326, + "grad_norm": 2.6347899436950684, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8918812274932861, + "num_tokens": 478867596.0, + "step": 12336 + }, + { + "epoch": 1.5693932069711232, + "grad_norm": 2.684173107147217, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9010522365570068, + "num_tokens": 478899852.0, + "step": 12337 + }, + { + "epoch": 1.5695204172497137, + "grad_norm": 2.4126927852630615, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.8993605375289917, + "num_tokens": 478940667.0, + "step": 12338 + }, + { + "epoch": 1.5696476275283042, + "grad_norm": 2.5164380073547363, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8926998376846313, + "num_tokens": 478980327.0, + "step": 12339 + }, + { + "epoch": 1.5697748378068948, + "grad_norm": 2.7229180335998535, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.886326789855957, + "num_tokens": 479020857.0, + "step": 12340 + }, + { + "epoch": 1.5699020480854853, + "grad_norm": 2.550602436065674, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8902641534805298, + "num_tokens": 479057690.0, + "step": 12341 + }, + { + "epoch": 1.5700292583640758, + "grad_norm": 2.6841893196105957, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8955556154251099, + "num_tokens": 479093798.0, + "step": 12342 + }, + { + "epoch": 1.5701564686426663, + "grad_norm": 2.6765689849853516, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8948122262954712, + "num_tokens": 479131589.0, + "step": 12343 + }, + { + "epoch": 1.5702836789212569, + "grad_norm": 2.6685385704040527, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9009882807731628, + "num_tokens": 479171387.0, + "step": 12344 + }, + { + "epoch": 1.5704108891998474, + "grad_norm": 2.4408204555511475, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8929213285446167, + "num_tokens": 479217684.0, + "step": 12345 + }, + { + "epoch": 1.570538099478438, + "grad_norm": 2.947974920272827, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8782550096511841, + "num_tokens": 479254661.0, + "step": 12346 + }, + { + "epoch": 1.5706653097570284, + "grad_norm": 2.9658873081207275, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8849170207977295, + "num_tokens": 479289369.0, + "step": 12347 + }, + { + "epoch": 1.570792520035619, + "grad_norm": 2.7213048934936523, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8952388167381287, + "num_tokens": 479326852.0, + "step": 12348 + }, + { + "epoch": 1.5709197303142095, + "grad_norm": 2.5756356716156006, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8914297819137573, + "num_tokens": 479367340.0, + "step": 12349 + }, + { + "epoch": 1.5710469405928, + "grad_norm": 2.5328688621520996, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8966706395149231, + "num_tokens": 479406492.0, + "step": 12350 + }, + { + "epoch": 1.5711741508713906, + "grad_norm": 2.7797188758850098, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8928632736206055, + "num_tokens": 479441292.0, + "step": 12351 + }, + { + "epoch": 1.5713013611499809, + "grad_norm": 2.3083949089050293, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8986870050430298, + "num_tokens": 479485734.0, + "step": 12352 + }, + { + "epoch": 1.5714285714285714, + "grad_norm": 2.5667366981506348, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8845449686050415, + "num_tokens": 479528064.0, + "step": 12353 + }, + { + "epoch": 1.571555781707162, + "grad_norm": 2.7709689140319824, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8931401371955872, + "num_tokens": 479564372.0, + "step": 12354 + }, + { + "epoch": 1.5716829919857525, + "grad_norm": 2.6572980880737305, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.888220489025116, + "num_tokens": 479606166.0, + "step": 12355 + }, + { + "epoch": 1.571810202264343, + "grad_norm": 2.875375986099243, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8838330507278442, + "num_tokens": 479640270.0, + "step": 12356 + }, + { + "epoch": 1.5719374125429335, + "grad_norm": 2.6603002548217773, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.8736458420753479, + "num_tokens": 479684638.0, + "step": 12357 + }, + { + "epoch": 1.5720646228215238, + "grad_norm": 2.507457971572876, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8905035257339478, + "num_tokens": 479729655.0, + "step": 12358 + }, + { + "epoch": 1.5721918331001143, + "grad_norm": 2.7257766723632812, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.881616473197937, + "num_tokens": 479766602.0, + "step": 12359 + }, + { + "epoch": 1.5723190433787049, + "grad_norm": 2.555417060852051, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8942021131515503, + "num_tokens": 479807190.0, + "step": 12360 + }, + { + "epoch": 1.5724462536572954, + "grad_norm": 2.618532180786133, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.885356068611145, + "num_tokens": 479847327.0, + "step": 12361 + }, + { + "epoch": 1.572573463935886, + "grad_norm": 2.690368890762329, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.884285032749176, + "num_tokens": 479888494.0, + "step": 12362 + }, + { + "epoch": 1.5727006742144765, + "grad_norm": 2.6042871475219727, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8974803686141968, + "num_tokens": 479927257.0, + "step": 12363 + }, + { + "epoch": 1.572827884493067, + "grad_norm": 2.3426077365875244, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8929497599601746, + "num_tokens": 479971992.0, + "step": 12364 + }, + { + "epoch": 1.5729550947716575, + "grad_norm": 2.68231201171875, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9049865007400513, + "num_tokens": 480007011.0, + "step": 12365 + }, + { + "epoch": 1.573082305050248, + "grad_norm": 2.6582982540130615, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8834066390991211, + "num_tokens": 480048019.0, + "step": 12366 + }, + { + "epoch": 1.5732095153288386, + "grad_norm": 2.397905111312866, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8804802298545837, + "num_tokens": 480093673.0, + "step": 12367 + }, + { + "epoch": 1.573336725607429, + "grad_norm": 2.6040384769439697, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8821027874946594, + "num_tokens": 480137494.0, + "step": 12368 + }, + { + "epoch": 1.5734639358860196, + "grad_norm": 2.7139956951141357, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8812320232391357, + "num_tokens": 480179157.0, + "step": 12369 + }, + { + "epoch": 1.5735911461646102, + "grad_norm": 2.5425093173980713, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8933636546134949, + "num_tokens": 480218978.0, + "step": 12370 + }, + { + "epoch": 1.5737183564432007, + "grad_norm": 2.431536912918091, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8980255722999573, + "num_tokens": 480256896.0, + "step": 12371 + }, + { + "epoch": 1.5738455667217912, + "grad_norm": 2.5427424907684326, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8885535001754761, + "num_tokens": 480298267.0, + "step": 12372 + }, + { + "epoch": 1.5739727770003817, + "grad_norm": 2.413266181945801, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.905520498752594, + "num_tokens": 480335023.0, + "step": 12373 + }, + { + "epoch": 1.5740999872789723, + "grad_norm": 2.671851873397827, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.881056547164917, + "num_tokens": 480373946.0, + "step": 12374 + }, + { + "epoch": 1.5742271975575628, + "grad_norm": 2.4178106784820557, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9033117294311523, + "num_tokens": 480412180.0, + "step": 12375 + }, + { + "epoch": 1.5743544078361533, + "grad_norm": 2.844299554824829, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8962177038192749, + "num_tokens": 480445767.0, + "step": 12376 + }, + { + "epoch": 1.5744816181147436, + "grad_norm": 2.7533938884735107, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8840286135673523, + "num_tokens": 480483634.0, + "step": 12377 + }, + { + "epoch": 1.5746088283933342, + "grad_norm": 2.836804151535034, + "learning_rate": 1e-06, + "loss": 0.3501, + "mean_token_accuracy": 0.8798969388008118, + "num_tokens": 480522794.0, + "step": 12378 + }, + { + "epoch": 1.5747360386719247, + "grad_norm": 2.6270575523376465, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8958028554916382, + "num_tokens": 480564737.0, + "step": 12379 + }, + { + "epoch": 1.5748632489505152, + "grad_norm": 2.4919378757476807, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8962369561195374, + "num_tokens": 480607086.0, + "step": 12380 + }, + { + "epoch": 1.5749904592291057, + "grad_norm": 2.6144182682037354, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8926367163658142, + "num_tokens": 480646314.0, + "step": 12381 + }, + { + "epoch": 1.5751176695076963, + "grad_norm": 2.693574905395508, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8896641731262207, + "num_tokens": 480685457.0, + "step": 12382 + }, + { + "epoch": 1.5752448797862866, + "grad_norm": 2.3944592475891113, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8870076537132263, + "num_tokens": 480730705.0, + "step": 12383 + }, + { + "epoch": 1.575372090064877, + "grad_norm": 2.5926151275634766, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8980361819267273, + "num_tokens": 480772005.0, + "step": 12384 + }, + { + "epoch": 1.5754993003434676, + "grad_norm": 2.559072971343994, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8886296153068542, + "num_tokens": 480812398.0, + "step": 12385 + }, + { + "epoch": 1.5756265106220582, + "grad_norm": 2.5486984252929688, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9011130332946777, + "num_tokens": 480851802.0, + "step": 12386 + }, + { + "epoch": 1.5757537209006487, + "grad_norm": 2.6706600189208984, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.88798987865448, + "num_tokens": 480890421.0, + "step": 12387 + }, + { + "epoch": 1.5758809311792392, + "grad_norm": 2.8381030559539795, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8839988708496094, + "num_tokens": 480925411.0, + "step": 12388 + }, + { + "epoch": 1.5760081414578297, + "grad_norm": 2.5354301929473877, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8913878202438354, + "num_tokens": 480964769.0, + "step": 12389 + }, + { + "epoch": 1.5761353517364203, + "grad_norm": 2.552398204803467, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8960123658180237, + "num_tokens": 481004893.0, + "step": 12390 + }, + { + "epoch": 1.5762625620150108, + "grad_norm": 2.5795700550079346, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8963042497634888, + "num_tokens": 481044802.0, + "step": 12391 + }, + { + "epoch": 1.5763897722936013, + "grad_norm": 2.5660431385040283, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8982765674591064, + "num_tokens": 481081063.0, + "step": 12392 + }, + { + "epoch": 1.5765169825721919, + "grad_norm": 2.5555014610290527, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.896669864654541, + "num_tokens": 481122611.0, + "step": 12393 + }, + { + "epoch": 1.5766441928507824, + "grad_norm": 2.815755844116211, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8976616859436035, + "num_tokens": 481155108.0, + "step": 12394 + }, + { + "epoch": 1.576771403129373, + "grad_norm": 2.3551106452941895, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8973579406738281, + "num_tokens": 481198952.0, + "step": 12395 + }, + { + "epoch": 1.5768986134079634, + "grad_norm": 2.883129835128784, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8877487778663635, + "num_tokens": 481232167.0, + "step": 12396 + }, + { + "epoch": 1.577025823686554, + "grad_norm": 2.6184375286102295, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8805155754089355, + "num_tokens": 481272712.0, + "step": 12397 + }, + { + "epoch": 1.5771530339651445, + "grad_norm": 2.7308239936828613, + "learning_rate": 1e-06, + "loss": 0.3615, + "mean_token_accuracy": 0.8755193948745728, + "num_tokens": 481315709.0, + "step": 12398 + }, + { + "epoch": 1.577280244243735, + "grad_norm": 2.571068286895752, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8940129280090332, + "num_tokens": 481352760.0, + "step": 12399 + }, + { + "epoch": 1.5774074545223256, + "grad_norm": 2.802189350128174, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8752825260162354, + "num_tokens": 481389844.0, + "step": 12400 + }, + { + "epoch": 1.5775346648009159, + "grad_norm": 2.5691609382629395, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8937468528747559, + "num_tokens": 481428465.0, + "step": 12401 + }, + { + "epoch": 1.5776618750795064, + "grad_norm": 2.8737778663635254, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8813773393630981, + "num_tokens": 481462474.0, + "step": 12402 + }, + { + "epoch": 1.577789085358097, + "grad_norm": 2.8129303455352783, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8800292015075684, + "num_tokens": 481499977.0, + "step": 12403 + }, + { + "epoch": 1.5779162956366874, + "grad_norm": 2.878563404083252, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8842194080352783, + "num_tokens": 481534661.0, + "step": 12404 + }, + { + "epoch": 1.578043505915278, + "grad_norm": 2.9707868099212646, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8763141632080078, + "num_tokens": 481567624.0, + "step": 12405 + }, + { + "epoch": 1.5781707161938685, + "grad_norm": 2.6703994274139404, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8818190097808838, + "num_tokens": 481610911.0, + "step": 12406 + }, + { + "epoch": 1.5782979264724588, + "grad_norm": 2.83916974067688, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8828161954879761, + "num_tokens": 481646150.0, + "step": 12407 + }, + { + "epoch": 1.5784251367510493, + "grad_norm": 2.6090688705444336, + "learning_rate": 1e-06, + "loss": 0.3269, + "mean_token_accuracy": 0.8856187462806702, + "num_tokens": 481688892.0, + "step": 12408 + }, + { + "epoch": 1.5785523470296399, + "grad_norm": 2.643258571624756, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.890321671962738, + "num_tokens": 481728538.0, + "step": 12409 + }, + { + "epoch": 1.5786795573082304, + "grad_norm": 2.490617513656616, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8793610334396362, + "num_tokens": 481773797.0, + "step": 12410 + }, + { + "epoch": 1.578806767586821, + "grad_norm": 2.465451240539551, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8865712881088257, + "num_tokens": 481819938.0, + "step": 12411 + }, + { + "epoch": 1.5789339778654115, + "grad_norm": 2.6471798419952393, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.893824577331543, + "num_tokens": 481860182.0, + "step": 12412 + }, + { + "epoch": 1.579061188144002, + "grad_norm": 2.3982138633728027, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8877489566802979, + "num_tokens": 481902515.0, + "step": 12413 + }, + { + "epoch": 1.5791883984225925, + "grad_norm": 2.605950355529785, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8865443468093872, + "num_tokens": 481943003.0, + "step": 12414 + }, + { + "epoch": 1.579315608701183, + "grad_norm": 2.664912462234497, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8881556987762451, + "num_tokens": 481982909.0, + "step": 12415 + }, + { + "epoch": 1.5794428189797736, + "grad_norm": 2.8017892837524414, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8843525648117065, + "num_tokens": 482019102.0, + "step": 12416 + }, + { + "epoch": 1.579570029258364, + "grad_norm": 2.723897933959961, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.885570228099823, + "num_tokens": 482057101.0, + "step": 12417 + }, + { + "epoch": 1.5796972395369546, + "grad_norm": 2.752274990081787, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8891431093215942, + "num_tokens": 482091898.0, + "step": 12418 + }, + { + "epoch": 1.5798244498155452, + "grad_norm": 2.5732975006103516, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8803625106811523, + "num_tokens": 482135744.0, + "step": 12419 + }, + { + "epoch": 1.5799516600941357, + "grad_norm": 2.4438462257385254, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.893764078617096, + "num_tokens": 482176842.0, + "step": 12420 + }, + { + "epoch": 1.5800788703727262, + "grad_norm": 2.5994749069213867, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.892838180065155, + "num_tokens": 482213119.0, + "step": 12421 + }, + { + "epoch": 1.5802060806513167, + "grad_norm": 2.469935655593872, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9041874408721924, + "num_tokens": 482250079.0, + "step": 12422 + }, + { + "epoch": 1.5803332909299073, + "grad_norm": 2.727302312850952, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8908213376998901, + "num_tokens": 482286177.0, + "step": 12423 + }, + { + "epoch": 1.5804605012084978, + "grad_norm": 2.5758447647094727, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.883379340171814, + "num_tokens": 482329743.0, + "step": 12424 + }, + { + "epoch": 1.5805877114870883, + "grad_norm": 2.6050469875335693, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8931485414505005, + "num_tokens": 482363939.0, + "step": 12425 + }, + { + "epoch": 1.5807149217656786, + "grad_norm": 2.9300899505615234, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8948512077331543, + "num_tokens": 482394250.0, + "step": 12426 + }, + { + "epoch": 1.5808421320442692, + "grad_norm": 2.6993672847747803, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8834801912307739, + "num_tokens": 482431367.0, + "step": 12427 + }, + { + "epoch": 1.5809693423228597, + "grad_norm": 2.367671489715576, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8981164693832397, + "num_tokens": 482476335.0, + "step": 12428 + }, + { + "epoch": 1.5810965526014502, + "grad_norm": 2.5892789363861084, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8907979726791382, + "num_tokens": 482514242.0, + "step": 12429 + }, + { + "epoch": 1.5812237628800407, + "grad_norm": 2.7887821197509766, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8761962652206421, + "num_tokens": 482551624.0, + "step": 12430 + }, + { + "epoch": 1.5813509731586313, + "grad_norm": 2.846724271774292, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8890393972396851, + "num_tokens": 482583140.0, + "step": 12431 + }, + { + "epoch": 1.5814781834372216, + "grad_norm": 2.53098726272583, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8989851474761963, + "num_tokens": 482622681.0, + "step": 12432 + }, + { + "epoch": 1.581605393715812, + "grad_norm": 2.660841941833496, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8893479108810425, + "num_tokens": 482663977.0, + "step": 12433 + }, + { + "epoch": 1.5817326039944026, + "grad_norm": 2.6694743633270264, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8885039687156677, + "num_tokens": 482701278.0, + "step": 12434 + }, + { + "epoch": 1.5818598142729932, + "grad_norm": 2.50146484375, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9019273519515991, + "num_tokens": 482739184.0, + "step": 12435 + }, + { + "epoch": 1.5819870245515837, + "grad_norm": 2.3842244148254395, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8847838640213013, + "num_tokens": 482787447.0, + "step": 12436 + }, + { + "epoch": 1.5821142348301742, + "grad_norm": 2.661642551422119, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8947193622589111, + "num_tokens": 482825074.0, + "step": 12437 + }, + { + "epoch": 1.5822414451087647, + "grad_norm": 2.5900373458862305, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.8994368314743042, + "num_tokens": 482860690.0, + "step": 12438 + }, + { + "epoch": 1.5823686553873553, + "grad_norm": 2.6992204189300537, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.8794597387313843, + "num_tokens": 482904808.0, + "step": 12439 + }, + { + "epoch": 1.5824958656659458, + "grad_norm": 2.894157886505127, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8911728858947754, + "num_tokens": 482936318.0, + "step": 12440 + }, + { + "epoch": 1.5826230759445363, + "grad_norm": 2.816624402999878, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8948684930801392, + "num_tokens": 482968710.0, + "step": 12441 + }, + { + "epoch": 1.5827502862231269, + "grad_norm": 2.5586254596710205, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8939856886863708, + "num_tokens": 483008714.0, + "step": 12442 + }, + { + "epoch": 1.5828774965017174, + "grad_norm": 2.6127123832702637, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8857505321502686, + "num_tokens": 483048216.0, + "step": 12443 + }, + { + "epoch": 1.583004706780308, + "grad_norm": 2.9277656078338623, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9012326598167419, + "num_tokens": 483080177.0, + "step": 12444 + }, + { + "epoch": 1.5831319170588984, + "grad_norm": 2.5014572143554688, + "learning_rate": 1e-06, + "loss": 0.3382, + "mean_token_accuracy": 0.8802777528762817, + "num_tokens": 483129128.0, + "step": 12445 + }, + { + "epoch": 1.583259127337489, + "grad_norm": 2.675278425216675, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.8930663466453552, + "num_tokens": 483164393.0, + "step": 12446 + }, + { + "epoch": 1.5833863376160795, + "grad_norm": 2.595060348510742, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8855959177017212, + "num_tokens": 483205299.0, + "step": 12447 + }, + { + "epoch": 1.58351354789467, + "grad_norm": 2.5836081504821777, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8916493654251099, + "num_tokens": 483243187.0, + "step": 12448 + }, + { + "epoch": 1.5836407581732606, + "grad_norm": 2.594578981399536, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.89130038022995, + "num_tokens": 483283931.0, + "step": 12449 + }, + { + "epoch": 1.5837679684518509, + "grad_norm": 2.4089624881744385, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.8939777612686157, + "num_tokens": 483326691.0, + "step": 12450 + }, + { + "epoch": 1.5838951787304414, + "grad_norm": 2.4826955795288086, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8947941064834595, + "num_tokens": 483369538.0, + "step": 12451 + }, + { + "epoch": 1.584022389009032, + "grad_norm": 2.5899360179901123, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8901174068450928, + "num_tokens": 483408484.0, + "step": 12452 + }, + { + "epoch": 1.5841495992876224, + "grad_norm": 2.631148338317871, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9029821157455444, + "num_tokens": 483441583.0, + "step": 12453 + }, + { + "epoch": 1.584276809566213, + "grad_norm": 2.4349613189697266, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9049673080444336, + "num_tokens": 483479654.0, + "step": 12454 + }, + { + "epoch": 1.5844040198448035, + "grad_norm": 2.7064590454101562, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8978356122970581, + "num_tokens": 483516050.0, + "step": 12455 + }, + { + "epoch": 1.5845312301233938, + "grad_norm": 2.9491024017333984, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8932633399963379, + "num_tokens": 483547133.0, + "step": 12456 + }, + { + "epoch": 1.5846584404019843, + "grad_norm": 2.731095552444458, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8838471174240112, + "num_tokens": 483584427.0, + "step": 12457 + }, + { + "epoch": 1.5847856506805749, + "grad_norm": 2.779723644256592, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8784708976745605, + "num_tokens": 483624033.0, + "step": 12458 + }, + { + "epoch": 1.5849128609591654, + "grad_norm": 2.5887413024902344, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.899965763092041, + "num_tokens": 483658024.0, + "step": 12459 + }, + { + "epoch": 1.585040071237756, + "grad_norm": 2.5356805324554443, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8942985534667969, + "num_tokens": 483696686.0, + "step": 12460 + }, + { + "epoch": 1.5851672815163464, + "grad_norm": 2.7329368591308594, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8965187072753906, + "num_tokens": 483733695.0, + "step": 12461 + }, + { + "epoch": 1.585294491794937, + "grad_norm": 2.554544448852539, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.899929404258728, + "num_tokens": 483770875.0, + "step": 12462 + }, + { + "epoch": 1.5854217020735275, + "grad_norm": 2.779017210006714, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8859649896621704, + "num_tokens": 483808773.0, + "step": 12463 + }, + { + "epoch": 1.585548912352118, + "grad_norm": 2.8179266452789307, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8902950286865234, + "num_tokens": 483847115.0, + "step": 12464 + }, + { + "epoch": 1.5856761226307086, + "grad_norm": 2.348210334777832, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9103190898895264, + "num_tokens": 483887893.0, + "step": 12465 + }, + { + "epoch": 1.585803332909299, + "grad_norm": 2.511644124984741, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8926050662994385, + "num_tokens": 483927972.0, + "step": 12466 + }, + { + "epoch": 1.5859305431878896, + "grad_norm": 2.745469570159912, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.889896035194397, + "num_tokens": 483965322.0, + "step": 12467 + }, + { + "epoch": 1.5860577534664801, + "grad_norm": 2.485611915588379, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8939473032951355, + "num_tokens": 484008681.0, + "step": 12468 + }, + { + "epoch": 1.5861849637450707, + "grad_norm": 2.8109652996063232, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8838534355163574, + "num_tokens": 484043855.0, + "step": 12469 + }, + { + "epoch": 1.5863121740236612, + "grad_norm": 2.809995651245117, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8981839418411255, + "num_tokens": 484076911.0, + "step": 12470 + }, + { + "epoch": 1.5864393843022517, + "grad_norm": 2.7010891437530518, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8980870246887207, + "num_tokens": 484111799.0, + "step": 12471 + }, + { + "epoch": 1.5865665945808423, + "grad_norm": 2.681159496307373, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8903517127037048, + "num_tokens": 484149893.0, + "step": 12472 + }, + { + "epoch": 1.5866938048594328, + "grad_norm": 2.5915348529815674, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8836005330085754, + "num_tokens": 484189887.0, + "step": 12473 + }, + { + "epoch": 1.5868210151380233, + "grad_norm": 2.31695294380188, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9019489884376526, + "num_tokens": 484233387.0, + "step": 12474 + }, + { + "epoch": 1.5869482254166136, + "grad_norm": 2.816218614578247, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8890385627746582, + "num_tokens": 484266370.0, + "step": 12475 + }, + { + "epoch": 1.5870754356952042, + "grad_norm": 2.588815212249756, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.8857307434082031, + "num_tokens": 484306765.0, + "step": 12476 + }, + { + "epoch": 1.5872026459737947, + "grad_norm": 2.789611577987671, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.901972234249115, + "num_tokens": 484341072.0, + "step": 12477 + }, + { + "epoch": 1.5873298562523852, + "grad_norm": 2.634063243865967, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8885027170181274, + "num_tokens": 484380280.0, + "step": 12478 + }, + { + "epoch": 1.5874570665309757, + "grad_norm": 2.493838310241699, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8968315124511719, + "num_tokens": 484415181.0, + "step": 12479 + }, + { + "epoch": 1.5875842768095663, + "grad_norm": 2.7221450805664062, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8783668875694275, + "num_tokens": 484455779.0, + "step": 12480 + }, + { + "epoch": 1.5877114870881566, + "grad_norm": 2.6401619911193848, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8906092047691345, + "num_tokens": 484494657.0, + "step": 12481 + }, + { + "epoch": 1.587838697366747, + "grad_norm": 2.5958127975463867, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.897158145904541, + "num_tokens": 484530329.0, + "step": 12482 + }, + { + "epoch": 1.5879659076453376, + "grad_norm": 2.5349063873291016, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8938050270080566, + "num_tokens": 484570397.0, + "step": 12483 + }, + { + "epoch": 1.5880931179239282, + "grad_norm": 2.5804569721221924, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8946877717971802, + "num_tokens": 484609933.0, + "step": 12484 + }, + { + "epoch": 1.5882203282025187, + "grad_norm": 2.797502279281616, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8906764984130859, + "num_tokens": 484647710.0, + "step": 12485 + }, + { + "epoch": 1.5883475384811092, + "grad_norm": 2.5599050521850586, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8859217166900635, + "num_tokens": 484686645.0, + "step": 12486 + }, + { + "epoch": 1.5884747487596997, + "grad_norm": 2.5367021560668945, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8940966129302979, + "num_tokens": 484722785.0, + "step": 12487 + }, + { + "epoch": 1.5886019590382903, + "grad_norm": 2.8818836212158203, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8811982870101929, + "num_tokens": 484761458.0, + "step": 12488 + }, + { + "epoch": 1.5887291693168808, + "grad_norm": 2.636320114135742, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8987582921981812, + "num_tokens": 484797031.0, + "step": 12489 + }, + { + "epoch": 1.5888563795954713, + "grad_norm": 2.5653557777404785, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.887630045413971, + "num_tokens": 484839867.0, + "step": 12490 + }, + { + "epoch": 1.5889835898740619, + "grad_norm": 2.3543426990509033, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9031979441642761, + "num_tokens": 484881817.0, + "step": 12491 + }, + { + "epoch": 1.5891108001526524, + "grad_norm": 2.5724949836730957, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8905927538871765, + "num_tokens": 484920271.0, + "step": 12492 + }, + { + "epoch": 1.589238010431243, + "grad_norm": 2.4054157733917236, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8950285911560059, + "num_tokens": 484963693.0, + "step": 12493 + }, + { + "epoch": 1.5893652207098334, + "grad_norm": 2.903007984161377, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8838937282562256, + "num_tokens": 485000742.0, + "step": 12494 + }, + { + "epoch": 1.589492430988424, + "grad_norm": 2.8160037994384766, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8940154314041138, + "num_tokens": 485034631.0, + "step": 12495 + }, + { + "epoch": 1.5896196412670145, + "grad_norm": 2.333052396774292, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9019283056259155, + "num_tokens": 485074667.0, + "step": 12496 + }, + { + "epoch": 1.589746851545605, + "grad_norm": 2.6328845024108887, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8927326202392578, + "num_tokens": 485111660.0, + "step": 12497 + }, + { + "epoch": 1.5898740618241956, + "grad_norm": 2.530634641647339, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8964633941650391, + "num_tokens": 485153830.0, + "step": 12498 + }, + { + "epoch": 1.5900012721027859, + "grad_norm": 2.956831693649292, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8769845366477966, + "num_tokens": 485185453.0, + "step": 12499 + }, + { + "epoch": 1.5901284823813764, + "grad_norm": 2.770498752593994, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8860585689544678, + "num_tokens": 485224169.0, + "step": 12500 + }, + { + "epoch": 1.590255692659967, + "grad_norm": 2.855982542037964, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9007399082183838, + "num_tokens": 485254201.0, + "step": 12501 + }, + { + "epoch": 1.5903829029385574, + "grad_norm": 2.4809107780456543, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8865836262702942, + "num_tokens": 485297772.0, + "step": 12502 + }, + { + "epoch": 1.590510113217148, + "grad_norm": 2.583972454071045, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.899734377861023, + "num_tokens": 485334097.0, + "step": 12503 + }, + { + "epoch": 1.5906373234957385, + "grad_norm": 2.9277026653289795, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8830679655075073, + "num_tokens": 485370456.0, + "step": 12504 + }, + { + "epoch": 1.5907645337743288, + "grad_norm": 2.8107845783233643, + "learning_rate": 1e-06, + "loss": 0.3423, + "mean_token_accuracy": 0.8780745267868042, + "num_tokens": 485410147.0, + "step": 12505 + }, + { + "epoch": 1.5908917440529193, + "grad_norm": 2.533992052078247, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8899823427200317, + "num_tokens": 485455604.0, + "step": 12506 + }, + { + "epoch": 1.5910189543315099, + "grad_norm": 2.986374616622925, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8885952234268188, + "num_tokens": 485487473.0, + "step": 12507 + }, + { + "epoch": 1.5911461646101004, + "grad_norm": 2.3683741092681885, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8958992958068848, + "num_tokens": 485529121.0, + "step": 12508 + }, + { + "epoch": 1.591273374888691, + "grad_norm": 3.0107133388519287, + "learning_rate": 1e-06, + "loss": 0.3516, + "mean_token_accuracy": 0.876897931098938, + "num_tokens": 485571826.0, + "step": 12509 + }, + { + "epoch": 1.5914005851672814, + "grad_norm": 2.6178998947143555, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.8992618322372437, + "num_tokens": 485607603.0, + "step": 12510 + }, + { + "epoch": 1.591527795445872, + "grad_norm": 2.834679126739502, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8816567063331604, + "num_tokens": 485644141.0, + "step": 12511 + }, + { + "epoch": 1.5916550057244625, + "grad_norm": 2.5973222255706787, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8913454413414001, + "num_tokens": 485683532.0, + "step": 12512 + }, + { + "epoch": 1.591782216003053, + "grad_norm": 2.529181957244873, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8964264392852783, + "num_tokens": 485719110.0, + "step": 12513 + }, + { + "epoch": 1.5919094262816436, + "grad_norm": 2.510805606842041, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8934755921363831, + "num_tokens": 485758845.0, + "step": 12514 + }, + { + "epoch": 1.592036636560234, + "grad_norm": 2.95983624458313, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.8756070733070374, + "num_tokens": 485798966.0, + "step": 12515 + }, + { + "epoch": 1.5921638468388246, + "grad_norm": 2.7797341346740723, + "learning_rate": 1e-06, + "loss": 0.3583, + "mean_token_accuracy": 0.8771397471427917, + "num_tokens": 485835670.0, + "step": 12516 + }, + { + "epoch": 1.5922910571174151, + "grad_norm": 2.6404080390930176, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8913828134536743, + "num_tokens": 485872064.0, + "step": 12517 + }, + { + "epoch": 1.5924182673960057, + "grad_norm": 2.6095492839813232, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8958383798599243, + "num_tokens": 485908730.0, + "step": 12518 + }, + { + "epoch": 1.5925454776745962, + "grad_norm": 2.5947697162628174, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8884768486022949, + "num_tokens": 485946172.0, + "step": 12519 + }, + { + "epoch": 1.5926726879531867, + "grad_norm": 2.6123857498168945, + "learning_rate": 1e-06, + "loss": 0.3531, + "mean_token_accuracy": 0.8729661703109741, + "num_tokens": 485992624.0, + "step": 12520 + }, + { + "epoch": 1.5927998982317773, + "grad_norm": 2.8841114044189453, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8990684747695923, + "num_tokens": 486029514.0, + "step": 12521 + }, + { + "epoch": 1.5929271085103678, + "grad_norm": 2.685129404067993, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8834775686264038, + "num_tokens": 486072698.0, + "step": 12522 + }, + { + "epoch": 1.5930543187889583, + "grad_norm": 2.7528038024902344, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9025027751922607, + "num_tokens": 486103175.0, + "step": 12523 + }, + { + "epoch": 1.5931815290675486, + "grad_norm": 2.810530662536621, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8922504782676697, + "num_tokens": 486139195.0, + "step": 12524 + }, + { + "epoch": 1.5933087393461391, + "grad_norm": 2.6887505054473877, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8959486484527588, + "num_tokens": 486172075.0, + "step": 12525 + }, + { + "epoch": 1.5934359496247297, + "grad_norm": 2.608858346939087, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8994331955909729, + "num_tokens": 486210789.0, + "step": 12526 + }, + { + "epoch": 1.5935631599033202, + "grad_norm": 2.7612736225128174, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8884545564651489, + "num_tokens": 486251336.0, + "step": 12527 + }, + { + "epoch": 1.5936903701819107, + "grad_norm": 2.7443466186523438, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8875126838684082, + "num_tokens": 486290121.0, + "step": 12528 + }, + { + "epoch": 1.5938175804605013, + "grad_norm": 2.5943644046783447, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8946070671081543, + "num_tokens": 486330850.0, + "step": 12529 + }, + { + "epoch": 1.5939447907390916, + "grad_norm": 2.6962413787841797, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8847101926803589, + "num_tokens": 486366575.0, + "step": 12530 + }, + { + "epoch": 1.594072001017682, + "grad_norm": 2.67887806892395, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8965858221054077, + "num_tokens": 486404882.0, + "step": 12531 + }, + { + "epoch": 1.5941992112962726, + "grad_norm": 2.405571937561035, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8963956832885742, + "num_tokens": 486447784.0, + "step": 12532 + }, + { + "epoch": 1.5943264215748632, + "grad_norm": 2.6681978702545166, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8790948987007141, + "num_tokens": 486490238.0, + "step": 12533 + }, + { + "epoch": 1.5944536318534537, + "grad_norm": 2.542804718017578, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8978334665298462, + "num_tokens": 486527613.0, + "step": 12534 + }, + { + "epoch": 1.5945808421320442, + "grad_norm": 2.4972290992736816, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8862766027450562, + "num_tokens": 486572340.0, + "step": 12535 + }, + { + "epoch": 1.5947080524106347, + "grad_norm": 2.6504619121551514, + "learning_rate": 1e-06, + "loss": 0.3555, + "mean_token_accuracy": 0.8702694177627563, + "num_tokens": 486614567.0, + "step": 12536 + }, + { + "epoch": 1.5948352626892253, + "grad_norm": 2.650102138519287, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.89448481798172, + "num_tokens": 486648632.0, + "step": 12537 + }, + { + "epoch": 1.5949624729678158, + "grad_norm": 2.4264016151428223, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8949739933013916, + "num_tokens": 486691433.0, + "step": 12538 + }, + { + "epoch": 1.5950896832464063, + "grad_norm": 2.598398208618164, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8889940977096558, + "num_tokens": 486729487.0, + "step": 12539 + }, + { + "epoch": 1.5952168935249968, + "grad_norm": 2.660583734512329, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8910229206085205, + "num_tokens": 486765230.0, + "step": 12540 + }, + { + "epoch": 1.5953441038035874, + "grad_norm": 2.734515428543091, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8920567035675049, + "num_tokens": 486800051.0, + "step": 12541 + }, + { + "epoch": 1.595471314082178, + "grad_norm": 2.4024646282196045, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8967642188072205, + "num_tokens": 486845810.0, + "step": 12542 + }, + { + "epoch": 1.5955985243607684, + "grad_norm": 2.8332152366638184, + "learning_rate": 1e-06, + "loss": 0.3751, + "mean_token_accuracy": 0.8681941628456116, + "num_tokens": 486885581.0, + "step": 12543 + }, + { + "epoch": 1.595725734639359, + "grad_norm": 2.5897629261016846, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9026439189910889, + "num_tokens": 486920966.0, + "step": 12544 + }, + { + "epoch": 1.5958529449179495, + "grad_norm": 2.5166659355163574, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8871088624000549, + "num_tokens": 486962994.0, + "step": 12545 + }, + { + "epoch": 1.59598015519654, + "grad_norm": 2.6120641231536865, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8823884725570679, + "num_tokens": 487004378.0, + "step": 12546 + }, + { + "epoch": 1.5961073654751305, + "grad_norm": 2.4468069076538086, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.8979713916778564, + "num_tokens": 487047085.0, + "step": 12547 + }, + { + "epoch": 1.5962345757537209, + "grad_norm": 2.8723907470703125, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8946841359138489, + "num_tokens": 487082560.0, + "step": 12548 + }, + { + "epoch": 1.5963617860323114, + "grad_norm": 2.275006055831909, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9014261364936829, + "num_tokens": 487124791.0, + "step": 12549 + }, + { + "epoch": 1.596488996310902, + "grad_norm": 2.6162962913513184, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8838768601417542, + "num_tokens": 487166407.0, + "step": 12550 + }, + { + "epoch": 1.5966162065894924, + "grad_norm": 2.838778257369995, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8927042484283447, + "num_tokens": 487199841.0, + "step": 12551 + }, + { + "epoch": 1.596743416868083, + "grad_norm": 2.7927215099334717, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8876115679740906, + "num_tokens": 487234648.0, + "step": 12552 + }, + { + "epoch": 1.5968706271466735, + "grad_norm": 2.7964396476745605, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8852293491363525, + "num_tokens": 487271351.0, + "step": 12553 + }, + { + "epoch": 1.5969978374252638, + "grad_norm": 2.4507558345794678, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9005117416381836, + "num_tokens": 487314760.0, + "step": 12554 + }, + { + "epoch": 1.5971250477038543, + "grad_norm": 2.736260414123535, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8950148820877075, + "num_tokens": 487348946.0, + "step": 12555 + }, + { + "epoch": 1.5972522579824449, + "grad_norm": 2.548227548599243, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8895106911659241, + "num_tokens": 487389581.0, + "step": 12556 + }, + { + "epoch": 1.5973794682610354, + "grad_norm": 2.652029275894165, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.887362003326416, + "num_tokens": 487426250.0, + "step": 12557 + }, + { + "epoch": 1.597506678539626, + "grad_norm": 2.6073994636535645, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8965017795562744, + "num_tokens": 487462140.0, + "step": 12558 + }, + { + "epoch": 1.5976338888182164, + "grad_norm": 2.8009767532348633, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8875645399093628, + "num_tokens": 487495400.0, + "step": 12559 + }, + { + "epoch": 1.597761099096807, + "grad_norm": 2.7239222526550293, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.88329017162323, + "num_tokens": 487534851.0, + "step": 12560 + }, + { + "epoch": 1.5978883093753975, + "grad_norm": 2.6555368900299072, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8817273378372192, + "num_tokens": 487574304.0, + "step": 12561 + }, + { + "epoch": 1.598015519653988, + "grad_norm": 2.8610243797302246, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8823847770690918, + "num_tokens": 487609804.0, + "step": 12562 + }, + { + "epoch": 1.5981427299325786, + "grad_norm": 3.0621402263641357, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8834728002548218, + "num_tokens": 487640975.0, + "step": 12563 + }, + { + "epoch": 1.598269940211169, + "grad_norm": 2.586930274963379, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8908485770225525, + "num_tokens": 487677858.0, + "step": 12564 + }, + { + "epoch": 1.5983971504897596, + "grad_norm": 3.0439341068267822, + "learning_rate": 1e-06, + "loss": 0.3633, + "mean_token_accuracy": 0.8711076378822327, + "num_tokens": 487713810.0, + "step": 12565 + }, + { + "epoch": 1.5985243607683501, + "grad_norm": 2.7571446895599365, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8891221284866333, + "num_tokens": 487750369.0, + "step": 12566 + }, + { + "epoch": 1.5986515710469407, + "grad_norm": 2.6509926319122314, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8825733661651611, + "num_tokens": 487793932.0, + "step": 12567 + }, + { + "epoch": 1.5987787813255312, + "grad_norm": 2.278378486633301, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.898794412612915, + "num_tokens": 487841023.0, + "step": 12568 + }, + { + "epoch": 1.5989059916041217, + "grad_norm": 2.595113515853882, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8975857496261597, + "num_tokens": 487880505.0, + "step": 12569 + }, + { + "epoch": 1.5990332018827123, + "grad_norm": 2.571152448654175, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.879851222038269, + "num_tokens": 487926943.0, + "step": 12570 + }, + { + "epoch": 1.5991604121613028, + "grad_norm": 2.7125110626220703, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8953840136528015, + "num_tokens": 487966362.0, + "step": 12571 + }, + { + "epoch": 1.5992876224398933, + "grad_norm": 2.718931198120117, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8930708169937134, + "num_tokens": 488002562.0, + "step": 12572 + }, + { + "epoch": 1.5994148327184836, + "grad_norm": 2.180922031402588, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9023458361625671, + "num_tokens": 488049936.0, + "step": 12573 + }, + { + "epoch": 1.5995420429970741, + "grad_norm": 2.529484510421753, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9021535515785217, + "num_tokens": 488086017.0, + "step": 12574 + }, + { + "epoch": 1.5996692532756647, + "grad_norm": 2.722285509109497, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8865270018577576, + "num_tokens": 488123360.0, + "step": 12575 + }, + { + "epoch": 1.5997964635542552, + "grad_norm": 2.994819402694702, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8842244744300842, + "num_tokens": 488155504.0, + "step": 12576 + }, + { + "epoch": 1.5999236738328457, + "grad_norm": 2.781031370162964, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8889212012290955, + "num_tokens": 488190691.0, + "step": 12577 + }, + { + "epoch": 1.6000508841114363, + "grad_norm": 2.707369804382324, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8966572284698486, + "num_tokens": 488224366.0, + "step": 12578 + }, + { + "epoch": 1.6001780943900266, + "grad_norm": 2.7335476875305176, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8841260075569153, + "num_tokens": 488262115.0, + "step": 12579 + }, + { + "epoch": 1.600305304668617, + "grad_norm": 2.57759165763855, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8997307419776917, + "num_tokens": 488301592.0, + "step": 12580 + }, + { + "epoch": 1.6004325149472076, + "grad_norm": 2.804701328277588, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8908023238182068, + "num_tokens": 488335270.0, + "step": 12581 + }, + { + "epoch": 1.6005597252257981, + "grad_norm": 2.556281566619873, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8853055834770203, + "num_tokens": 488375565.0, + "step": 12582 + }, + { + "epoch": 1.6006869355043887, + "grad_norm": 2.6361186504364014, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8910484313964844, + "num_tokens": 488415205.0, + "step": 12583 + }, + { + "epoch": 1.6008141457829792, + "grad_norm": 2.6537225246429443, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8875155448913574, + "num_tokens": 488457119.0, + "step": 12584 + }, + { + "epoch": 1.6009413560615697, + "grad_norm": 2.6176540851593018, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9007145762443542, + "num_tokens": 488493990.0, + "step": 12585 + }, + { + "epoch": 1.6010685663401603, + "grad_norm": 2.755434989929199, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8823139667510986, + "num_tokens": 488530257.0, + "step": 12586 + }, + { + "epoch": 1.6011957766187508, + "grad_norm": 2.9744505882263184, + "learning_rate": 1e-06, + "loss": 0.3705, + "mean_token_accuracy": 0.8730154037475586, + "num_tokens": 488569281.0, + "step": 12587 + }, + { + "epoch": 1.6013229868973413, + "grad_norm": 2.801518201828003, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8930881023406982, + "num_tokens": 488602701.0, + "step": 12588 + }, + { + "epoch": 1.6014501971759318, + "grad_norm": 2.80859375, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8927351832389832, + "num_tokens": 488638730.0, + "step": 12589 + }, + { + "epoch": 1.6015774074545224, + "grad_norm": 2.4023544788360596, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8952276706695557, + "num_tokens": 488682896.0, + "step": 12590 + }, + { + "epoch": 1.601704617733113, + "grad_norm": 2.627312183380127, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.887959897518158, + "num_tokens": 488724699.0, + "step": 12591 + }, + { + "epoch": 1.6018318280117034, + "grad_norm": 2.519958972930908, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8912017941474915, + "num_tokens": 488766085.0, + "step": 12592 + }, + { + "epoch": 1.601959038290294, + "grad_norm": 2.6307830810546875, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8842941522598267, + "num_tokens": 488808107.0, + "step": 12593 + }, + { + "epoch": 1.6020862485688845, + "grad_norm": 2.667383909225464, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8969629406929016, + "num_tokens": 488846483.0, + "step": 12594 + }, + { + "epoch": 1.602213458847475, + "grad_norm": 2.5371639728546143, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8953663110733032, + "num_tokens": 488887089.0, + "step": 12595 + }, + { + "epoch": 1.6023406691260655, + "grad_norm": 2.800485372543335, + "learning_rate": 1e-06, + "loss": 0.3626, + "mean_token_accuracy": 0.874240517616272, + "num_tokens": 488924600.0, + "step": 12596 + }, + { + "epoch": 1.6024678794046558, + "grad_norm": 2.5342133045196533, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8966960906982422, + "num_tokens": 488966530.0, + "step": 12597 + }, + { + "epoch": 1.6025950896832464, + "grad_norm": 2.3064026832580566, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9054194092750549, + "num_tokens": 489010524.0, + "step": 12598 + }, + { + "epoch": 1.602722299961837, + "grad_norm": 2.5321006774902344, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9002737998962402, + "num_tokens": 489050174.0, + "step": 12599 + }, + { + "epoch": 1.6028495102404274, + "grad_norm": 3.2608208656311035, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8772866725921631, + "num_tokens": 489083661.0, + "step": 12600 + }, + { + "epoch": 1.602976720519018, + "grad_norm": 2.639162302017212, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.891792893409729, + "num_tokens": 489122444.0, + "step": 12601 + }, + { + "epoch": 1.6031039307976085, + "grad_norm": 2.50113844871521, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9069482088088989, + "num_tokens": 489159844.0, + "step": 12602 + }, + { + "epoch": 1.6032311410761988, + "grad_norm": 2.7947847843170166, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8896609544754028, + "num_tokens": 489194261.0, + "step": 12603 + }, + { + "epoch": 1.6033583513547893, + "grad_norm": 2.656235933303833, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8827845454216003, + "num_tokens": 489232302.0, + "step": 12604 + }, + { + "epoch": 1.6034855616333799, + "grad_norm": 2.7033092975616455, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8950420618057251, + "num_tokens": 489269420.0, + "step": 12605 + }, + { + "epoch": 1.6036127719119704, + "grad_norm": 2.5145907402038574, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.890751302242279, + "num_tokens": 489313063.0, + "step": 12606 + }, + { + "epoch": 1.603739982190561, + "grad_norm": 2.6092264652252197, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.8996652364730835, + "num_tokens": 489348588.0, + "step": 12607 + }, + { + "epoch": 1.6038671924691514, + "grad_norm": 2.495698928833008, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8851460218429565, + "num_tokens": 489394950.0, + "step": 12608 + }, + { + "epoch": 1.603994402747742, + "grad_norm": 2.5971779823303223, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8954781889915466, + "num_tokens": 489434960.0, + "step": 12609 + }, + { + "epoch": 1.6041216130263325, + "grad_norm": 2.7282180786132812, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.892583429813385, + "num_tokens": 489475671.0, + "step": 12610 + }, + { + "epoch": 1.604248823304923, + "grad_norm": 2.7262916564941406, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8886767625808716, + "num_tokens": 489512933.0, + "step": 12611 + }, + { + "epoch": 1.6043760335835135, + "grad_norm": 2.8310434818267822, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8884928822517395, + "num_tokens": 489549660.0, + "step": 12612 + }, + { + "epoch": 1.604503243862104, + "grad_norm": 2.6946218013763428, + "learning_rate": 1e-06, + "loss": 0.3691, + "mean_token_accuracy": 0.8718705177307129, + "num_tokens": 489592749.0, + "step": 12613 + }, + { + "epoch": 1.6046304541406946, + "grad_norm": 2.915198802947998, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8904213905334473, + "num_tokens": 489625462.0, + "step": 12614 + }, + { + "epoch": 1.6047576644192851, + "grad_norm": 2.6561126708984375, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8844588398933411, + "num_tokens": 489666008.0, + "step": 12615 + }, + { + "epoch": 1.6048848746978757, + "grad_norm": 2.602754831314087, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8892216682434082, + "num_tokens": 489707230.0, + "step": 12616 + }, + { + "epoch": 1.6050120849764662, + "grad_norm": 2.420534133911133, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.8994737267494202, + "num_tokens": 489750350.0, + "step": 12617 + }, + { + "epoch": 1.6051392952550567, + "grad_norm": 2.574772834777832, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.894483208656311, + "num_tokens": 489792639.0, + "step": 12618 + }, + { + "epoch": 1.6052665055336472, + "grad_norm": 2.4763965606689453, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.896531879901886, + "num_tokens": 489836163.0, + "step": 12619 + }, + { + "epoch": 1.6053937158122378, + "grad_norm": 2.6585397720336914, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8903189301490784, + "num_tokens": 489872009.0, + "step": 12620 + }, + { + "epoch": 1.6055209260908283, + "grad_norm": 2.8139665126800537, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8853988647460938, + "num_tokens": 489910840.0, + "step": 12621 + }, + { + "epoch": 1.6056481363694186, + "grad_norm": 2.766688108444214, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8966435790061951, + "num_tokens": 489944478.0, + "step": 12622 + }, + { + "epoch": 1.6057753466480091, + "grad_norm": 3.005976676940918, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8957430124282837, + "num_tokens": 489981986.0, + "step": 12623 + }, + { + "epoch": 1.6059025569265997, + "grad_norm": 2.5842740535736084, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8982077836990356, + "num_tokens": 490017858.0, + "step": 12624 + }, + { + "epoch": 1.6060297672051902, + "grad_norm": 2.6144320964813232, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9018096923828125, + "num_tokens": 490056239.0, + "step": 12625 + }, + { + "epoch": 1.6061569774837807, + "grad_norm": 2.6231234073638916, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8830620050430298, + "num_tokens": 490097172.0, + "step": 12626 + }, + { + "epoch": 1.6062841877623713, + "grad_norm": 2.6337127685546875, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.899095892906189, + "num_tokens": 490133546.0, + "step": 12627 + }, + { + "epoch": 1.6064113980409616, + "grad_norm": 2.5671091079711914, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8970251679420471, + "num_tokens": 490172204.0, + "step": 12628 + }, + { + "epoch": 1.606538608319552, + "grad_norm": 2.5917327404022217, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8850106000900269, + "num_tokens": 490212756.0, + "step": 12629 + }, + { + "epoch": 1.6066658185981426, + "grad_norm": 2.718019962310791, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.868635892868042, + "num_tokens": 490255189.0, + "step": 12630 + }, + { + "epoch": 1.6067930288767331, + "grad_norm": 2.5279836654663086, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8915162086486816, + "num_tokens": 490300332.0, + "step": 12631 + }, + { + "epoch": 1.6069202391553237, + "grad_norm": 2.763510227203369, + "learning_rate": 1e-06, + "loss": 0.3724, + "mean_token_accuracy": 0.868465781211853, + "num_tokens": 490342504.0, + "step": 12632 + }, + { + "epoch": 1.6070474494339142, + "grad_norm": 2.753910541534424, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8934764862060547, + "num_tokens": 490378736.0, + "step": 12633 + }, + { + "epoch": 1.6071746597125047, + "grad_norm": 2.6062960624694824, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.8970902562141418, + "num_tokens": 490417193.0, + "step": 12634 + }, + { + "epoch": 1.6073018699910953, + "grad_norm": 2.4489777088165283, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8944579362869263, + "num_tokens": 490456901.0, + "step": 12635 + }, + { + "epoch": 1.6074290802696858, + "grad_norm": 2.66166353225708, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8929340243339539, + "num_tokens": 490493455.0, + "step": 12636 + }, + { + "epoch": 1.6075562905482763, + "grad_norm": 2.3737480640411377, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8968878984451294, + "num_tokens": 490538951.0, + "step": 12637 + }, + { + "epoch": 1.6076835008268668, + "grad_norm": 2.7292633056640625, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8957201838493347, + "num_tokens": 490573016.0, + "step": 12638 + }, + { + "epoch": 1.6078107111054574, + "grad_norm": 2.5633962154388428, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8934124708175659, + "num_tokens": 490612447.0, + "step": 12639 + }, + { + "epoch": 1.607937921384048, + "grad_norm": 2.7463791370391846, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8866944313049316, + "num_tokens": 490649011.0, + "step": 12640 + }, + { + "epoch": 1.6080651316626384, + "grad_norm": 2.783433437347412, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8931735157966614, + "num_tokens": 490684626.0, + "step": 12641 + }, + { + "epoch": 1.608192341941229, + "grad_norm": 2.6433253288269043, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8932185173034668, + "num_tokens": 490722867.0, + "step": 12642 + }, + { + "epoch": 1.6083195522198195, + "grad_norm": 2.551293134689331, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8875194787979126, + "num_tokens": 490762034.0, + "step": 12643 + }, + { + "epoch": 1.60844676249841, + "grad_norm": 2.7285850048065186, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8782947063446045, + "num_tokens": 490801904.0, + "step": 12644 + }, + { + "epoch": 1.6085739727770005, + "grad_norm": 2.776704788208008, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9019532203674316, + "num_tokens": 490835285.0, + "step": 12645 + }, + { + "epoch": 1.6087011830555908, + "grad_norm": 2.7593984603881836, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8937292098999023, + "num_tokens": 490872327.0, + "step": 12646 + }, + { + "epoch": 1.6088283933341814, + "grad_norm": 2.448871612548828, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.904046893119812, + "num_tokens": 490911728.0, + "step": 12647 + }, + { + "epoch": 1.608955603612772, + "grad_norm": 2.5166566371917725, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.897685706615448, + "num_tokens": 490952399.0, + "step": 12648 + }, + { + "epoch": 1.6090828138913624, + "grad_norm": 2.8206958770751953, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8902174234390259, + "num_tokens": 490988360.0, + "step": 12649 + }, + { + "epoch": 1.609210024169953, + "grad_norm": 2.728276252746582, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8837878704071045, + "num_tokens": 491027415.0, + "step": 12650 + }, + { + "epoch": 1.6093372344485435, + "grad_norm": 2.4564874172210693, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9064478278160095, + "num_tokens": 491065985.0, + "step": 12651 + }, + { + "epoch": 1.6094644447271338, + "grad_norm": 2.805050849914551, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8784952163696289, + "num_tokens": 491103222.0, + "step": 12652 + }, + { + "epoch": 1.6095916550057243, + "grad_norm": 2.4265336990356445, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8945885300636292, + "num_tokens": 491147065.0, + "step": 12653 + }, + { + "epoch": 1.6097188652843148, + "grad_norm": 2.6286544799804688, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8965676426887512, + "num_tokens": 491181282.0, + "step": 12654 + }, + { + "epoch": 1.6098460755629054, + "grad_norm": 2.5476186275482178, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8877776861190796, + "num_tokens": 491229576.0, + "step": 12655 + }, + { + "epoch": 1.609973285841496, + "grad_norm": 2.479964017868042, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8950097560882568, + "num_tokens": 491268990.0, + "step": 12656 + }, + { + "epoch": 1.6101004961200864, + "grad_norm": 2.4795725345611572, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.897294819355011, + "num_tokens": 491306730.0, + "step": 12657 + }, + { + "epoch": 1.610227706398677, + "grad_norm": 2.61220645904541, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8908765912055969, + "num_tokens": 491345657.0, + "step": 12658 + }, + { + "epoch": 1.6103549166772675, + "grad_norm": 2.6684908866882324, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8796895146369934, + "num_tokens": 491385697.0, + "step": 12659 + }, + { + "epoch": 1.610482126955858, + "grad_norm": 2.5878167152404785, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.889462947845459, + "num_tokens": 491426538.0, + "step": 12660 + }, + { + "epoch": 1.6106093372344485, + "grad_norm": 2.3931331634521484, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.888759970664978, + "num_tokens": 491468630.0, + "step": 12661 + }, + { + "epoch": 1.610736547513039, + "grad_norm": 2.5709385871887207, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8899543881416321, + "num_tokens": 491509878.0, + "step": 12662 + }, + { + "epoch": 1.6108637577916296, + "grad_norm": 3.0022921562194824, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8875993490219116, + "num_tokens": 491540798.0, + "step": 12663 + }, + { + "epoch": 1.6109909680702201, + "grad_norm": 2.9407169818878174, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8787010312080383, + "num_tokens": 491576410.0, + "step": 12664 + }, + { + "epoch": 1.6111181783488107, + "grad_norm": 2.7726049423217773, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8892454504966736, + "num_tokens": 491612290.0, + "step": 12665 + }, + { + "epoch": 1.6112453886274012, + "grad_norm": 2.526040554046631, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8842976689338684, + "num_tokens": 491656393.0, + "step": 12666 + }, + { + "epoch": 1.6113725989059917, + "grad_norm": 2.534668207168579, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8886027336120605, + "num_tokens": 491698641.0, + "step": 12667 + }, + { + "epoch": 1.6114998091845822, + "grad_norm": 2.415548801422119, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.894084095954895, + "num_tokens": 491741119.0, + "step": 12668 + }, + { + "epoch": 1.6116270194631728, + "grad_norm": 2.6987380981445312, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.885797381401062, + "num_tokens": 491781752.0, + "step": 12669 + }, + { + "epoch": 1.6117542297417633, + "grad_norm": 2.575911045074463, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8946263790130615, + "num_tokens": 491820029.0, + "step": 12670 + }, + { + "epoch": 1.6118814400203536, + "grad_norm": 2.708467960357666, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8789601922035217, + "num_tokens": 491857595.0, + "step": 12671 + }, + { + "epoch": 1.6120086502989441, + "grad_norm": 2.7162654399871826, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8883979320526123, + "num_tokens": 491893381.0, + "step": 12672 + }, + { + "epoch": 1.6121358605775347, + "grad_norm": 2.6990861892700195, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8919157981872559, + "num_tokens": 491928304.0, + "step": 12673 + }, + { + "epoch": 1.6122630708561252, + "grad_norm": 2.386080741882324, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8927675485610962, + "num_tokens": 491972748.0, + "step": 12674 + }, + { + "epoch": 1.6123902811347157, + "grad_norm": 2.6669363975524902, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8883906602859497, + "num_tokens": 492011443.0, + "step": 12675 + }, + { + "epoch": 1.6125174914133062, + "grad_norm": 2.5214955806732178, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8898897171020508, + "num_tokens": 492055838.0, + "step": 12676 + }, + { + "epoch": 1.6126447016918966, + "grad_norm": 2.326388120651245, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9038863182067871, + "num_tokens": 492099961.0, + "step": 12677 + }, + { + "epoch": 1.612771911970487, + "grad_norm": 2.9363765716552734, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8860397934913635, + "num_tokens": 492133312.0, + "step": 12678 + }, + { + "epoch": 1.6128991222490776, + "grad_norm": 2.6433005332946777, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8906695246696472, + "num_tokens": 492173432.0, + "step": 12679 + }, + { + "epoch": 1.6130263325276681, + "grad_norm": 2.6485373973846436, + "learning_rate": 1e-06, + "loss": 0.3645, + "mean_token_accuracy": 0.8712259531021118, + "num_tokens": 492214578.0, + "step": 12680 + }, + { + "epoch": 1.6131535428062587, + "grad_norm": 2.556881904602051, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8922285437583923, + "num_tokens": 492254804.0, + "step": 12681 + }, + { + "epoch": 1.6132807530848492, + "grad_norm": 2.725764036178589, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8874924182891846, + "num_tokens": 492291106.0, + "step": 12682 + }, + { + "epoch": 1.6134079633634397, + "grad_norm": 2.474761486053467, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9042478799819946, + "num_tokens": 492328793.0, + "step": 12683 + }, + { + "epoch": 1.6135351736420303, + "grad_norm": 2.4187588691711426, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8971137404441833, + "num_tokens": 492371705.0, + "step": 12684 + }, + { + "epoch": 1.6136623839206208, + "grad_norm": 2.5842134952545166, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8862185478210449, + "num_tokens": 492411608.0, + "step": 12685 + }, + { + "epoch": 1.6137895941992113, + "grad_norm": 2.685819149017334, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8824003338813782, + "num_tokens": 492450826.0, + "step": 12686 + }, + { + "epoch": 1.6139168044778018, + "grad_norm": 2.4766855239868164, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9011915922164917, + "num_tokens": 492491024.0, + "step": 12687 + }, + { + "epoch": 1.6140440147563924, + "grad_norm": 2.660339832305908, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8988639116287231, + "num_tokens": 492527546.0, + "step": 12688 + }, + { + "epoch": 1.614171225034983, + "grad_norm": 2.614795684814453, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8989353179931641, + "num_tokens": 492565825.0, + "step": 12689 + }, + { + "epoch": 1.6142984353135734, + "grad_norm": 2.514303207397461, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.89452064037323, + "num_tokens": 492604797.0, + "step": 12690 + }, + { + "epoch": 1.614425645592164, + "grad_norm": 2.662407875061035, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.8804997205734253, + "num_tokens": 492644665.0, + "step": 12691 + }, + { + "epoch": 1.6145528558707545, + "grad_norm": 3.0466220378875732, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.882125973701477, + "num_tokens": 492679464.0, + "step": 12692 + }, + { + "epoch": 1.614680066149345, + "grad_norm": 2.6006362438201904, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8979566097259521, + "num_tokens": 492716215.0, + "step": 12693 + }, + { + "epoch": 1.6148072764279355, + "grad_norm": 2.597287178039551, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8900146484375, + "num_tokens": 492754176.0, + "step": 12694 + }, + { + "epoch": 1.6149344867065258, + "grad_norm": 2.628016471862793, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.90488600730896, + "num_tokens": 492787488.0, + "step": 12695 + }, + { + "epoch": 1.6150616969851164, + "grad_norm": 2.493993043899536, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8975379467010498, + "num_tokens": 492830203.0, + "step": 12696 + }, + { + "epoch": 1.615188907263707, + "grad_norm": 2.438230037689209, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8842703104019165, + "num_tokens": 492875145.0, + "step": 12697 + }, + { + "epoch": 1.6153161175422974, + "grad_norm": 2.307232618331909, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8962478637695312, + "num_tokens": 492919578.0, + "step": 12698 + }, + { + "epoch": 1.615443327820888, + "grad_norm": 2.417526960372925, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8888564705848694, + "num_tokens": 492966098.0, + "step": 12699 + }, + { + "epoch": 1.6155705380994785, + "grad_norm": 2.4442615509033203, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9042133092880249, + "num_tokens": 492999988.0, + "step": 12700 + }, + { + "epoch": 1.6156977483780688, + "grad_norm": 2.7397491931915283, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8901969790458679, + "num_tokens": 493036541.0, + "step": 12701 + }, + { + "epoch": 1.6158249586566593, + "grad_norm": 2.571887969970703, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8885751962661743, + "num_tokens": 493075350.0, + "step": 12702 + }, + { + "epoch": 1.6159521689352498, + "grad_norm": 2.7756941318511963, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8799921870231628, + "num_tokens": 493116676.0, + "step": 12703 + }, + { + "epoch": 1.6160793792138404, + "grad_norm": 2.4459714889526367, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8928459882736206, + "num_tokens": 493159799.0, + "step": 12704 + }, + { + "epoch": 1.616206589492431, + "grad_norm": 2.3690502643585205, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8996458053588867, + "num_tokens": 493204901.0, + "step": 12705 + }, + { + "epoch": 1.6163337997710214, + "grad_norm": 2.6129534244537354, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8926767110824585, + "num_tokens": 493246752.0, + "step": 12706 + }, + { + "epoch": 1.616461010049612, + "grad_norm": 2.723663568496704, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8831554055213928, + "num_tokens": 493289011.0, + "step": 12707 + }, + { + "epoch": 1.6165882203282025, + "grad_norm": 2.7180542945861816, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.8997101783752441, + "num_tokens": 493321626.0, + "step": 12708 + }, + { + "epoch": 1.616715430606793, + "grad_norm": 2.4364047050476074, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8998763561248779, + "num_tokens": 493358668.0, + "step": 12709 + }, + { + "epoch": 1.6168426408853835, + "grad_norm": 2.5142009258270264, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8990989923477173, + "num_tokens": 493398219.0, + "step": 12710 + }, + { + "epoch": 1.616969851163974, + "grad_norm": 2.5458922386169434, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8905806541442871, + "num_tokens": 493439664.0, + "step": 12711 + }, + { + "epoch": 1.6170970614425646, + "grad_norm": 2.6340408325195312, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.902681827545166, + "num_tokens": 493477147.0, + "step": 12712 + }, + { + "epoch": 1.6172242717211551, + "grad_norm": 2.626863479614258, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.882354199886322, + "num_tokens": 493515611.0, + "step": 12713 + }, + { + "epoch": 1.6173514819997457, + "grad_norm": 2.7871036529541016, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8980410099029541, + "num_tokens": 493550968.0, + "step": 12714 + }, + { + "epoch": 1.6174786922783362, + "grad_norm": 2.8010518550872803, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8863449096679688, + "num_tokens": 493590174.0, + "step": 12715 + }, + { + "epoch": 1.6176059025569267, + "grad_norm": 2.4277455806732178, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9062435030937195, + "num_tokens": 493628427.0, + "step": 12716 + }, + { + "epoch": 1.6177331128355172, + "grad_norm": 2.5821993350982666, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8838059902191162, + "num_tokens": 493670420.0, + "step": 12717 + }, + { + "epoch": 1.6178603231141078, + "grad_norm": 2.75931715965271, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.885084331035614, + "num_tokens": 493707684.0, + "step": 12718 + }, + { + "epoch": 1.6179875333926983, + "grad_norm": 2.8329050540924072, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8950711488723755, + "num_tokens": 493740332.0, + "step": 12719 + }, + { + "epoch": 1.6181147436712886, + "grad_norm": 2.6865456104278564, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8915929198265076, + "num_tokens": 493777585.0, + "step": 12720 + }, + { + "epoch": 1.6182419539498791, + "grad_norm": 2.5622620582580566, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8834636211395264, + "num_tokens": 493819139.0, + "step": 12721 + }, + { + "epoch": 1.6183691642284697, + "grad_norm": 2.791980028152466, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8871722221374512, + "num_tokens": 493850764.0, + "step": 12722 + }, + { + "epoch": 1.6184963745070602, + "grad_norm": 2.6136271953582764, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8993405103683472, + "num_tokens": 493887092.0, + "step": 12723 + }, + { + "epoch": 1.6186235847856507, + "grad_norm": 2.472834587097168, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8965638875961304, + "num_tokens": 493929216.0, + "step": 12724 + }, + { + "epoch": 1.6187507950642412, + "grad_norm": 2.581997871398926, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8994160890579224, + "num_tokens": 493966183.0, + "step": 12725 + }, + { + "epoch": 1.6188780053428315, + "grad_norm": 2.6119232177734375, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.902133584022522, + "num_tokens": 494003706.0, + "step": 12726 + }, + { + "epoch": 1.619005215621422, + "grad_norm": 2.6821606159210205, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8976863026618958, + "num_tokens": 494039606.0, + "step": 12727 + }, + { + "epoch": 1.6191324259000126, + "grad_norm": 2.653521776199341, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8992130160331726, + "num_tokens": 494076180.0, + "step": 12728 + }, + { + "epoch": 1.6192596361786031, + "grad_norm": 2.544764280319214, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8949933052062988, + "num_tokens": 494114951.0, + "step": 12729 + }, + { + "epoch": 1.6193868464571937, + "grad_norm": 2.5700507164001465, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8799725770950317, + "num_tokens": 494159695.0, + "step": 12730 + }, + { + "epoch": 1.6195140567357842, + "grad_norm": 2.6018447875976562, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8844866752624512, + "num_tokens": 494204604.0, + "step": 12731 + }, + { + "epoch": 1.6196412670143747, + "grad_norm": 2.751358985900879, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8769801259040833, + "num_tokens": 494243042.0, + "step": 12732 + }, + { + "epoch": 1.6197684772929652, + "grad_norm": 2.3170344829559326, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8998678922653198, + "num_tokens": 494286840.0, + "step": 12733 + }, + { + "epoch": 1.6198956875715558, + "grad_norm": 2.8145532608032227, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8802435398101807, + "num_tokens": 494323876.0, + "step": 12734 + }, + { + "epoch": 1.6200228978501463, + "grad_norm": 2.338505268096924, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8953670263290405, + "num_tokens": 494371748.0, + "step": 12735 + }, + { + "epoch": 1.6201501081287368, + "grad_norm": 2.727161407470703, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8939521312713623, + "num_tokens": 494406605.0, + "step": 12736 + }, + { + "epoch": 1.6202773184073274, + "grad_norm": 2.7669429779052734, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8896484375, + "num_tokens": 494442615.0, + "step": 12737 + }, + { + "epoch": 1.6204045286859179, + "grad_norm": 2.3069941997528076, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8994383811950684, + "num_tokens": 494485101.0, + "step": 12738 + }, + { + "epoch": 1.6205317389645084, + "grad_norm": 2.8235716819763184, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8911861777305603, + "num_tokens": 494518662.0, + "step": 12739 + }, + { + "epoch": 1.620658949243099, + "grad_norm": 2.5558829307556152, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8879774808883667, + "num_tokens": 494560500.0, + "step": 12740 + }, + { + "epoch": 1.6207861595216895, + "grad_norm": 2.4983580112457275, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8881065845489502, + "num_tokens": 494604766.0, + "step": 12741 + }, + { + "epoch": 1.62091336980028, + "grad_norm": 2.5525693893432617, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8909921646118164, + "num_tokens": 494644663.0, + "step": 12742 + }, + { + "epoch": 1.6210405800788705, + "grad_norm": 2.422403573989868, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9029943943023682, + "num_tokens": 494684801.0, + "step": 12743 + }, + { + "epoch": 1.6211677903574608, + "grad_norm": 2.644831895828247, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8939135074615479, + "num_tokens": 494725928.0, + "step": 12744 + }, + { + "epoch": 1.6212950006360514, + "grad_norm": 2.4444217681884766, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8989229202270508, + "num_tokens": 494766552.0, + "step": 12745 + }, + { + "epoch": 1.621422210914642, + "grad_norm": 2.6333460807800293, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8890843391418457, + "num_tokens": 494802937.0, + "step": 12746 + }, + { + "epoch": 1.6215494211932324, + "grad_norm": 2.5524230003356934, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8958603143692017, + "num_tokens": 494840368.0, + "step": 12747 + }, + { + "epoch": 1.621676631471823, + "grad_norm": 2.5392701625823975, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8850198984146118, + "num_tokens": 494884012.0, + "step": 12748 + }, + { + "epoch": 1.6218038417504135, + "grad_norm": 2.7511367797851562, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8880951404571533, + "num_tokens": 494918425.0, + "step": 12749 + }, + { + "epoch": 1.6219310520290038, + "grad_norm": 2.7457115650177, + "learning_rate": 1e-06, + "loss": 0.3308, + "mean_token_accuracy": 0.8815745711326599, + "num_tokens": 494956372.0, + "step": 12750 + }, + { + "epoch": 1.6220582623075943, + "grad_norm": 2.5178511142730713, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8972244262695312, + "num_tokens": 494997948.0, + "step": 12751 + }, + { + "epoch": 1.6221854725861848, + "grad_norm": 2.5429952144622803, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8942518830299377, + "num_tokens": 495037721.0, + "step": 12752 + }, + { + "epoch": 1.6223126828647754, + "grad_norm": 2.661973237991333, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.885075569152832, + "num_tokens": 495073357.0, + "step": 12753 + }, + { + "epoch": 1.622439893143366, + "grad_norm": 2.776089906692505, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8861255049705505, + "num_tokens": 495108806.0, + "step": 12754 + }, + { + "epoch": 1.6225671034219564, + "grad_norm": 2.501767635345459, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8937077522277832, + "num_tokens": 495146891.0, + "step": 12755 + }, + { + "epoch": 1.622694313700547, + "grad_norm": 2.881655216217041, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8938592672348022, + "num_tokens": 495177587.0, + "step": 12756 + }, + { + "epoch": 1.6228215239791375, + "grad_norm": 2.4254074096679688, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9052300453186035, + "num_tokens": 495218864.0, + "step": 12757 + }, + { + "epoch": 1.622948734257728, + "grad_norm": 2.755437135696411, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.8768539428710938, + "num_tokens": 495258082.0, + "step": 12758 + }, + { + "epoch": 1.6230759445363185, + "grad_norm": 2.520995616912842, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8932042121887207, + "num_tokens": 495297466.0, + "step": 12759 + }, + { + "epoch": 1.623203154814909, + "grad_norm": 2.8464436531066895, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8911725282669067, + "num_tokens": 495331509.0, + "step": 12760 + }, + { + "epoch": 1.6233303650934996, + "grad_norm": 2.804690361022949, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8835690021514893, + "num_tokens": 495363539.0, + "step": 12761 + }, + { + "epoch": 1.6234575753720901, + "grad_norm": 2.3684844970703125, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9022176265716553, + "num_tokens": 495408785.0, + "step": 12762 + }, + { + "epoch": 1.6235847856506807, + "grad_norm": 2.4682109355926514, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8892195224761963, + "num_tokens": 495450889.0, + "step": 12763 + }, + { + "epoch": 1.6237119959292712, + "grad_norm": 2.6142992973327637, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.894576370716095, + "num_tokens": 495485555.0, + "step": 12764 + }, + { + "epoch": 1.6238392062078617, + "grad_norm": 2.415844440460205, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8924709558486938, + "num_tokens": 495531635.0, + "step": 12765 + }, + { + "epoch": 1.6239664164864522, + "grad_norm": 2.69272518157959, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8945649862289429, + "num_tokens": 495573496.0, + "step": 12766 + }, + { + "epoch": 1.6240936267650428, + "grad_norm": 2.4669454097747803, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8945907950401306, + "num_tokens": 495616891.0, + "step": 12767 + }, + { + "epoch": 1.6242208370436333, + "grad_norm": 2.4628853797912598, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.8974826335906982, + "num_tokens": 495656238.0, + "step": 12768 + }, + { + "epoch": 1.6243480473222236, + "grad_norm": 2.6648740768432617, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8810532689094543, + "num_tokens": 495696330.0, + "step": 12769 + }, + { + "epoch": 1.6244752576008141, + "grad_norm": 2.7942702770233154, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8824657797813416, + "num_tokens": 495733311.0, + "step": 12770 + }, + { + "epoch": 1.6246024678794047, + "grad_norm": 2.511462926864624, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8974744081497192, + "num_tokens": 495771862.0, + "step": 12771 + }, + { + "epoch": 1.6247296781579952, + "grad_norm": 2.689051628112793, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8943559527397156, + "num_tokens": 495813583.0, + "step": 12772 + }, + { + "epoch": 1.6248568884365857, + "grad_norm": 2.7973461151123047, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.884183943271637, + "num_tokens": 495850447.0, + "step": 12773 + }, + { + "epoch": 1.6249840987151762, + "grad_norm": 2.670410633087158, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8897199630737305, + "num_tokens": 495885844.0, + "step": 12774 + }, + { + "epoch": 1.6251113089937665, + "grad_norm": 2.592197895050049, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8901515007019043, + "num_tokens": 495925010.0, + "step": 12775 + }, + { + "epoch": 1.625238519272357, + "grad_norm": 2.469593048095703, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8951412439346313, + "num_tokens": 495966229.0, + "step": 12776 + }, + { + "epoch": 1.6253657295509476, + "grad_norm": 2.4117889404296875, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8905775547027588, + "num_tokens": 496010931.0, + "step": 12777 + }, + { + "epoch": 1.6254929398295381, + "grad_norm": 2.6936285495758057, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8801720142364502, + "num_tokens": 496052696.0, + "step": 12778 + }, + { + "epoch": 1.6256201501081287, + "grad_norm": 2.5514767169952393, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8881829380989075, + "num_tokens": 496094210.0, + "step": 12779 + }, + { + "epoch": 1.6257473603867192, + "grad_norm": 2.578087091445923, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.888798177242279, + "num_tokens": 496132603.0, + "step": 12780 + }, + { + "epoch": 1.6258745706653097, + "grad_norm": 2.8397743701934814, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8943691849708557, + "num_tokens": 496166352.0, + "step": 12781 + }, + { + "epoch": 1.6260017809439002, + "grad_norm": 2.4810328483581543, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8898148536682129, + "num_tokens": 496208834.0, + "step": 12782 + }, + { + "epoch": 1.6261289912224908, + "grad_norm": 2.435530185699463, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8918161392211914, + "num_tokens": 496248167.0, + "step": 12783 + }, + { + "epoch": 1.6262562015010813, + "grad_norm": 2.5223190784454346, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8929701447486877, + "num_tokens": 496289877.0, + "step": 12784 + }, + { + "epoch": 1.6263834117796718, + "grad_norm": 2.521052122116089, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8808243274688721, + "num_tokens": 496333693.0, + "step": 12785 + }, + { + "epoch": 1.6265106220582624, + "grad_norm": 2.58860182762146, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8948267698287964, + "num_tokens": 496370488.0, + "step": 12786 + }, + { + "epoch": 1.6266378323368529, + "grad_norm": 2.467761516571045, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8952815532684326, + "num_tokens": 496412138.0, + "step": 12787 + }, + { + "epoch": 1.6267650426154434, + "grad_norm": 2.376391649246216, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8896371722221375, + "num_tokens": 496457411.0, + "step": 12788 + }, + { + "epoch": 1.626892252894034, + "grad_norm": 2.49391770362854, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8903716206550598, + "num_tokens": 496498798.0, + "step": 12789 + }, + { + "epoch": 1.6270194631726245, + "grad_norm": 2.8150110244750977, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8768677711486816, + "num_tokens": 496535647.0, + "step": 12790 + }, + { + "epoch": 1.627146673451215, + "grad_norm": 2.5673372745513916, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8944380283355713, + "num_tokens": 496575322.0, + "step": 12791 + }, + { + "epoch": 1.6272738837298055, + "grad_norm": 2.5941755771636963, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8956810235977173, + "num_tokens": 496610486.0, + "step": 12792 + }, + { + "epoch": 1.6274010940083958, + "grad_norm": 2.8299129009246826, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.886344313621521, + "num_tokens": 496646530.0, + "step": 12793 + }, + { + "epoch": 1.6275283042869864, + "grad_norm": 2.5918221473693848, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9024576544761658, + "num_tokens": 496683359.0, + "step": 12794 + }, + { + "epoch": 1.6276555145655769, + "grad_norm": 2.469949722290039, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9027631282806396, + "num_tokens": 496721891.0, + "step": 12795 + }, + { + "epoch": 1.6277827248441674, + "grad_norm": 2.5787951946258545, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.903857946395874, + "num_tokens": 496759756.0, + "step": 12796 + }, + { + "epoch": 1.627909935122758, + "grad_norm": 2.670074939727783, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.893187940120697, + "num_tokens": 496797117.0, + "step": 12797 + }, + { + "epoch": 1.6280371454013485, + "grad_norm": 2.8693931102752686, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.888129711151123, + "num_tokens": 496830564.0, + "step": 12798 + }, + { + "epoch": 1.6281643556799388, + "grad_norm": 2.5985705852508545, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8942527770996094, + "num_tokens": 496870117.0, + "step": 12799 + }, + { + "epoch": 1.6282915659585293, + "grad_norm": 2.4178273677825928, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9010919332504272, + "num_tokens": 496910484.0, + "step": 12800 + }, + { + "epoch": 1.6284187762371198, + "grad_norm": 2.7096481323242188, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8863298892974854, + "num_tokens": 496948700.0, + "step": 12801 + }, + { + "epoch": 1.6285459865157104, + "grad_norm": 2.559523105621338, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.891109049320221, + "num_tokens": 496985807.0, + "step": 12802 + }, + { + "epoch": 1.628673196794301, + "grad_norm": 2.7753617763519287, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8924586772918701, + "num_tokens": 497018708.0, + "step": 12803 + }, + { + "epoch": 1.6288004070728914, + "grad_norm": 2.695275068283081, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8797642588615417, + "num_tokens": 497057486.0, + "step": 12804 + }, + { + "epoch": 1.628927617351482, + "grad_norm": 2.801940441131592, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8835371732711792, + "num_tokens": 497094956.0, + "step": 12805 + }, + { + "epoch": 1.6290548276300725, + "grad_norm": 2.6198034286499023, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8907943367958069, + "num_tokens": 497136501.0, + "step": 12806 + }, + { + "epoch": 1.629182037908663, + "grad_norm": 2.850672960281372, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8911282420158386, + "num_tokens": 497176329.0, + "step": 12807 + }, + { + "epoch": 1.6293092481872535, + "grad_norm": 2.784114122390747, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8911137580871582, + "num_tokens": 497209668.0, + "step": 12808 + }, + { + "epoch": 1.629436458465844, + "grad_norm": 2.804857015609741, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8837842345237732, + "num_tokens": 497243749.0, + "step": 12809 + }, + { + "epoch": 1.6295636687444346, + "grad_norm": 2.742741823196411, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8841912746429443, + "num_tokens": 497278811.0, + "step": 12810 + }, + { + "epoch": 1.6296908790230251, + "grad_norm": 2.703444719314575, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8800052404403687, + "num_tokens": 497316294.0, + "step": 12811 + }, + { + "epoch": 1.6298180893016156, + "grad_norm": 2.8944215774536133, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8806939721107483, + "num_tokens": 497351964.0, + "step": 12812 + }, + { + "epoch": 1.6299452995802062, + "grad_norm": 3.108511447906494, + "learning_rate": 1e-06, + "loss": 0.3369, + "mean_token_accuracy": 0.8789919018745422, + "num_tokens": 497391013.0, + "step": 12813 + }, + { + "epoch": 1.6300725098587967, + "grad_norm": 2.581162691116333, + "learning_rate": 1e-06, + "loss": 0.332, + "mean_token_accuracy": 0.8815134763717651, + "num_tokens": 497433689.0, + "step": 12814 + }, + { + "epoch": 1.6301997201373872, + "grad_norm": 2.406109094619751, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8936342000961304, + "num_tokens": 497474941.0, + "step": 12815 + }, + { + "epoch": 1.6303269304159778, + "grad_norm": 2.55836820602417, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8960438966751099, + "num_tokens": 497514223.0, + "step": 12816 + }, + { + "epoch": 1.630454140694568, + "grad_norm": 2.662837505340576, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8804415464401245, + "num_tokens": 497555691.0, + "step": 12817 + }, + { + "epoch": 1.6305813509731586, + "grad_norm": 2.59376859664917, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8888645172119141, + "num_tokens": 497594016.0, + "step": 12818 + }, + { + "epoch": 1.6307085612517491, + "grad_norm": 2.5651111602783203, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8945879936218262, + "num_tokens": 497632915.0, + "step": 12819 + }, + { + "epoch": 1.6308357715303397, + "grad_norm": 2.705746650695801, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8916714787483215, + "num_tokens": 497671023.0, + "step": 12820 + }, + { + "epoch": 1.6309629818089302, + "grad_norm": 2.4942126274108887, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8896676301956177, + "num_tokens": 497714439.0, + "step": 12821 + }, + { + "epoch": 1.6310901920875207, + "grad_norm": 2.7394416332244873, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.886805534362793, + "num_tokens": 497753648.0, + "step": 12822 + }, + { + "epoch": 1.6312174023661112, + "grad_norm": 2.4887712001800537, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8904141187667847, + "num_tokens": 497793931.0, + "step": 12823 + }, + { + "epoch": 1.6313446126447015, + "grad_norm": 2.5679402351379395, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9008193016052246, + "num_tokens": 497830435.0, + "step": 12824 + }, + { + "epoch": 1.631471822923292, + "grad_norm": 2.822166681289673, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8870096206665039, + "num_tokens": 497865112.0, + "step": 12825 + }, + { + "epoch": 1.6315990332018826, + "grad_norm": 2.7344748973846436, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8981226682662964, + "num_tokens": 497905853.0, + "step": 12826 + }, + { + "epoch": 1.6317262434804731, + "grad_norm": 2.4305903911590576, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.891682505607605, + "num_tokens": 497946962.0, + "step": 12827 + }, + { + "epoch": 1.6318534537590637, + "grad_norm": 2.3066389560699463, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9059145450592041, + "num_tokens": 497985448.0, + "step": 12828 + }, + { + "epoch": 1.6319806640376542, + "grad_norm": 2.6893835067749023, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8932147026062012, + "num_tokens": 498023020.0, + "step": 12829 + }, + { + "epoch": 1.6321078743162447, + "grad_norm": 2.403168201446533, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9002016186714172, + "num_tokens": 498066063.0, + "step": 12830 + }, + { + "epoch": 1.6322350845948352, + "grad_norm": 2.7995827198028564, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8895394206047058, + "num_tokens": 498104351.0, + "step": 12831 + }, + { + "epoch": 1.6323622948734258, + "grad_norm": 2.6191437244415283, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8865039944648743, + "num_tokens": 498145444.0, + "step": 12832 + }, + { + "epoch": 1.6324895051520163, + "grad_norm": 2.53462815284729, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.886931300163269, + "num_tokens": 498194393.0, + "step": 12833 + }, + { + "epoch": 1.6326167154306068, + "grad_norm": 2.3972840309143066, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9026373624801636, + "num_tokens": 498236467.0, + "step": 12834 + }, + { + "epoch": 1.6327439257091974, + "grad_norm": 2.592292070388794, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8974372148513794, + "num_tokens": 498274715.0, + "step": 12835 + }, + { + "epoch": 1.6328711359877879, + "grad_norm": 2.432668447494507, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8973158597946167, + "num_tokens": 498316592.0, + "step": 12836 + }, + { + "epoch": 1.6329983462663784, + "grad_norm": 2.68921160697937, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8973008990287781, + "num_tokens": 498352753.0, + "step": 12837 + }, + { + "epoch": 1.633125556544969, + "grad_norm": 2.854884386062622, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8862981796264648, + "num_tokens": 498387040.0, + "step": 12838 + }, + { + "epoch": 1.6332527668235595, + "grad_norm": 2.5028936862945557, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8871984481811523, + "num_tokens": 498431444.0, + "step": 12839 + }, + { + "epoch": 1.63337997710215, + "grad_norm": 2.8200974464416504, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8892481923103333, + "num_tokens": 498465012.0, + "step": 12840 + }, + { + "epoch": 1.6335071873807405, + "grad_norm": 2.8154115676879883, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8892738819122314, + "num_tokens": 498500557.0, + "step": 12841 + }, + { + "epoch": 1.6336343976593308, + "grad_norm": 2.599370002746582, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8871866464614868, + "num_tokens": 498541229.0, + "step": 12842 + }, + { + "epoch": 1.6337616079379214, + "grad_norm": 2.473461627960205, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8865172863006592, + "num_tokens": 498582821.0, + "step": 12843 + }, + { + "epoch": 1.6338888182165119, + "grad_norm": 2.924111843109131, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8784210681915283, + "num_tokens": 498617413.0, + "step": 12844 + }, + { + "epoch": 1.6340160284951024, + "grad_norm": 2.9074184894561768, + "learning_rate": 1e-06, + "loss": 0.368, + "mean_token_accuracy": 0.8690652251243591, + "num_tokens": 498654176.0, + "step": 12845 + }, + { + "epoch": 1.634143238773693, + "grad_norm": 2.6892709732055664, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8862041234970093, + "num_tokens": 498692370.0, + "step": 12846 + }, + { + "epoch": 1.6342704490522835, + "grad_norm": 2.769573926925659, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8813811540603638, + "num_tokens": 498730131.0, + "step": 12847 + }, + { + "epoch": 1.6343976593308738, + "grad_norm": 2.2888219356536865, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8974903225898743, + "num_tokens": 498773748.0, + "step": 12848 + }, + { + "epoch": 1.6345248696094643, + "grad_norm": 2.5761895179748535, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8974428176879883, + "num_tokens": 498810008.0, + "step": 12849 + }, + { + "epoch": 1.6346520798880548, + "grad_norm": 2.439757823944092, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.8995152115821838, + "num_tokens": 498850753.0, + "step": 12850 + }, + { + "epoch": 1.6347792901666454, + "grad_norm": 2.760162115097046, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8860061168670654, + "num_tokens": 498886431.0, + "step": 12851 + }, + { + "epoch": 1.6349065004452359, + "grad_norm": 2.485893726348877, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9035707712173462, + "num_tokens": 498922364.0, + "step": 12852 + }, + { + "epoch": 1.6350337107238264, + "grad_norm": 2.390418767929077, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8845381736755371, + "num_tokens": 498969752.0, + "step": 12853 + }, + { + "epoch": 1.635160921002417, + "grad_norm": 2.3603217601776123, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8950101137161255, + "num_tokens": 499010896.0, + "step": 12854 + }, + { + "epoch": 1.6352881312810075, + "grad_norm": 2.571275234222412, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8947657346725464, + "num_tokens": 499045483.0, + "step": 12855 + }, + { + "epoch": 1.635415341559598, + "grad_norm": 2.6987247467041016, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8830587863922119, + "num_tokens": 499083620.0, + "step": 12856 + }, + { + "epoch": 1.6355425518381885, + "grad_norm": 2.647810220718384, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8821601271629333, + "num_tokens": 499125159.0, + "step": 12857 + }, + { + "epoch": 1.635669762116779, + "grad_norm": 2.6144280433654785, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.884992778301239, + "num_tokens": 499165180.0, + "step": 12858 + }, + { + "epoch": 1.6357969723953696, + "grad_norm": 2.7405128479003906, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8828537464141846, + "num_tokens": 499201956.0, + "step": 12859 + }, + { + "epoch": 1.6359241826739601, + "grad_norm": 2.7815046310424805, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8856655955314636, + "num_tokens": 499238500.0, + "step": 12860 + }, + { + "epoch": 1.6360513929525506, + "grad_norm": 2.7217891216278076, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8925026655197144, + "num_tokens": 499271804.0, + "step": 12861 + }, + { + "epoch": 1.6361786032311412, + "grad_norm": 2.522425651550293, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8985889554023743, + "num_tokens": 499310340.0, + "step": 12862 + }, + { + "epoch": 1.6363058135097317, + "grad_norm": 2.6756272315979004, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8837699294090271, + "num_tokens": 499347220.0, + "step": 12863 + }, + { + "epoch": 1.6364330237883222, + "grad_norm": 2.6517817974090576, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8858934640884399, + "num_tokens": 499385249.0, + "step": 12864 + }, + { + "epoch": 1.6365602340669128, + "grad_norm": 2.7268052101135254, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8943742513656616, + "num_tokens": 499421093.0, + "step": 12865 + }, + { + "epoch": 1.636687444345503, + "grad_norm": 2.724184036254883, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8795961737632751, + "num_tokens": 499461554.0, + "step": 12866 + }, + { + "epoch": 1.6368146546240936, + "grad_norm": 2.4906630516052246, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8910298347473145, + "num_tokens": 499506323.0, + "step": 12867 + }, + { + "epoch": 1.6369418649026841, + "grad_norm": 2.5386576652526855, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8927242159843445, + "num_tokens": 499546341.0, + "step": 12868 + }, + { + "epoch": 1.6370690751812746, + "grad_norm": 2.7515037059783936, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8868957757949829, + "num_tokens": 499582034.0, + "step": 12869 + }, + { + "epoch": 1.6371962854598652, + "grad_norm": 2.7834346294403076, + "learning_rate": 1e-06, + "loss": 0.3617, + "mean_token_accuracy": 0.8732177019119263, + "num_tokens": 499625991.0, + "step": 12870 + }, + { + "epoch": 1.6373234957384557, + "grad_norm": 2.7092137336730957, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8889660239219666, + "num_tokens": 499662728.0, + "step": 12871 + }, + { + "epoch": 1.6374507060170462, + "grad_norm": 2.606395959854126, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8901659250259399, + "num_tokens": 499703612.0, + "step": 12872 + }, + { + "epoch": 1.6375779162956365, + "grad_norm": 2.776034116744995, + "learning_rate": 1e-06, + "loss": 0.3151, + "mean_token_accuracy": 0.8890228271484375, + "num_tokens": 499746588.0, + "step": 12873 + }, + { + "epoch": 1.637705126574227, + "grad_norm": 2.7258121967315674, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9032956957817078, + "num_tokens": 499777482.0, + "step": 12874 + }, + { + "epoch": 1.6378323368528176, + "grad_norm": 2.5717530250549316, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8836994171142578, + "num_tokens": 499822790.0, + "step": 12875 + }, + { + "epoch": 1.6379595471314081, + "grad_norm": 2.500180721282959, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8962143063545227, + "num_tokens": 499864769.0, + "step": 12876 + }, + { + "epoch": 1.6380867574099987, + "grad_norm": 2.6615793704986572, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8907806873321533, + "num_tokens": 499903336.0, + "step": 12877 + }, + { + "epoch": 1.6382139676885892, + "grad_norm": 2.846466302871704, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8864403963088989, + "num_tokens": 499940821.0, + "step": 12878 + }, + { + "epoch": 1.6383411779671797, + "grad_norm": 2.746027946472168, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8793222904205322, + "num_tokens": 499979434.0, + "step": 12879 + }, + { + "epoch": 1.6384683882457702, + "grad_norm": 2.404855489730835, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8957757949829102, + "num_tokens": 500022762.0, + "step": 12880 + }, + { + "epoch": 1.6385955985243608, + "grad_norm": 2.534346103668213, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8902695178985596, + "num_tokens": 500064088.0, + "step": 12881 + }, + { + "epoch": 1.6387228088029513, + "grad_norm": 2.425525188446045, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8946070671081543, + "num_tokens": 500105979.0, + "step": 12882 + }, + { + "epoch": 1.6388500190815418, + "grad_norm": 2.8483471870422363, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.896543562412262, + "num_tokens": 500139696.0, + "step": 12883 + }, + { + "epoch": 1.6389772293601323, + "grad_norm": 2.594442844390869, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8869664072990417, + "num_tokens": 500182698.0, + "step": 12884 + }, + { + "epoch": 1.6391044396387229, + "grad_norm": 2.519876480102539, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.895796000957489, + "num_tokens": 500226353.0, + "step": 12885 + }, + { + "epoch": 1.6392316499173134, + "grad_norm": 2.6086971759796143, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8942887783050537, + "num_tokens": 500263381.0, + "step": 12886 + }, + { + "epoch": 1.639358860195904, + "grad_norm": 2.5827279090881348, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8884057998657227, + "num_tokens": 500305016.0, + "step": 12887 + }, + { + "epoch": 1.6394860704744945, + "grad_norm": 2.6568214893341064, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8934873342514038, + "num_tokens": 500341247.0, + "step": 12888 + }, + { + "epoch": 1.639613280753085, + "grad_norm": 2.5642282962799072, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8892468214035034, + "num_tokens": 500383835.0, + "step": 12889 + }, + { + "epoch": 1.6397404910316755, + "grad_norm": 2.4252161979675293, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8985315561294556, + "num_tokens": 500423736.0, + "step": 12890 + }, + { + "epoch": 1.6398677013102658, + "grad_norm": 2.439206838607788, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8980095982551575, + "num_tokens": 500465267.0, + "step": 12891 + }, + { + "epoch": 1.6399949115888564, + "grad_norm": 2.3863837718963623, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9011463522911072, + "num_tokens": 500509900.0, + "step": 12892 + }, + { + "epoch": 1.6401221218674469, + "grad_norm": 2.778141736984253, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8794870376586914, + "num_tokens": 500544214.0, + "step": 12893 + }, + { + "epoch": 1.6402493321460374, + "grad_norm": 2.2888991832733154, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8970465660095215, + "num_tokens": 500590597.0, + "step": 12894 + }, + { + "epoch": 1.640376542424628, + "grad_norm": 2.789891481399536, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8888987898826599, + "num_tokens": 500628028.0, + "step": 12895 + }, + { + "epoch": 1.6405037527032185, + "grad_norm": 2.822068929672241, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9068390727043152, + "num_tokens": 500657913.0, + "step": 12896 + }, + { + "epoch": 1.6406309629818088, + "grad_norm": 2.7726056575775146, + "learning_rate": 1e-06, + "loss": 0.3362, + "mean_token_accuracy": 0.8785841464996338, + "num_tokens": 500693812.0, + "step": 12897 + }, + { + "epoch": 1.6407581732603993, + "grad_norm": 2.774282693862915, + "learning_rate": 1e-06, + "loss": 0.337, + "mean_token_accuracy": 0.8800165057182312, + "num_tokens": 500733237.0, + "step": 12898 + }, + { + "epoch": 1.6408853835389898, + "grad_norm": 2.545117139816284, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8979136943817139, + "num_tokens": 500774053.0, + "step": 12899 + }, + { + "epoch": 1.6410125938175804, + "grad_norm": 2.7637312412261963, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8883184194564819, + "num_tokens": 500810719.0, + "step": 12900 + }, + { + "epoch": 1.6411398040961709, + "grad_norm": 2.5527610778808594, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.896405816078186, + "num_tokens": 500850364.0, + "step": 12901 + }, + { + "epoch": 1.6412670143747614, + "grad_norm": 2.8119449615478516, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8926110863685608, + "num_tokens": 500881853.0, + "step": 12902 + }, + { + "epoch": 1.641394224653352, + "grad_norm": 2.6525471210479736, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8945541381835938, + "num_tokens": 500917486.0, + "step": 12903 + }, + { + "epoch": 1.6415214349319425, + "grad_norm": 2.719149112701416, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.898343026638031, + "num_tokens": 500952678.0, + "step": 12904 + }, + { + "epoch": 1.641648645210533, + "grad_norm": 2.512214183807373, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8931268453598022, + "num_tokens": 500994186.0, + "step": 12905 + }, + { + "epoch": 1.6417758554891235, + "grad_norm": 2.417168140411377, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.90199875831604, + "num_tokens": 501034567.0, + "step": 12906 + }, + { + "epoch": 1.641903065767714, + "grad_norm": 2.739813804626465, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8870223760604858, + "num_tokens": 501069054.0, + "step": 12907 + }, + { + "epoch": 1.6420302760463046, + "grad_norm": 2.3914408683776855, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9093197584152222, + "num_tokens": 501108831.0, + "step": 12908 + }, + { + "epoch": 1.642157486324895, + "grad_norm": 2.8408873081207275, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8804654479026794, + "num_tokens": 501144617.0, + "step": 12909 + }, + { + "epoch": 1.6422846966034856, + "grad_norm": 2.488419532775879, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9027159214019775, + "num_tokens": 501182397.0, + "step": 12910 + }, + { + "epoch": 1.6424119068820762, + "grad_norm": 2.4560606479644775, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9047821760177612, + "num_tokens": 501224670.0, + "step": 12911 + }, + { + "epoch": 1.6425391171606667, + "grad_norm": 2.6091153621673584, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8870724439620972, + "num_tokens": 501261952.0, + "step": 12912 + }, + { + "epoch": 1.6426663274392572, + "grad_norm": 2.579570770263672, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8879619240760803, + "num_tokens": 501304295.0, + "step": 12913 + }, + { + "epoch": 1.6427935377178478, + "grad_norm": 2.4085519313812256, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8917688131332397, + "num_tokens": 501351842.0, + "step": 12914 + }, + { + "epoch": 1.642920747996438, + "grad_norm": 2.529791831970215, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8864922523498535, + "num_tokens": 501396026.0, + "step": 12915 + }, + { + "epoch": 1.6430479582750286, + "grad_norm": 2.5270066261291504, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8933330178260803, + "num_tokens": 501435340.0, + "step": 12916 + }, + { + "epoch": 1.6431751685536191, + "grad_norm": 2.6747336387634277, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8927516937255859, + "num_tokens": 501470450.0, + "step": 12917 + }, + { + "epoch": 1.6433023788322096, + "grad_norm": 2.5405871868133545, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8868333101272583, + "num_tokens": 501512909.0, + "step": 12918 + }, + { + "epoch": 1.6434295891108002, + "grad_norm": 2.64080810546875, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8763514757156372, + "num_tokens": 501553023.0, + "step": 12919 + }, + { + "epoch": 1.6435567993893907, + "grad_norm": 2.5698440074920654, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8825148940086365, + "num_tokens": 501599009.0, + "step": 12920 + }, + { + "epoch": 1.6436840096679812, + "grad_norm": 2.5421907901763916, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.892548143863678, + "num_tokens": 501638473.0, + "step": 12921 + }, + { + "epoch": 1.6438112199465715, + "grad_norm": 2.727914333343506, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8882707953453064, + "num_tokens": 501676228.0, + "step": 12922 + }, + { + "epoch": 1.643938430225162, + "grad_norm": 2.6826281547546387, + "learning_rate": 1e-06, + "loss": 0.3365, + "mean_token_accuracy": 0.8790715336799622, + "num_tokens": 501718657.0, + "step": 12923 + }, + { + "epoch": 1.6440656405037526, + "grad_norm": 2.631659984588623, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.9003141522407532, + "num_tokens": 501759696.0, + "step": 12924 + }, + { + "epoch": 1.6441928507823431, + "grad_norm": 2.658097982406616, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.89406418800354, + "num_tokens": 501798172.0, + "step": 12925 + }, + { + "epoch": 1.6443200610609336, + "grad_norm": 2.7622385025024414, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8758306503295898, + "num_tokens": 501839199.0, + "step": 12926 + }, + { + "epoch": 1.6444472713395242, + "grad_norm": 2.8395917415618896, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8863115310668945, + "num_tokens": 501872393.0, + "step": 12927 + }, + { + "epoch": 1.6445744816181147, + "grad_norm": 2.5532732009887695, + "learning_rate": 1e-06, + "loss": 0.349, + "mean_token_accuracy": 0.8780690431594849, + "num_tokens": 501916040.0, + "step": 12928 + }, + { + "epoch": 1.6447016918967052, + "grad_norm": 2.5631446838378906, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8959431648254395, + "num_tokens": 501954075.0, + "step": 12929 + }, + { + "epoch": 1.6448289021752958, + "grad_norm": 2.824361801147461, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8780592083930969, + "num_tokens": 501989241.0, + "step": 12930 + }, + { + "epoch": 1.6449561124538863, + "grad_norm": 2.662384510040283, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8924538493156433, + "num_tokens": 502026107.0, + "step": 12931 + }, + { + "epoch": 1.6450833227324768, + "grad_norm": 2.8970885276794434, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.880435049533844, + "num_tokens": 502065140.0, + "step": 12932 + }, + { + "epoch": 1.6452105330110673, + "grad_norm": 2.360668897628784, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9025924801826477, + "num_tokens": 502105559.0, + "step": 12933 + }, + { + "epoch": 1.6453377432896579, + "grad_norm": 2.7101199626922607, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8909106850624084, + "num_tokens": 502143724.0, + "step": 12934 + }, + { + "epoch": 1.6454649535682484, + "grad_norm": 2.7779245376586914, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8819891214370728, + "num_tokens": 502180828.0, + "step": 12935 + }, + { + "epoch": 1.645592163846839, + "grad_norm": 2.703568696975708, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8786388635635376, + "num_tokens": 502222288.0, + "step": 12936 + }, + { + "epoch": 1.6457193741254295, + "grad_norm": 2.7121644020080566, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8901287317276001, + "num_tokens": 502257629.0, + "step": 12937 + }, + { + "epoch": 1.64584658440402, + "grad_norm": 2.46718692779541, + "learning_rate": 1e-06, + "loss": 0.3332, + "mean_token_accuracy": 0.8809025883674622, + "num_tokens": 502303265.0, + "step": 12938 + }, + { + "epoch": 1.6459737946826105, + "grad_norm": 3.066094398498535, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.886884868144989, + "num_tokens": 502333566.0, + "step": 12939 + }, + { + "epoch": 1.6461010049612008, + "grad_norm": 2.584533452987671, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8886010646820068, + "num_tokens": 502375063.0, + "step": 12940 + }, + { + "epoch": 1.6462282152397913, + "grad_norm": 2.436105728149414, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8864046335220337, + "num_tokens": 502418346.0, + "step": 12941 + }, + { + "epoch": 1.6463554255183819, + "grad_norm": 2.8478848934173584, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8871004581451416, + "num_tokens": 502450558.0, + "step": 12942 + }, + { + "epoch": 1.6464826357969724, + "grad_norm": 2.4395980834960938, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8894412517547607, + "num_tokens": 502493693.0, + "step": 12943 + }, + { + "epoch": 1.646609846075563, + "grad_norm": 2.725390672683716, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.893172025680542, + "num_tokens": 502528810.0, + "step": 12944 + }, + { + "epoch": 1.6467370563541535, + "grad_norm": 2.5898900032043457, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9039283990859985, + "num_tokens": 502564402.0, + "step": 12945 + }, + { + "epoch": 1.6468642666327438, + "grad_norm": 2.6330060958862305, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8920058012008667, + "num_tokens": 502603030.0, + "step": 12946 + }, + { + "epoch": 1.6469914769113343, + "grad_norm": 2.558271646499634, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8894850611686707, + "num_tokens": 502644736.0, + "step": 12947 + }, + { + "epoch": 1.6471186871899248, + "grad_norm": 2.694183349609375, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8918948769569397, + "num_tokens": 502679827.0, + "step": 12948 + }, + { + "epoch": 1.6472458974685154, + "grad_norm": 2.612518548965454, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8774904012680054, + "num_tokens": 502721062.0, + "step": 12949 + }, + { + "epoch": 1.6473731077471059, + "grad_norm": 2.7734389305114746, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8943066596984863, + "num_tokens": 502756528.0, + "step": 12950 + }, + { + "epoch": 1.6475003180256964, + "grad_norm": 2.5413920879364014, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8865812420845032, + "num_tokens": 502798301.0, + "step": 12951 + }, + { + "epoch": 1.647627528304287, + "grad_norm": 2.5102672576904297, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8871208429336548, + "num_tokens": 502841897.0, + "step": 12952 + }, + { + "epoch": 1.6477547385828775, + "grad_norm": 2.536714792251587, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8874077796936035, + "num_tokens": 502884232.0, + "step": 12953 + }, + { + "epoch": 1.647881948861468, + "grad_norm": 2.5735561847686768, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8902661800384521, + "num_tokens": 502925945.0, + "step": 12954 + }, + { + "epoch": 1.6480091591400585, + "grad_norm": 2.3675646781921387, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.896856427192688, + "num_tokens": 502970151.0, + "step": 12955 + }, + { + "epoch": 1.648136369418649, + "grad_norm": 2.5690860748291016, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8885183334350586, + "num_tokens": 503010188.0, + "step": 12956 + }, + { + "epoch": 1.6482635796972396, + "grad_norm": 2.5679023265838623, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.892290472984314, + "num_tokens": 503048052.0, + "step": 12957 + }, + { + "epoch": 1.64839078997583, + "grad_norm": 2.461534261703491, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8811803460121155, + "num_tokens": 503094839.0, + "step": 12958 + }, + { + "epoch": 1.6485180002544206, + "grad_norm": 2.6288130283355713, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8912328481674194, + "num_tokens": 503137363.0, + "step": 12959 + }, + { + "epoch": 1.6486452105330112, + "grad_norm": 2.6115362644195557, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8884178400039673, + "num_tokens": 503179452.0, + "step": 12960 + }, + { + "epoch": 1.6487724208116017, + "grad_norm": 2.534376621246338, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8955321311950684, + "num_tokens": 503219154.0, + "step": 12961 + }, + { + "epoch": 1.6488996310901922, + "grad_norm": 2.6960842609405518, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8927643299102783, + "num_tokens": 503254547.0, + "step": 12962 + }, + { + "epoch": 1.6490268413687827, + "grad_norm": 2.4186606407165527, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8872504830360413, + "num_tokens": 503296827.0, + "step": 12963 + }, + { + "epoch": 1.649154051647373, + "grad_norm": 3.0291996002197266, + "learning_rate": 1e-06, + "loss": 0.3264, + "mean_token_accuracy": 0.8837000131607056, + "num_tokens": 503328057.0, + "step": 12964 + }, + { + "epoch": 1.6492812619259636, + "grad_norm": 2.3988749980926514, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8947046995162964, + "num_tokens": 503372299.0, + "step": 12965 + }, + { + "epoch": 1.649408472204554, + "grad_norm": 2.841541051864624, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8975013494491577, + "num_tokens": 503408774.0, + "step": 12966 + }, + { + "epoch": 1.6495356824831446, + "grad_norm": 2.7512972354888916, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8860936164855957, + "num_tokens": 503446155.0, + "step": 12967 + }, + { + "epoch": 1.6496628927617352, + "grad_norm": 3.020514488220215, + "learning_rate": 1e-06, + "loss": 0.3469, + "mean_token_accuracy": 0.8833981156349182, + "num_tokens": 503482878.0, + "step": 12968 + }, + { + "epoch": 1.6497901030403257, + "grad_norm": 2.535449981689453, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8941658139228821, + "num_tokens": 503521924.0, + "step": 12969 + }, + { + "epoch": 1.6499173133189162, + "grad_norm": 2.4790356159210205, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8804744482040405, + "num_tokens": 503566496.0, + "step": 12970 + }, + { + "epoch": 1.6500445235975065, + "grad_norm": 2.7166824340820312, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8905956149101257, + "num_tokens": 503597445.0, + "step": 12971 + }, + { + "epoch": 1.650171733876097, + "grad_norm": 2.5523812770843506, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8989567756652832, + "num_tokens": 503634582.0, + "step": 12972 + }, + { + "epoch": 1.6502989441546876, + "grad_norm": 2.8288028240203857, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8908872008323669, + "num_tokens": 503668113.0, + "step": 12973 + }, + { + "epoch": 1.6504261544332781, + "grad_norm": 2.77624773979187, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8895905017852783, + "num_tokens": 503703993.0, + "step": 12974 + }, + { + "epoch": 1.6505533647118686, + "grad_norm": 2.4991931915283203, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9010416269302368, + "num_tokens": 503742063.0, + "step": 12975 + }, + { + "epoch": 1.6506805749904592, + "grad_norm": 2.775474786758423, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8791176080703735, + "num_tokens": 503778295.0, + "step": 12976 + }, + { + "epoch": 1.6508077852690497, + "grad_norm": 2.6277358531951904, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8890156149864197, + "num_tokens": 503817649.0, + "step": 12977 + }, + { + "epoch": 1.6509349955476402, + "grad_norm": 2.7505743503570557, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8836776614189148, + "num_tokens": 503854356.0, + "step": 12978 + }, + { + "epoch": 1.6510622058262308, + "grad_norm": 2.760545015335083, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8851317167282104, + "num_tokens": 503891133.0, + "step": 12979 + }, + { + "epoch": 1.6511894161048213, + "grad_norm": 2.686767101287842, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8868193626403809, + "num_tokens": 503932854.0, + "step": 12980 + }, + { + "epoch": 1.6513166263834118, + "grad_norm": 2.7230710983276367, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8911718726158142, + "num_tokens": 503971519.0, + "step": 12981 + }, + { + "epoch": 1.6514438366620023, + "grad_norm": 2.5041580200195312, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8945415019989014, + "num_tokens": 504010942.0, + "step": 12982 + }, + { + "epoch": 1.6515710469405929, + "grad_norm": 2.510206460952759, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8925955295562744, + "num_tokens": 504055277.0, + "step": 12983 + }, + { + "epoch": 1.6516982572191834, + "grad_norm": 2.598330497741699, + "learning_rate": 1e-06, + "loss": 0.3341, + "mean_token_accuracy": 0.8831284046173096, + "num_tokens": 504096975.0, + "step": 12984 + }, + { + "epoch": 1.651825467497774, + "grad_norm": 2.521198272705078, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8949131965637207, + "num_tokens": 504138485.0, + "step": 12985 + }, + { + "epoch": 1.6519526777763645, + "grad_norm": 2.53393292427063, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8968613147735596, + "num_tokens": 504176309.0, + "step": 12986 + }, + { + "epoch": 1.652079888054955, + "grad_norm": 2.5815515518188477, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8855628967285156, + "num_tokens": 504219480.0, + "step": 12987 + }, + { + "epoch": 1.6522070983335455, + "grad_norm": 2.812333345413208, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8920465707778931, + "num_tokens": 504255903.0, + "step": 12988 + }, + { + "epoch": 1.6523343086121358, + "grad_norm": 2.8287734985351562, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8885063529014587, + "num_tokens": 504289638.0, + "step": 12989 + }, + { + "epoch": 1.6524615188907263, + "grad_norm": 2.5016119480133057, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8748260140419006, + "num_tokens": 504337004.0, + "step": 12990 + }, + { + "epoch": 1.6525887291693169, + "grad_norm": 2.3661701679229736, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8810818195343018, + "num_tokens": 504383526.0, + "step": 12991 + }, + { + "epoch": 1.6527159394479074, + "grad_norm": 2.901984453201294, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8906137943267822, + "num_tokens": 504416937.0, + "step": 12992 + }, + { + "epoch": 1.652843149726498, + "grad_norm": 2.9126827716827393, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9005051255226135, + "num_tokens": 504460365.0, + "step": 12993 + }, + { + "epoch": 1.6529703600050885, + "grad_norm": 2.796220302581787, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8961148858070374, + "num_tokens": 504495090.0, + "step": 12994 + }, + { + "epoch": 1.6530975702836788, + "grad_norm": 2.5210378170013428, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8935588598251343, + "num_tokens": 504535646.0, + "step": 12995 + }, + { + "epoch": 1.6532247805622693, + "grad_norm": 2.5539674758911133, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.889989972114563, + "num_tokens": 504576361.0, + "step": 12996 + }, + { + "epoch": 1.6533519908408598, + "grad_norm": 2.6812245845794678, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8817442655563354, + "num_tokens": 504614127.0, + "step": 12997 + }, + { + "epoch": 1.6534792011194503, + "grad_norm": 2.6262879371643066, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8981002569198608, + "num_tokens": 504651423.0, + "step": 12998 + }, + { + "epoch": 1.6536064113980409, + "grad_norm": 2.469909191131592, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8970557451248169, + "num_tokens": 504691322.0, + "step": 12999 + }, + { + "epoch": 1.6537336216766314, + "grad_norm": 2.5581905841827393, + "learning_rate": 1e-06, + "loss": 0.3334, + "mean_token_accuracy": 0.8833733201026917, + "num_tokens": 504733348.0, + "step": 13000 + }, + { + "epoch": 1.653860831955222, + "grad_norm": 2.6222636699676514, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.898417592048645, + "num_tokens": 504771925.0, + "step": 13001 + }, + { + "epoch": 1.6539880422338125, + "grad_norm": 2.512404680252075, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.8983496427536011, + "num_tokens": 504810723.0, + "step": 13002 + }, + { + "epoch": 1.654115252512403, + "grad_norm": 2.6225321292877197, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.885797381401062, + "num_tokens": 504853105.0, + "step": 13003 + }, + { + "epoch": 1.6542424627909935, + "grad_norm": 2.643817186355591, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8885527849197388, + "num_tokens": 504890972.0, + "step": 13004 + }, + { + "epoch": 1.654369673069584, + "grad_norm": 2.773738384246826, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9051276445388794, + "num_tokens": 504922700.0, + "step": 13005 + }, + { + "epoch": 1.6544968833481746, + "grad_norm": 2.660845994949341, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8882423639297485, + "num_tokens": 504960136.0, + "step": 13006 + }, + { + "epoch": 1.654624093626765, + "grad_norm": 2.3576958179473877, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8979417681694031, + "num_tokens": 505001870.0, + "step": 13007 + }, + { + "epoch": 1.6547513039053556, + "grad_norm": 2.486297607421875, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9062947034835815, + "num_tokens": 505040643.0, + "step": 13008 + }, + { + "epoch": 1.6548785141839462, + "grad_norm": 2.702202081680298, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8934217095375061, + "num_tokens": 505079697.0, + "step": 13009 + }, + { + "epoch": 1.6550057244625367, + "grad_norm": 2.797396659851074, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8966353535652161, + "num_tokens": 505114838.0, + "step": 13010 + }, + { + "epoch": 1.6551329347411272, + "grad_norm": 2.5190932750701904, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8972861766815186, + "num_tokens": 505154112.0, + "step": 13011 + }, + { + "epoch": 1.6552601450197177, + "grad_norm": 2.5875046253204346, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8871648907661438, + "num_tokens": 505195630.0, + "step": 13012 + }, + { + "epoch": 1.655387355298308, + "grad_norm": 2.496795654296875, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.907108724117279, + "num_tokens": 505232717.0, + "step": 13013 + }, + { + "epoch": 1.6555145655768986, + "grad_norm": 2.5994856357574463, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8914549350738525, + "num_tokens": 505273719.0, + "step": 13014 + }, + { + "epoch": 1.655641775855489, + "grad_norm": 2.7302448749542236, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8830348253250122, + "num_tokens": 505309889.0, + "step": 13015 + }, + { + "epoch": 1.6557689861340796, + "grad_norm": 2.6590416431427, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8943299055099487, + "num_tokens": 505347328.0, + "step": 13016 + }, + { + "epoch": 1.6558961964126702, + "grad_norm": 2.5483057498931885, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.900780439376831, + "num_tokens": 505383205.0, + "step": 13017 + }, + { + "epoch": 1.6560234066912607, + "grad_norm": 2.470612049102783, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8996511697769165, + "num_tokens": 505423419.0, + "step": 13018 + }, + { + "epoch": 1.6561506169698512, + "grad_norm": 2.5656158924102783, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8952297568321228, + "num_tokens": 505461399.0, + "step": 13019 + }, + { + "epoch": 1.6562778272484415, + "grad_norm": 2.415860176086426, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.893540620803833, + "num_tokens": 505504264.0, + "step": 13020 + }, + { + "epoch": 1.656405037527032, + "grad_norm": 3.2685396671295166, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8902840614318848, + "num_tokens": 505542442.0, + "step": 13021 + }, + { + "epoch": 1.6565322478056226, + "grad_norm": 2.5079779624938965, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9063734412193298, + "num_tokens": 505581582.0, + "step": 13022 + }, + { + "epoch": 1.656659458084213, + "grad_norm": 2.5742392539978027, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8852511644363403, + "num_tokens": 505623455.0, + "step": 13023 + }, + { + "epoch": 1.6567866683628036, + "grad_norm": 2.6026206016540527, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8986012935638428, + "num_tokens": 505660911.0, + "step": 13024 + }, + { + "epoch": 1.6569138786413942, + "grad_norm": 2.580904245376587, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8966504335403442, + "num_tokens": 505701092.0, + "step": 13025 + }, + { + "epoch": 1.6570410889199847, + "grad_norm": 2.5393214225769043, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8934749364852905, + "num_tokens": 505742427.0, + "step": 13026 + }, + { + "epoch": 1.6571682991985752, + "grad_norm": 2.4090123176574707, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8975791931152344, + "num_tokens": 505785408.0, + "step": 13027 + }, + { + "epoch": 1.6572955094771658, + "grad_norm": 2.6612467765808105, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8931717872619629, + "num_tokens": 505820705.0, + "step": 13028 + }, + { + "epoch": 1.6574227197557563, + "grad_norm": 2.452350616455078, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8834922909736633, + "num_tokens": 505864648.0, + "step": 13029 + }, + { + "epoch": 1.6575499300343468, + "grad_norm": 2.5867650508880615, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8941252827644348, + "num_tokens": 505902637.0, + "step": 13030 + }, + { + "epoch": 1.6576771403129373, + "grad_norm": 2.6597697734832764, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8980588912963867, + "num_tokens": 505937401.0, + "step": 13031 + }, + { + "epoch": 1.6578043505915279, + "grad_norm": 2.7656805515289307, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8942215442657471, + "num_tokens": 505973889.0, + "step": 13032 + }, + { + "epoch": 1.6579315608701184, + "grad_norm": 2.830822706222534, + "learning_rate": 1e-06, + "loss": 0.3404, + "mean_token_accuracy": 0.8788342475891113, + "num_tokens": 506009861.0, + "step": 13033 + }, + { + "epoch": 1.658058771148709, + "grad_norm": 2.8132824897766113, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8904470205307007, + "num_tokens": 506043643.0, + "step": 13034 + }, + { + "epoch": 1.6581859814272994, + "grad_norm": 2.742044448852539, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8793375492095947, + "num_tokens": 506084448.0, + "step": 13035 + }, + { + "epoch": 1.65831319170589, + "grad_norm": 2.915177822113037, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8873594403266907, + "num_tokens": 506116429.0, + "step": 13036 + }, + { + "epoch": 1.6584404019844805, + "grad_norm": 2.597277879714966, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8904849886894226, + "num_tokens": 506154104.0, + "step": 13037 + }, + { + "epoch": 1.6585676122630708, + "grad_norm": 2.8503615856170654, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8802220821380615, + "num_tokens": 506191221.0, + "step": 13038 + }, + { + "epoch": 1.6586948225416613, + "grad_norm": 2.6141138076782227, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8921108245849609, + "num_tokens": 506229390.0, + "step": 13039 + }, + { + "epoch": 1.6588220328202519, + "grad_norm": 2.4099433422088623, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9066542387008667, + "num_tokens": 506268993.0, + "step": 13040 + }, + { + "epoch": 1.6589492430988424, + "grad_norm": 2.808351755142212, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8797698020935059, + "num_tokens": 506308981.0, + "step": 13041 + }, + { + "epoch": 1.659076453377433, + "grad_norm": 2.6315724849700928, + "learning_rate": 1e-06, + "loss": 0.3256, + "mean_token_accuracy": 0.8835006952285767, + "num_tokens": 506349239.0, + "step": 13042 + }, + { + "epoch": 1.6592036636560235, + "grad_norm": 2.584012269973755, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8926489949226379, + "num_tokens": 506387100.0, + "step": 13043 + }, + { + "epoch": 1.6593308739346138, + "grad_norm": 2.8293514251708984, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8832587003707886, + "num_tokens": 506423873.0, + "step": 13044 + }, + { + "epoch": 1.6594580842132043, + "grad_norm": 2.681352138519287, + "learning_rate": 1e-06, + "loss": 0.3428, + "mean_token_accuracy": 0.8794683814048767, + "num_tokens": 506467655.0, + "step": 13045 + }, + { + "epoch": 1.6595852944917948, + "grad_norm": 2.924992084503174, + "learning_rate": 1e-06, + "loss": 0.3337, + "mean_token_accuracy": 0.8813570737838745, + "num_tokens": 506502259.0, + "step": 13046 + }, + { + "epoch": 1.6597125047703853, + "grad_norm": 2.9206769466400146, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8774129152297974, + "num_tokens": 506537286.0, + "step": 13047 + }, + { + "epoch": 1.6598397150489759, + "grad_norm": 2.661795139312744, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8838528394699097, + "num_tokens": 506577252.0, + "step": 13048 + }, + { + "epoch": 1.6599669253275664, + "grad_norm": 2.8902289867401123, + "learning_rate": 1e-06, + "loss": 0.3593, + "mean_token_accuracy": 0.8722472190856934, + "num_tokens": 506617039.0, + "step": 13049 + }, + { + "epoch": 1.660094135606157, + "grad_norm": 2.5987441539764404, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8803236484527588, + "num_tokens": 506658204.0, + "step": 13050 + }, + { + "epoch": 1.6602213458847475, + "grad_norm": 2.3712193965911865, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9014800786972046, + "num_tokens": 506696465.0, + "step": 13051 + }, + { + "epoch": 1.660348556163338, + "grad_norm": 2.653646469116211, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8984508514404297, + "num_tokens": 506732273.0, + "step": 13052 + }, + { + "epoch": 1.6604757664419285, + "grad_norm": 2.4504597187042236, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8887979984283447, + "num_tokens": 506775518.0, + "step": 13053 + }, + { + "epoch": 1.660602976720519, + "grad_norm": 2.6105244159698486, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.890417218208313, + "num_tokens": 506813825.0, + "step": 13054 + }, + { + "epoch": 1.6607301869991096, + "grad_norm": 2.5008111000061035, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9055963754653931, + "num_tokens": 506852514.0, + "step": 13055 + }, + { + "epoch": 1.6608573972777, + "grad_norm": 2.8885250091552734, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8846497535705566, + "num_tokens": 506885699.0, + "step": 13056 + }, + { + "epoch": 1.6609846075562906, + "grad_norm": 2.5980031490325928, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.889880895614624, + "num_tokens": 506924055.0, + "step": 13057 + }, + { + "epoch": 1.6611118178348812, + "grad_norm": 2.6338987350463867, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.88759446144104, + "num_tokens": 506964162.0, + "step": 13058 + }, + { + "epoch": 1.6612390281134717, + "grad_norm": 2.7131028175354004, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8853703141212463, + "num_tokens": 507004955.0, + "step": 13059 + }, + { + "epoch": 1.6613662383920622, + "grad_norm": 2.813652276992798, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8910869359970093, + "num_tokens": 507039950.0, + "step": 13060 + }, + { + "epoch": 1.6614934486706527, + "grad_norm": 2.575241804122925, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8894343376159668, + "num_tokens": 507079080.0, + "step": 13061 + }, + { + "epoch": 1.661620658949243, + "grad_norm": 2.6065940856933594, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8883329033851624, + "num_tokens": 507116789.0, + "step": 13062 + }, + { + "epoch": 1.6617478692278336, + "grad_norm": 2.8026344776153564, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8720725774765015, + "num_tokens": 507158944.0, + "step": 13063 + }, + { + "epoch": 1.661875079506424, + "grad_norm": 2.527660846710205, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8936580419540405, + "num_tokens": 507199640.0, + "step": 13064 + }, + { + "epoch": 1.6620022897850146, + "grad_norm": 2.6509432792663574, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8995718359947205, + "num_tokens": 507235832.0, + "step": 13065 + }, + { + "epoch": 1.6621295000636052, + "grad_norm": 2.771559953689575, + "learning_rate": 1e-06, + "loss": 0.3658, + "mean_token_accuracy": 0.8712062835693359, + "num_tokens": 507274326.0, + "step": 13066 + }, + { + "epoch": 1.6622567103421957, + "grad_norm": 2.5094716548919678, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8958180546760559, + "num_tokens": 507314591.0, + "step": 13067 + }, + { + "epoch": 1.662383920620786, + "grad_norm": 2.6056761741638184, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8967717289924622, + "num_tokens": 507352229.0, + "step": 13068 + }, + { + "epoch": 1.6625111308993765, + "grad_norm": 2.8735010623931885, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8939635753631592, + "num_tokens": 507395544.0, + "step": 13069 + }, + { + "epoch": 1.662638341177967, + "grad_norm": 2.633613348007202, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9029673337936401, + "num_tokens": 507432047.0, + "step": 13070 + }, + { + "epoch": 1.6627655514565576, + "grad_norm": 2.898134469985962, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8869754672050476, + "num_tokens": 507468595.0, + "step": 13071 + }, + { + "epoch": 1.662892761735148, + "grad_norm": 2.6895554065704346, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8930891752243042, + "num_tokens": 507506330.0, + "step": 13072 + }, + { + "epoch": 1.6630199720137386, + "grad_norm": 2.8308231830596924, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8825478553771973, + "num_tokens": 507544064.0, + "step": 13073 + }, + { + "epoch": 1.6631471822923292, + "grad_norm": 2.4896976947784424, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.8994020223617554, + "num_tokens": 507584594.0, + "step": 13074 + }, + { + "epoch": 1.6632743925709197, + "grad_norm": 2.6608452796936035, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8782087564468384, + "num_tokens": 507627099.0, + "step": 13075 + }, + { + "epoch": 1.6634016028495102, + "grad_norm": 2.7131025791168213, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8880763649940491, + "num_tokens": 507670511.0, + "step": 13076 + }, + { + "epoch": 1.6635288131281007, + "grad_norm": 2.6190638542175293, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8904430866241455, + "num_tokens": 507711875.0, + "step": 13077 + }, + { + "epoch": 1.6636560234066913, + "grad_norm": 2.6634974479675293, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8857659101486206, + "num_tokens": 507752394.0, + "step": 13078 + }, + { + "epoch": 1.6637832336852818, + "grad_norm": 2.783794403076172, + "learning_rate": 1e-06, + "loss": 0.3517, + "mean_token_accuracy": 0.8729420304298401, + "num_tokens": 507792977.0, + "step": 13079 + }, + { + "epoch": 1.6639104439638723, + "grad_norm": 2.623802900314331, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8938541412353516, + "num_tokens": 507830430.0, + "step": 13080 + }, + { + "epoch": 1.6640376542424629, + "grad_norm": 2.398714542388916, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9000704288482666, + "num_tokens": 507873577.0, + "step": 13081 + }, + { + "epoch": 1.6641648645210534, + "grad_norm": 2.366457223892212, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8939046859741211, + "num_tokens": 507916859.0, + "step": 13082 + }, + { + "epoch": 1.664292074799644, + "grad_norm": 2.7055399417877197, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8924026489257812, + "num_tokens": 507951568.0, + "step": 13083 + }, + { + "epoch": 1.6644192850782344, + "grad_norm": 2.701979398727417, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8746889233589172, + "num_tokens": 507992694.0, + "step": 13084 + }, + { + "epoch": 1.664546495356825, + "grad_norm": 2.7707691192626953, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8926202654838562, + "num_tokens": 508028551.0, + "step": 13085 + }, + { + "epoch": 1.6646737056354155, + "grad_norm": 2.4819085597991943, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8929286599159241, + "num_tokens": 508070249.0, + "step": 13086 + }, + { + "epoch": 1.6648009159140058, + "grad_norm": 2.349579334259033, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.8986616134643555, + "num_tokens": 508112182.0, + "step": 13087 + }, + { + "epoch": 1.6649281261925963, + "grad_norm": 3.042393684387207, + "learning_rate": 1e-06, + "loss": 0.374, + "mean_token_accuracy": 0.8665204048156738, + "num_tokens": 508148924.0, + "step": 13088 + }, + { + "epoch": 1.6650553364711869, + "grad_norm": 2.3553168773651123, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9021459221839905, + "num_tokens": 508190756.0, + "step": 13089 + }, + { + "epoch": 1.6651825467497774, + "grad_norm": 2.5431838035583496, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8866877555847168, + "num_tokens": 508235418.0, + "step": 13090 + }, + { + "epoch": 1.665309757028368, + "grad_norm": 2.8101396560668945, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8938087224960327, + "num_tokens": 508271958.0, + "step": 13091 + }, + { + "epoch": 1.6654369673069584, + "grad_norm": 2.75872540473938, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8831198215484619, + "num_tokens": 508311435.0, + "step": 13092 + }, + { + "epoch": 1.6655641775855488, + "grad_norm": 2.704533815383911, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8921875357627869, + "num_tokens": 508349255.0, + "step": 13093 + }, + { + "epoch": 1.6656913878641393, + "grad_norm": 2.7788100242614746, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8742369413375854, + "num_tokens": 508388358.0, + "step": 13094 + }, + { + "epoch": 1.6658185981427298, + "grad_norm": 2.6979477405548096, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9056131839752197, + "num_tokens": 508419088.0, + "step": 13095 + }, + { + "epoch": 1.6659458084213203, + "grad_norm": 3.157266855239868, + "learning_rate": 1e-06, + "loss": 0.3396, + "mean_token_accuracy": 0.8832112550735474, + "num_tokens": 508451310.0, + "step": 13096 + }, + { + "epoch": 1.6660730186999109, + "grad_norm": 2.5877418518066406, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8918828368186951, + "num_tokens": 508489507.0, + "step": 13097 + }, + { + "epoch": 1.6662002289785014, + "grad_norm": 3.1497135162353516, + "learning_rate": 1e-06, + "loss": 0.3696, + "mean_token_accuracy": 0.8689534068107605, + "num_tokens": 508523093.0, + "step": 13098 + }, + { + "epoch": 1.666327439257092, + "grad_norm": 2.5915322303771973, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8867838382720947, + "num_tokens": 508565332.0, + "step": 13099 + }, + { + "epoch": 1.6664546495356825, + "grad_norm": 2.9643797874450684, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.879862904548645, + "num_tokens": 508599779.0, + "step": 13100 + }, + { + "epoch": 1.666581859814273, + "grad_norm": 2.640648126602173, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8919498920440674, + "num_tokens": 508637623.0, + "step": 13101 + }, + { + "epoch": 1.6667090700928635, + "grad_norm": 2.5115010738372803, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8955411911010742, + "num_tokens": 508679227.0, + "step": 13102 + }, + { + "epoch": 1.666836280371454, + "grad_norm": 2.4778683185577393, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8959340453147888, + "num_tokens": 508720379.0, + "step": 13103 + }, + { + "epoch": 1.6669634906500446, + "grad_norm": 2.477229356765747, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8912202715873718, + "num_tokens": 508763316.0, + "step": 13104 + }, + { + "epoch": 1.667090700928635, + "grad_norm": 2.592038154602051, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8933416604995728, + "num_tokens": 508804598.0, + "step": 13105 + }, + { + "epoch": 1.6672179112072256, + "grad_norm": 2.4078307151794434, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.898943305015564, + "num_tokens": 508846773.0, + "step": 13106 + }, + { + "epoch": 1.6673451214858162, + "grad_norm": 2.5917181968688965, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8825914263725281, + "num_tokens": 508888841.0, + "step": 13107 + }, + { + "epoch": 1.6674723317644067, + "grad_norm": 2.6552515029907227, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8903781771659851, + "num_tokens": 508926033.0, + "step": 13108 + }, + { + "epoch": 1.6675995420429972, + "grad_norm": 2.665626049041748, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8945543169975281, + "num_tokens": 508963866.0, + "step": 13109 + }, + { + "epoch": 1.6677267523215877, + "grad_norm": 2.544752836227417, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9009180068969727, + "num_tokens": 509001867.0, + "step": 13110 + }, + { + "epoch": 1.667853962600178, + "grad_norm": 2.474182605743408, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.897169291973114, + "num_tokens": 509043244.0, + "step": 13111 + }, + { + "epoch": 1.6679811728787686, + "grad_norm": 2.544691324234009, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8979501724243164, + "num_tokens": 509084356.0, + "step": 13112 + }, + { + "epoch": 1.668108383157359, + "grad_norm": 2.543125629425049, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8979023694992065, + "num_tokens": 509124459.0, + "step": 13113 + }, + { + "epoch": 1.6682355934359496, + "grad_norm": 2.6817703247070312, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8996753692626953, + "num_tokens": 509161202.0, + "step": 13114 + }, + { + "epoch": 1.6683628037145402, + "grad_norm": 2.7081236839294434, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8945140838623047, + "num_tokens": 509197829.0, + "step": 13115 + }, + { + "epoch": 1.6684900139931307, + "grad_norm": 2.8867673873901367, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.881496250629425, + "num_tokens": 509231742.0, + "step": 13116 + }, + { + "epoch": 1.668617224271721, + "grad_norm": 2.7516531944274902, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8809860944747925, + "num_tokens": 509271010.0, + "step": 13117 + }, + { + "epoch": 1.6687444345503115, + "grad_norm": 2.5768237113952637, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.881452739238739, + "num_tokens": 509314547.0, + "step": 13118 + }, + { + "epoch": 1.668871644828902, + "grad_norm": 2.4469830989837646, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9028027057647705, + "num_tokens": 509355743.0, + "step": 13119 + }, + { + "epoch": 1.6689988551074926, + "grad_norm": 2.654581308364868, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8894834518432617, + "num_tokens": 509393003.0, + "step": 13120 + }, + { + "epoch": 1.669126065386083, + "grad_norm": 2.6689999103546143, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8845983743667603, + "num_tokens": 509431556.0, + "step": 13121 + }, + { + "epoch": 1.6692532756646736, + "grad_norm": 2.9195146560668945, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8891152143478394, + "num_tokens": 509466723.0, + "step": 13122 + }, + { + "epoch": 1.6693804859432642, + "grad_norm": 2.535646677017212, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8926241397857666, + "num_tokens": 509503917.0, + "step": 13123 + }, + { + "epoch": 1.6695076962218547, + "grad_norm": 2.546104907989502, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8872788548469543, + "num_tokens": 509543719.0, + "step": 13124 + }, + { + "epoch": 1.6696349065004452, + "grad_norm": 2.4955697059631348, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8826829791069031, + "num_tokens": 509584547.0, + "step": 13125 + }, + { + "epoch": 1.6697621167790357, + "grad_norm": 2.8459665775299072, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8910009860992432, + "num_tokens": 509619435.0, + "step": 13126 + }, + { + "epoch": 1.6698893270576263, + "grad_norm": 2.6250507831573486, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.884469747543335, + "num_tokens": 509665310.0, + "step": 13127 + }, + { + "epoch": 1.6700165373362168, + "grad_norm": 2.7315680980682373, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8840341567993164, + "num_tokens": 509702654.0, + "step": 13128 + }, + { + "epoch": 1.6701437476148073, + "grad_norm": 2.716357707977295, + "learning_rate": 1e-06, + "loss": 0.3664, + "mean_token_accuracy": 0.8733364343643188, + "num_tokens": 509742878.0, + "step": 13129 + }, + { + "epoch": 1.6702709578933979, + "grad_norm": 2.632607936859131, + "learning_rate": 1e-06, + "loss": 0.3483, + "mean_token_accuracy": 0.8742991089820862, + "num_tokens": 509783670.0, + "step": 13130 + }, + { + "epoch": 1.6703981681719884, + "grad_norm": 2.6013102531433105, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8977500200271606, + "num_tokens": 509823344.0, + "step": 13131 + }, + { + "epoch": 1.670525378450579, + "grad_norm": 2.562831401824951, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8935495615005493, + "num_tokens": 509861752.0, + "step": 13132 + }, + { + "epoch": 1.6706525887291694, + "grad_norm": 2.661348581314087, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8908399343490601, + "num_tokens": 509896930.0, + "step": 13133 + }, + { + "epoch": 1.67077979900776, + "grad_norm": 2.5693047046661377, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8995459675788879, + "num_tokens": 509934002.0, + "step": 13134 + }, + { + "epoch": 1.6709070092863505, + "grad_norm": 2.485342025756836, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9026502966880798, + "num_tokens": 509970533.0, + "step": 13135 + }, + { + "epoch": 1.6710342195649408, + "grad_norm": 2.4034066200256348, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8910113573074341, + "num_tokens": 510013264.0, + "step": 13136 + }, + { + "epoch": 1.6711614298435313, + "grad_norm": 2.632377862930298, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8933461904525757, + "num_tokens": 510051252.0, + "step": 13137 + }, + { + "epoch": 1.6712886401221219, + "grad_norm": 3.0107762813568115, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8844852447509766, + "num_tokens": 510083524.0, + "step": 13138 + }, + { + "epoch": 1.6714158504007124, + "grad_norm": 2.497598648071289, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8946735858917236, + "num_tokens": 510126396.0, + "step": 13139 + }, + { + "epoch": 1.671543060679303, + "grad_norm": 3.5900156497955322, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8887712955474854, + "num_tokens": 510167232.0, + "step": 13140 + }, + { + "epoch": 1.6716702709578934, + "grad_norm": 2.7024545669555664, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8923903107643127, + "num_tokens": 510202441.0, + "step": 13141 + }, + { + "epoch": 1.6717974812364838, + "grad_norm": 2.839188575744629, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8794434070587158, + "num_tokens": 510239191.0, + "step": 13142 + }, + { + "epoch": 1.6719246915150743, + "grad_norm": 2.9196224212646484, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8866152763366699, + "num_tokens": 510274463.0, + "step": 13143 + }, + { + "epoch": 1.6720519017936648, + "grad_norm": 2.5103886127471924, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8861538171768188, + "num_tokens": 510317276.0, + "step": 13144 + }, + { + "epoch": 1.6721791120722553, + "grad_norm": 2.4332997798919678, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9009345173835754, + "num_tokens": 510358593.0, + "step": 13145 + }, + { + "epoch": 1.6723063223508459, + "grad_norm": 2.6416471004486084, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.8798154592514038, + "num_tokens": 510400911.0, + "step": 13146 + }, + { + "epoch": 1.6724335326294364, + "grad_norm": 2.5013527870178223, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9006916284561157, + "num_tokens": 510437721.0, + "step": 13147 + }, + { + "epoch": 1.672560742908027, + "grad_norm": 2.5242984294891357, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8908979892730713, + "num_tokens": 510485373.0, + "step": 13148 + }, + { + "epoch": 1.6726879531866174, + "grad_norm": 2.9403867721557617, + "learning_rate": 1e-06, + "loss": 0.3711, + "mean_token_accuracy": 0.8687730431556702, + "num_tokens": 510522709.0, + "step": 13149 + }, + { + "epoch": 1.672815163465208, + "grad_norm": 2.839259147644043, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8838542103767395, + "num_tokens": 510557372.0, + "step": 13150 + }, + { + "epoch": 1.6729423737437985, + "grad_norm": 2.406280040740967, + "learning_rate": 1e-06, + "loss": 0.242, + "mean_token_accuracy": 0.9116728901863098, + "num_tokens": 510596560.0, + "step": 13151 + }, + { + "epoch": 1.673069584022389, + "grad_norm": 2.855116128921509, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8795721530914307, + "num_tokens": 510634032.0, + "step": 13152 + }, + { + "epoch": 1.6731967943009796, + "grad_norm": 2.830514907836914, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8886539936065674, + "num_tokens": 510670499.0, + "step": 13153 + }, + { + "epoch": 1.67332400457957, + "grad_norm": 2.618720769882202, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9000267386436462, + "num_tokens": 510705663.0, + "step": 13154 + }, + { + "epoch": 1.6734512148581606, + "grad_norm": 2.6924331188201904, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.890781044960022, + "num_tokens": 510744539.0, + "step": 13155 + }, + { + "epoch": 1.6735784251367511, + "grad_norm": 2.5882890224456787, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8851016759872437, + "num_tokens": 510785453.0, + "step": 13156 + }, + { + "epoch": 1.6737056354153417, + "grad_norm": 2.5730183124542236, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8952344655990601, + "num_tokens": 510824070.0, + "step": 13157 + }, + { + "epoch": 1.6738328456939322, + "grad_norm": 2.8469784259796143, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8924708962440491, + "num_tokens": 510858023.0, + "step": 13158 + }, + { + "epoch": 1.6739600559725227, + "grad_norm": 2.707826614379883, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8951790928840637, + "num_tokens": 510899431.0, + "step": 13159 + }, + { + "epoch": 1.674087266251113, + "grad_norm": 2.780144453048706, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8878326416015625, + "num_tokens": 510935438.0, + "step": 13160 + }, + { + "epoch": 1.6742144765297036, + "grad_norm": 2.872647762298584, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8951163291931152, + "num_tokens": 510968758.0, + "step": 13161 + }, + { + "epoch": 1.674341686808294, + "grad_norm": 2.500436305999756, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.893744170665741, + "num_tokens": 511007292.0, + "step": 13162 + }, + { + "epoch": 1.6744688970868846, + "grad_norm": 2.7260243892669678, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8808760643005371, + "num_tokens": 511046615.0, + "step": 13163 + }, + { + "epoch": 1.6745961073654752, + "grad_norm": 2.6466116905212402, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8853529691696167, + "num_tokens": 511085495.0, + "step": 13164 + }, + { + "epoch": 1.6747233176440657, + "grad_norm": 2.2993197441101074, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9135596752166748, + "num_tokens": 511124133.0, + "step": 13165 + }, + { + "epoch": 1.674850527922656, + "grad_norm": 2.510560989379883, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.8995710611343384, + "num_tokens": 511162632.0, + "step": 13166 + }, + { + "epoch": 1.6749777382012465, + "grad_norm": 3.0919809341430664, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8875945806503296, + "num_tokens": 511195838.0, + "step": 13167 + }, + { + "epoch": 1.675104948479837, + "grad_norm": 2.818744659423828, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8913478851318359, + "num_tokens": 511229159.0, + "step": 13168 + }, + { + "epoch": 1.6752321587584276, + "grad_norm": 2.5421998500823975, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8989077806472778, + "num_tokens": 511265915.0, + "step": 13169 + }, + { + "epoch": 1.675359369037018, + "grad_norm": 2.2786014080047607, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9085354804992676, + "num_tokens": 511308831.0, + "step": 13170 + }, + { + "epoch": 1.6754865793156086, + "grad_norm": 2.6067938804626465, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9034901261329651, + "num_tokens": 511343737.0, + "step": 13171 + }, + { + "epoch": 1.6756137895941992, + "grad_norm": 2.421628475189209, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8961217403411865, + "num_tokens": 511387594.0, + "step": 13172 + }, + { + "epoch": 1.6757409998727897, + "grad_norm": 2.9560229778289795, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8832900524139404, + "num_tokens": 511424285.0, + "step": 13173 + }, + { + "epoch": 1.6758682101513802, + "grad_norm": 2.371454954147339, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8992164134979248, + "num_tokens": 511469142.0, + "step": 13174 + }, + { + "epoch": 1.6759954204299707, + "grad_norm": 2.95824933052063, + "learning_rate": 1e-06, + "loss": 0.3479, + "mean_token_accuracy": 0.8762080669403076, + "num_tokens": 511501287.0, + "step": 13175 + }, + { + "epoch": 1.6761226307085613, + "grad_norm": 3.4086084365844727, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8981385231018066, + "num_tokens": 511538027.0, + "step": 13176 + }, + { + "epoch": 1.6762498409871518, + "grad_norm": 2.3507251739501953, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8973830342292786, + "num_tokens": 511583065.0, + "step": 13177 + }, + { + "epoch": 1.6763770512657423, + "grad_norm": 2.7597599029541016, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8809374570846558, + "num_tokens": 511623102.0, + "step": 13178 + }, + { + "epoch": 1.6765042615443329, + "grad_norm": 3.1384482383728027, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.8797130584716797, + "num_tokens": 511653384.0, + "step": 13179 + }, + { + "epoch": 1.6766314718229234, + "grad_norm": 2.5434415340423584, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.8990930914878845, + "num_tokens": 511689980.0, + "step": 13180 + }, + { + "epoch": 1.676758682101514, + "grad_norm": 2.554025173187256, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.886970043182373, + "num_tokens": 511729030.0, + "step": 13181 + }, + { + "epoch": 1.6768858923801044, + "grad_norm": 2.404247760772705, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8978490233421326, + "num_tokens": 511766811.0, + "step": 13182 + }, + { + "epoch": 1.677013102658695, + "grad_norm": 2.6658594608306885, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8901865482330322, + "num_tokens": 511803734.0, + "step": 13183 + }, + { + "epoch": 1.6771403129372855, + "grad_norm": 2.520787239074707, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9005396366119385, + "num_tokens": 511840777.0, + "step": 13184 + }, + { + "epoch": 1.6772675232158758, + "grad_norm": 2.665494918823242, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8850560188293457, + "num_tokens": 511880302.0, + "step": 13185 + }, + { + "epoch": 1.6773947334944663, + "grad_norm": 2.7683966159820557, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.892624020576477, + "num_tokens": 511915815.0, + "step": 13186 + }, + { + "epoch": 1.6775219437730569, + "grad_norm": 2.602196216583252, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8868533372879028, + "num_tokens": 511956606.0, + "step": 13187 + }, + { + "epoch": 1.6776491540516474, + "grad_norm": 2.7779030799865723, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.899705171585083, + "num_tokens": 511988691.0, + "step": 13188 + }, + { + "epoch": 1.677776364330238, + "grad_norm": 2.773200511932373, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8904045820236206, + "num_tokens": 512022984.0, + "step": 13189 + }, + { + "epoch": 1.6779035746088284, + "grad_norm": 2.7134335041046143, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.879906415939331, + "num_tokens": 512065894.0, + "step": 13190 + }, + { + "epoch": 1.6780307848874187, + "grad_norm": 2.830880641937256, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8970063924789429, + "num_tokens": 512098686.0, + "step": 13191 + }, + { + "epoch": 1.6781579951660093, + "grad_norm": 2.6575217247009277, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.8834977149963379, + "num_tokens": 512139738.0, + "step": 13192 + }, + { + "epoch": 1.6782852054445998, + "grad_norm": 2.611496686935425, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8781206607818604, + "num_tokens": 512183525.0, + "step": 13193 + }, + { + "epoch": 1.6784124157231903, + "grad_norm": 2.352032423019409, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8890231847763062, + "num_tokens": 512231361.0, + "step": 13194 + }, + { + "epoch": 1.6785396260017809, + "grad_norm": 2.8412911891937256, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8898267149925232, + "num_tokens": 512265996.0, + "step": 13195 + }, + { + "epoch": 1.6786668362803714, + "grad_norm": 2.6836681365966797, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8821501135826111, + "num_tokens": 512308101.0, + "step": 13196 + }, + { + "epoch": 1.678794046558962, + "grad_norm": 2.923842191696167, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8718727827072144, + "num_tokens": 512348339.0, + "step": 13197 + }, + { + "epoch": 1.6789212568375524, + "grad_norm": 2.4724507331848145, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8982230424880981, + "num_tokens": 512389294.0, + "step": 13198 + }, + { + "epoch": 1.679048467116143, + "grad_norm": 2.67169189453125, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8910582065582275, + "num_tokens": 512426624.0, + "step": 13199 + }, + { + "epoch": 1.6791756773947335, + "grad_norm": 2.595088481903076, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8926699161529541, + "num_tokens": 512467361.0, + "step": 13200 + }, + { + "epoch": 1.679302887673324, + "grad_norm": 2.6671481132507324, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8871164321899414, + "num_tokens": 512505979.0, + "step": 13201 + }, + { + "epoch": 1.6794300979519146, + "grad_norm": 2.6851766109466553, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8899358510971069, + "num_tokens": 512541613.0, + "step": 13202 + }, + { + "epoch": 1.679557308230505, + "grad_norm": 2.5719640254974365, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8991385102272034, + "num_tokens": 512578192.0, + "step": 13203 + }, + { + "epoch": 1.6796845185090956, + "grad_norm": 2.2858169078826904, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.8992286920547485, + "num_tokens": 512622750.0, + "step": 13204 + }, + { + "epoch": 1.6798117287876861, + "grad_norm": 2.6188838481903076, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8845614194869995, + "num_tokens": 512663555.0, + "step": 13205 + }, + { + "epoch": 1.6799389390662767, + "grad_norm": 2.5938470363616943, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.893958330154419, + "num_tokens": 512702700.0, + "step": 13206 + }, + { + "epoch": 1.6800661493448672, + "grad_norm": 2.600618362426758, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.8988999724388123, + "num_tokens": 512739231.0, + "step": 13207 + }, + { + "epoch": 1.6801933596234577, + "grad_norm": 2.7378251552581787, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8671423196792603, + "num_tokens": 512778289.0, + "step": 13208 + }, + { + "epoch": 1.680320569902048, + "grad_norm": 2.7913036346435547, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.889040470123291, + "num_tokens": 512814340.0, + "step": 13209 + }, + { + "epoch": 1.6804477801806386, + "grad_norm": 2.418889045715332, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8766286373138428, + "num_tokens": 512862196.0, + "step": 13210 + }, + { + "epoch": 1.680574990459229, + "grad_norm": 2.6081390380859375, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8986133337020874, + "num_tokens": 512901594.0, + "step": 13211 + }, + { + "epoch": 1.6807022007378196, + "grad_norm": 2.5868871212005615, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.893246591091156, + "num_tokens": 512941294.0, + "step": 13212 + }, + { + "epoch": 1.6808294110164101, + "grad_norm": 2.574108362197876, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.9004455804824829, + "num_tokens": 512977985.0, + "step": 13213 + }, + { + "epoch": 1.6809566212950007, + "grad_norm": 2.493924617767334, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.907696008682251, + "num_tokens": 513017606.0, + "step": 13214 + }, + { + "epoch": 1.681083831573591, + "grad_norm": 2.562373161315918, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9008303880691528, + "num_tokens": 513055582.0, + "step": 13215 + }, + { + "epoch": 1.6812110418521815, + "grad_norm": 2.526263475418091, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8896968364715576, + "num_tokens": 513096362.0, + "step": 13216 + }, + { + "epoch": 1.681338252130772, + "grad_norm": 2.744767427444458, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.882770299911499, + "num_tokens": 513135659.0, + "step": 13217 + }, + { + "epoch": 1.6814654624093626, + "grad_norm": 2.7235968112945557, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8883428573608398, + "num_tokens": 513173693.0, + "step": 13218 + }, + { + "epoch": 1.681592672687953, + "grad_norm": 2.5596201419830322, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8922426700592041, + "num_tokens": 513213289.0, + "step": 13219 + }, + { + "epoch": 1.6817198829665436, + "grad_norm": 2.4855704307556152, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9077849388122559, + "num_tokens": 513249695.0, + "step": 13220 + }, + { + "epoch": 1.6818470932451342, + "grad_norm": 2.5596747398376465, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8843929767608643, + "num_tokens": 513293360.0, + "step": 13221 + }, + { + "epoch": 1.6819743035237247, + "grad_norm": 2.7979698181152344, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8825060129165649, + "num_tokens": 513327647.0, + "step": 13222 + }, + { + "epoch": 1.6821015138023152, + "grad_norm": 2.7642061710357666, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.882195234298706, + "num_tokens": 513369140.0, + "step": 13223 + }, + { + "epoch": 1.6822287240809057, + "grad_norm": 2.6969776153564453, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9007556438446045, + "num_tokens": 513402913.0, + "step": 13224 + }, + { + "epoch": 1.6823559343594963, + "grad_norm": 2.863206148147583, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8946208953857422, + "num_tokens": 513432749.0, + "step": 13225 + }, + { + "epoch": 1.6824831446380868, + "grad_norm": 2.698686361312866, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8985388278961182, + "num_tokens": 513465708.0, + "step": 13226 + }, + { + "epoch": 1.6826103549166773, + "grad_norm": 2.5815768241882324, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8819109201431274, + "num_tokens": 513509501.0, + "step": 13227 + }, + { + "epoch": 1.6827375651952678, + "grad_norm": 2.5100200176239014, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8963479995727539, + "num_tokens": 513549775.0, + "step": 13228 + }, + { + "epoch": 1.6828647754738584, + "grad_norm": 2.6101462841033936, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8856256008148193, + "num_tokens": 513591069.0, + "step": 13229 + }, + { + "epoch": 1.682991985752449, + "grad_norm": 2.5393030643463135, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8990780115127563, + "num_tokens": 513627135.0, + "step": 13230 + }, + { + "epoch": 1.6831191960310394, + "grad_norm": 2.6908042430877686, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8919045329093933, + "num_tokens": 513666188.0, + "step": 13231 + }, + { + "epoch": 1.68324640630963, + "grad_norm": 2.554547071456909, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8837013244628906, + "num_tokens": 513709726.0, + "step": 13232 + }, + { + "epoch": 1.6833736165882205, + "grad_norm": 2.515277624130249, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8949167728424072, + "num_tokens": 513748024.0, + "step": 13233 + }, + { + "epoch": 1.6835008268668108, + "grad_norm": 2.774932861328125, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8887112140655518, + "num_tokens": 513784150.0, + "step": 13234 + }, + { + "epoch": 1.6836280371454013, + "grad_norm": 2.79436993598938, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8948133587837219, + "num_tokens": 513819953.0, + "step": 13235 + }, + { + "epoch": 1.6837552474239919, + "grad_norm": 2.6116554737091064, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8987283110618591, + "num_tokens": 513856937.0, + "step": 13236 + }, + { + "epoch": 1.6838824577025824, + "grad_norm": 2.6353113651275635, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8930797576904297, + "num_tokens": 513895320.0, + "step": 13237 + }, + { + "epoch": 1.684009667981173, + "grad_norm": 2.420124053955078, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8920361995697021, + "num_tokens": 513940629.0, + "step": 13238 + }, + { + "epoch": 1.6841368782597634, + "grad_norm": 2.441755771636963, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8926900625228882, + "num_tokens": 513984114.0, + "step": 13239 + }, + { + "epoch": 1.6842640885383537, + "grad_norm": 2.539614200592041, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8931521773338318, + "num_tokens": 514023760.0, + "step": 13240 + }, + { + "epoch": 1.6843912988169443, + "grad_norm": 2.554561138153076, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8989067077636719, + "num_tokens": 514064156.0, + "step": 13241 + }, + { + "epoch": 1.6845185090955348, + "grad_norm": 2.4636423587799072, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.9010528922080994, + "num_tokens": 514105087.0, + "step": 13242 + }, + { + "epoch": 1.6846457193741253, + "grad_norm": 2.542846202850342, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.892136812210083, + "num_tokens": 514144226.0, + "step": 13243 + }, + { + "epoch": 1.6847729296527159, + "grad_norm": 2.4923157691955566, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8945299386978149, + "num_tokens": 514184270.0, + "step": 13244 + }, + { + "epoch": 1.6849001399313064, + "grad_norm": 2.4834342002868652, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8991409540176392, + "num_tokens": 514224120.0, + "step": 13245 + }, + { + "epoch": 1.685027350209897, + "grad_norm": 2.6270227432250977, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8943779468536377, + "num_tokens": 514260491.0, + "step": 13246 + }, + { + "epoch": 1.6851545604884874, + "grad_norm": 2.392078399658203, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.897588312625885, + "num_tokens": 514302709.0, + "step": 13247 + }, + { + "epoch": 1.685281770767078, + "grad_norm": 2.748258352279663, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8915925025939941, + "num_tokens": 514339783.0, + "step": 13248 + }, + { + "epoch": 1.6854089810456685, + "grad_norm": 2.470097780227661, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8865203261375427, + "num_tokens": 514384613.0, + "step": 13249 + }, + { + "epoch": 1.685536191324259, + "grad_norm": 2.633141040802002, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8910386562347412, + "num_tokens": 514421576.0, + "step": 13250 + }, + { + "epoch": 1.6856634016028496, + "grad_norm": 2.6867854595184326, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8862396478652954, + "num_tokens": 514459742.0, + "step": 13251 + }, + { + "epoch": 1.68579061188144, + "grad_norm": 2.849780321121216, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8973696231842041, + "num_tokens": 514494238.0, + "step": 13252 + }, + { + "epoch": 1.6859178221600306, + "grad_norm": 2.5144097805023193, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8909361362457275, + "num_tokens": 514535509.0, + "step": 13253 + }, + { + "epoch": 1.6860450324386211, + "grad_norm": 2.6722419261932373, + "learning_rate": 1e-06, + "loss": 0.3325, + "mean_token_accuracy": 0.88033127784729, + "num_tokens": 514576327.0, + "step": 13254 + }, + { + "epoch": 1.6861722427172117, + "grad_norm": 3.0408785343170166, + "learning_rate": 1e-06, + "loss": 0.3573, + "mean_token_accuracy": 0.8767904043197632, + "num_tokens": 514611783.0, + "step": 13255 + }, + { + "epoch": 1.6862994529958022, + "grad_norm": 2.572113037109375, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8947142958641052, + "num_tokens": 514652302.0, + "step": 13256 + }, + { + "epoch": 1.6864266632743927, + "grad_norm": 2.3584043979644775, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9029717445373535, + "num_tokens": 514693938.0, + "step": 13257 + }, + { + "epoch": 1.686553873552983, + "grad_norm": 2.485241413116455, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8976799249649048, + "num_tokens": 514735014.0, + "step": 13258 + }, + { + "epoch": 1.6866810838315736, + "grad_norm": 2.817807197570801, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8837682008743286, + "num_tokens": 514775052.0, + "step": 13259 + }, + { + "epoch": 1.686808294110164, + "grad_norm": 2.602550745010376, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8864197731018066, + "num_tokens": 514815598.0, + "step": 13260 + }, + { + "epoch": 1.6869355043887546, + "grad_norm": 2.555366277694702, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.8850558996200562, + "num_tokens": 514858781.0, + "step": 13261 + }, + { + "epoch": 1.6870627146673451, + "grad_norm": 2.717074394226074, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8846873044967651, + "num_tokens": 514898588.0, + "step": 13262 + }, + { + "epoch": 1.6871899249459357, + "grad_norm": 2.8267407417297363, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8835541605949402, + "num_tokens": 514932379.0, + "step": 13263 + }, + { + "epoch": 1.687317135224526, + "grad_norm": 2.7129967212677, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.879418671131134, + "num_tokens": 514972613.0, + "step": 13264 + }, + { + "epoch": 1.6874443455031165, + "grad_norm": 2.5667660236358643, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8991848230361938, + "num_tokens": 515012382.0, + "step": 13265 + }, + { + "epoch": 1.687571555781707, + "grad_norm": 2.58083176612854, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8952099084854126, + "num_tokens": 515052700.0, + "step": 13266 + }, + { + "epoch": 1.6876987660602976, + "grad_norm": 2.6212785243988037, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8817428946495056, + "num_tokens": 515095896.0, + "step": 13267 + }, + { + "epoch": 1.687825976338888, + "grad_norm": 2.5225331783294678, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9006048440933228, + "num_tokens": 515135573.0, + "step": 13268 + }, + { + "epoch": 1.6879531866174786, + "grad_norm": 2.5390870571136475, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8988361358642578, + "num_tokens": 515176523.0, + "step": 13269 + }, + { + "epoch": 1.6880803968960691, + "grad_norm": 2.4906187057495117, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8824142217636108, + "num_tokens": 515222544.0, + "step": 13270 + }, + { + "epoch": 1.6882076071746597, + "grad_norm": 2.64880108833313, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8929554224014282, + "num_tokens": 515260515.0, + "step": 13271 + }, + { + "epoch": 1.6883348174532502, + "grad_norm": 2.736353635787964, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.881873607635498, + "num_tokens": 515300072.0, + "step": 13272 + }, + { + "epoch": 1.6884620277318407, + "grad_norm": 2.721775531768799, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8847849369049072, + "num_tokens": 515340749.0, + "step": 13273 + }, + { + "epoch": 1.6885892380104313, + "grad_norm": 2.542147159576416, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8955641984939575, + "num_tokens": 515379193.0, + "step": 13274 + }, + { + "epoch": 1.6887164482890218, + "grad_norm": 2.7158515453338623, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8882990479469299, + "num_tokens": 515419885.0, + "step": 13275 + }, + { + "epoch": 1.6888436585676123, + "grad_norm": 2.4281563758850098, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.894716203212738, + "num_tokens": 515464146.0, + "step": 13276 + }, + { + "epoch": 1.6889708688462028, + "grad_norm": 2.5180513858795166, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9001048803329468, + "num_tokens": 515503284.0, + "step": 13277 + }, + { + "epoch": 1.6890980791247934, + "grad_norm": 2.6358895301818848, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8914058208465576, + "num_tokens": 515540415.0, + "step": 13278 + }, + { + "epoch": 1.689225289403384, + "grad_norm": 2.7203428745269775, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8852195739746094, + "num_tokens": 515576512.0, + "step": 13279 + }, + { + "epoch": 1.6893524996819744, + "grad_norm": 2.5745794773101807, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.896312952041626, + "num_tokens": 515616074.0, + "step": 13280 + }, + { + "epoch": 1.689479709960565, + "grad_norm": 2.7841601371765137, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.893110990524292, + "num_tokens": 515653735.0, + "step": 13281 + }, + { + "epoch": 1.6896069202391555, + "grad_norm": 2.7348484992980957, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.884692907333374, + "num_tokens": 515694107.0, + "step": 13282 + }, + { + "epoch": 1.6897341305177458, + "grad_norm": 2.620360851287842, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8838186264038086, + "num_tokens": 515729109.0, + "step": 13283 + }, + { + "epoch": 1.6898613407963363, + "grad_norm": 2.6794216632843018, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9031816124916077, + "num_tokens": 515762497.0, + "step": 13284 + }, + { + "epoch": 1.6899885510749268, + "grad_norm": 2.7663357257843018, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8870177268981934, + "num_tokens": 515802866.0, + "step": 13285 + }, + { + "epoch": 1.6901157613535174, + "grad_norm": 2.7947943210601807, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8812874555587769, + "num_tokens": 515842600.0, + "step": 13286 + }, + { + "epoch": 1.690242971632108, + "grad_norm": 2.57619309425354, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8953943252563477, + "num_tokens": 515878614.0, + "step": 13287 + }, + { + "epoch": 1.6903701819106984, + "grad_norm": 2.439481258392334, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8874019980430603, + "num_tokens": 515922962.0, + "step": 13288 + }, + { + "epoch": 1.6904973921892887, + "grad_norm": 2.435044527053833, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8899602890014648, + "num_tokens": 515965416.0, + "step": 13289 + }, + { + "epoch": 1.6906246024678793, + "grad_norm": 2.5094048976898193, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8926892280578613, + "num_tokens": 516003688.0, + "step": 13290 + }, + { + "epoch": 1.6907518127464698, + "grad_norm": 2.4995920658111572, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.903383731842041, + "num_tokens": 516041339.0, + "step": 13291 + }, + { + "epoch": 1.6908790230250603, + "grad_norm": 2.583249092102051, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8889174461364746, + "num_tokens": 516086309.0, + "step": 13292 + }, + { + "epoch": 1.6910062333036509, + "grad_norm": 3.010205030441284, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8857272863388062, + "num_tokens": 516121383.0, + "step": 13293 + }, + { + "epoch": 1.6911334435822414, + "grad_norm": 2.495630979537964, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8885303735733032, + "num_tokens": 516163035.0, + "step": 13294 + }, + { + "epoch": 1.691260653860832, + "grad_norm": 2.7849690914154053, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8770240545272827, + "num_tokens": 516199553.0, + "step": 13295 + }, + { + "epoch": 1.6913878641394224, + "grad_norm": 2.3756253719329834, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8900800943374634, + "num_tokens": 516241383.0, + "step": 13296 + }, + { + "epoch": 1.691515074418013, + "grad_norm": 2.3836302757263184, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8960952758789062, + "num_tokens": 516283919.0, + "step": 13297 + }, + { + "epoch": 1.6916422846966035, + "grad_norm": 2.4702038764953613, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8957701325416565, + "num_tokens": 516325941.0, + "step": 13298 + }, + { + "epoch": 1.691769494975194, + "grad_norm": 2.4340357780456543, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.8994928002357483, + "num_tokens": 516367436.0, + "step": 13299 + }, + { + "epoch": 1.6918967052537845, + "grad_norm": 2.48290753364563, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8943063020706177, + "num_tokens": 516408966.0, + "step": 13300 + }, + { + "epoch": 1.692023915532375, + "grad_norm": 2.428417444229126, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9043195247650146, + "num_tokens": 516448599.0, + "step": 13301 + }, + { + "epoch": 1.6921511258109656, + "grad_norm": 2.6273033618927, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8934352397918701, + "num_tokens": 516487408.0, + "step": 13302 + }, + { + "epoch": 1.6922783360895561, + "grad_norm": 2.494755983352661, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8932311534881592, + "num_tokens": 516532190.0, + "step": 13303 + }, + { + "epoch": 1.6924055463681467, + "grad_norm": 2.5055782794952393, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8939895033836365, + "num_tokens": 516569857.0, + "step": 13304 + }, + { + "epoch": 1.6925327566467372, + "grad_norm": 2.8782801628112793, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8913163542747498, + "num_tokens": 516603899.0, + "step": 13305 + }, + { + "epoch": 1.6926599669253277, + "grad_norm": 2.6782968044281006, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8927494287490845, + "num_tokens": 516641407.0, + "step": 13306 + }, + { + "epoch": 1.692787177203918, + "grad_norm": 2.8005025386810303, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.88005530834198, + "num_tokens": 516676477.0, + "step": 13307 + }, + { + "epoch": 1.6929143874825086, + "grad_norm": 2.656341075897217, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8910229802131653, + "num_tokens": 516714535.0, + "step": 13308 + }, + { + "epoch": 1.693041597761099, + "grad_norm": 2.6412386894226074, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8918951749801636, + "num_tokens": 516750392.0, + "step": 13309 + }, + { + "epoch": 1.6931688080396896, + "grad_norm": 2.6617794036865234, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8937649726867676, + "num_tokens": 516785824.0, + "step": 13310 + }, + { + "epoch": 1.6932960183182801, + "grad_norm": 2.4670188426971436, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8951719999313354, + "num_tokens": 516827272.0, + "step": 13311 + }, + { + "epoch": 1.6934232285968707, + "grad_norm": 2.5034244060516357, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.895986020565033, + "num_tokens": 516867653.0, + "step": 13312 + }, + { + "epoch": 1.693550438875461, + "grad_norm": 2.4832892417907715, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8974395990371704, + "num_tokens": 516907788.0, + "step": 13313 + }, + { + "epoch": 1.6936776491540515, + "grad_norm": 2.4347777366638184, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8908621668815613, + "num_tokens": 516949242.0, + "step": 13314 + }, + { + "epoch": 1.693804859432642, + "grad_norm": 2.727865219116211, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8945790529251099, + "num_tokens": 516988528.0, + "step": 13315 + }, + { + "epoch": 1.6939320697112326, + "grad_norm": 2.4374594688415527, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8883069157600403, + "num_tokens": 517035227.0, + "step": 13316 + }, + { + "epoch": 1.694059279989823, + "grad_norm": 2.492525577545166, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8997154831886292, + "num_tokens": 517074261.0, + "step": 13317 + }, + { + "epoch": 1.6941864902684136, + "grad_norm": 2.601409435272217, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8823928833007812, + "num_tokens": 517115665.0, + "step": 13318 + }, + { + "epoch": 1.6943137005470041, + "grad_norm": 2.9196043014526367, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8847008943557739, + "num_tokens": 517149086.0, + "step": 13319 + }, + { + "epoch": 1.6944409108255947, + "grad_norm": 2.549278497695923, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8775568604469299, + "num_tokens": 517190480.0, + "step": 13320 + }, + { + "epoch": 1.6945681211041852, + "grad_norm": 2.5261926651000977, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8885444402694702, + "num_tokens": 517234556.0, + "step": 13321 + }, + { + "epoch": 1.6946953313827757, + "grad_norm": 2.8658907413482666, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8808098435401917, + "num_tokens": 517274215.0, + "step": 13322 + }, + { + "epoch": 1.6948225416613663, + "grad_norm": 2.4823782444000244, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.885176420211792, + "num_tokens": 517320512.0, + "step": 13323 + }, + { + "epoch": 1.6949497519399568, + "grad_norm": 2.6588871479034424, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8934284448623657, + "num_tokens": 517358641.0, + "step": 13324 + }, + { + "epoch": 1.6950769622185473, + "grad_norm": 2.564918279647827, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8917987942695618, + "num_tokens": 517397136.0, + "step": 13325 + }, + { + "epoch": 1.6952041724971378, + "grad_norm": 2.666273593902588, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8856987953186035, + "num_tokens": 517435338.0, + "step": 13326 + }, + { + "epoch": 1.6953313827757284, + "grad_norm": 3.0218427181243896, + "learning_rate": 1e-06, + "loss": 0.3685, + "mean_token_accuracy": 0.8704535961151123, + "num_tokens": 517468955.0, + "step": 13327 + }, + { + "epoch": 1.695458593054319, + "grad_norm": 2.8117167949676514, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8855020999908447, + "num_tokens": 517503736.0, + "step": 13328 + }, + { + "epoch": 1.6955858033329094, + "grad_norm": 2.890817880630493, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8754149675369263, + "num_tokens": 517536630.0, + "step": 13329 + }, + { + "epoch": 1.6957130136115, + "grad_norm": 2.4608373641967773, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9036182165145874, + "num_tokens": 517575255.0, + "step": 13330 + }, + { + "epoch": 1.6958402238900905, + "grad_norm": 2.2813398838043213, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8967125415802002, + "num_tokens": 517620311.0, + "step": 13331 + }, + { + "epoch": 1.6959674341686808, + "grad_norm": 2.6681671142578125, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.897010862827301, + "num_tokens": 517656453.0, + "step": 13332 + }, + { + "epoch": 1.6960946444472713, + "grad_norm": 2.7150185108184814, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8783441781997681, + "num_tokens": 517695483.0, + "step": 13333 + }, + { + "epoch": 1.6962218547258618, + "grad_norm": 2.6143996715545654, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8891428709030151, + "num_tokens": 517735585.0, + "step": 13334 + }, + { + "epoch": 1.6963490650044524, + "grad_norm": 2.56257700920105, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8931654095649719, + "num_tokens": 517774857.0, + "step": 13335 + }, + { + "epoch": 1.696476275283043, + "grad_norm": 2.659241199493408, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8937383890151978, + "num_tokens": 517809820.0, + "step": 13336 + }, + { + "epoch": 1.6966034855616334, + "grad_norm": 2.5887680053710938, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8949215412139893, + "num_tokens": 517847004.0, + "step": 13337 + }, + { + "epoch": 1.6967306958402237, + "grad_norm": 2.7316060066223145, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8893226981163025, + "num_tokens": 517885890.0, + "step": 13338 + }, + { + "epoch": 1.6968579061188143, + "grad_norm": 2.568265914916992, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.888252854347229, + "num_tokens": 517930514.0, + "step": 13339 + }, + { + "epoch": 1.6969851163974048, + "grad_norm": 3.064357042312622, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.8776930570602417, + "num_tokens": 517965536.0, + "step": 13340 + }, + { + "epoch": 1.6971123266759953, + "grad_norm": 2.6982059478759766, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8900230526924133, + "num_tokens": 518007092.0, + "step": 13341 + }, + { + "epoch": 1.6972395369545858, + "grad_norm": 3.0506391525268555, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8869383335113525, + "num_tokens": 518036547.0, + "step": 13342 + }, + { + "epoch": 1.6973667472331764, + "grad_norm": 2.737281322479248, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8863387107849121, + "num_tokens": 518075685.0, + "step": 13343 + }, + { + "epoch": 1.697493957511767, + "grad_norm": 2.6139261722564697, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8981325626373291, + "num_tokens": 518112775.0, + "step": 13344 + }, + { + "epoch": 1.6976211677903574, + "grad_norm": 2.530522346496582, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9019951224327087, + "num_tokens": 518158730.0, + "step": 13345 + }, + { + "epoch": 1.697748378068948, + "grad_norm": 2.6523118019104004, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8802824020385742, + "num_tokens": 518200692.0, + "step": 13346 + }, + { + "epoch": 1.6978755883475385, + "grad_norm": 2.587651491165161, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8973706960678101, + "num_tokens": 518242175.0, + "step": 13347 + }, + { + "epoch": 1.698002798626129, + "grad_norm": 2.7663841247558594, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9009417295455933, + "num_tokens": 518279216.0, + "step": 13348 + }, + { + "epoch": 1.6981300089047195, + "grad_norm": 2.4618709087371826, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8911003470420837, + "num_tokens": 518320588.0, + "step": 13349 + }, + { + "epoch": 1.69825721918331, + "grad_norm": 2.5316381454467773, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.8961303234100342, + "num_tokens": 518358646.0, + "step": 13350 + }, + { + "epoch": 1.6983844294619006, + "grad_norm": 2.5626025199890137, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8940925598144531, + "num_tokens": 518404974.0, + "step": 13351 + }, + { + "epoch": 1.6985116397404911, + "grad_norm": 2.889132499694824, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8843469619750977, + "num_tokens": 518440061.0, + "step": 13352 + }, + { + "epoch": 1.6986388500190817, + "grad_norm": 2.6929726600646973, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8938169479370117, + "num_tokens": 518473575.0, + "step": 13353 + }, + { + "epoch": 1.6987660602976722, + "grad_norm": 2.548426866531372, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8900015950202942, + "num_tokens": 518510374.0, + "step": 13354 + }, + { + "epoch": 1.6988932705762627, + "grad_norm": 2.542051315307617, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8993010520935059, + "num_tokens": 518549070.0, + "step": 13355 + }, + { + "epoch": 1.699020480854853, + "grad_norm": 2.8026251792907715, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8861762285232544, + "num_tokens": 518585789.0, + "step": 13356 + }, + { + "epoch": 1.6991476911334435, + "grad_norm": 2.732877731323242, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.873218297958374, + "num_tokens": 518629257.0, + "step": 13357 + }, + { + "epoch": 1.699274901412034, + "grad_norm": 2.7283620834350586, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8864424824714661, + "num_tokens": 518668192.0, + "step": 13358 + }, + { + "epoch": 1.6994021116906246, + "grad_norm": 2.8577823638916016, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8723435401916504, + "num_tokens": 518706177.0, + "step": 13359 + }, + { + "epoch": 1.6995293219692151, + "grad_norm": 2.5761916637420654, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8912941217422485, + "num_tokens": 518743765.0, + "step": 13360 + }, + { + "epoch": 1.6996565322478057, + "grad_norm": 2.5097243785858154, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8903907537460327, + "num_tokens": 518785927.0, + "step": 13361 + }, + { + "epoch": 1.699783742526396, + "grad_norm": 2.6494948863983154, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8961190581321716, + "num_tokens": 518820383.0, + "step": 13362 + }, + { + "epoch": 1.6999109528049865, + "grad_norm": 2.77736496925354, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.893279492855072, + "num_tokens": 518854590.0, + "step": 13363 + }, + { + "epoch": 1.700038163083577, + "grad_norm": 2.648524045944214, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8848578929901123, + "num_tokens": 518896225.0, + "step": 13364 + }, + { + "epoch": 1.7001653733621676, + "grad_norm": 2.5408551692962646, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8887600302696228, + "num_tokens": 518937702.0, + "step": 13365 + }, + { + "epoch": 1.700292583640758, + "grad_norm": 2.6948487758636475, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9030543565750122, + "num_tokens": 518970011.0, + "step": 13366 + }, + { + "epoch": 1.7004197939193486, + "grad_norm": 2.6874961853027344, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8858789205551147, + "num_tokens": 519009581.0, + "step": 13367 + }, + { + "epoch": 1.7005470041979391, + "grad_norm": 2.498753547668457, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8888390064239502, + "num_tokens": 519050504.0, + "step": 13368 + }, + { + "epoch": 1.7006742144765297, + "grad_norm": 2.5453941822052, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8876369595527649, + "num_tokens": 519090700.0, + "step": 13369 + }, + { + "epoch": 1.7008014247551202, + "grad_norm": 2.5719828605651855, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8928504586219788, + "num_tokens": 519133110.0, + "step": 13370 + }, + { + "epoch": 1.7009286350337107, + "grad_norm": 2.9973225593566895, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8896668553352356, + "num_tokens": 519164122.0, + "step": 13371 + }, + { + "epoch": 1.7010558453123013, + "grad_norm": 2.4047889709472656, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.902530312538147, + "num_tokens": 519203154.0, + "step": 13372 + }, + { + "epoch": 1.7011830555908918, + "grad_norm": 3.0191266536712646, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8741244673728943, + "num_tokens": 519246688.0, + "step": 13373 + }, + { + "epoch": 1.7013102658694823, + "grad_norm": 2.4847335815429688, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.889127790927887, + "num_tokens": 519293821.0, + "step": 13374 + }, + { + "epoch": 1.7014374761480728, + "grad_norm": 2.5417466163635254, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8949105739593506, + "num_tokens": 519333373.0, + "step": 13375 + }, + { + "epoch": 1.7015646864266634, + "grad_norm": 2.5266473293304443, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8961116075515747, + "num_tokens": 519374064.0, + "step": 13376 + }, + { + "epoch": 1.701691896705254, + "grad_norm": 2.679499387741089, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8906122446060181, + "num_tokens": 519412728.0, + "step": 13377 + }, + { + "epoch": 1.7018191069838444, + "grad_norm": 2.620518922805786, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8953986763954163, + "num_tokens": 519449286.0, + "step": 13378 + }, + { + "epoch": 1.701946317262435, + "grad_norm": 2.595677137374878, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8978623151779175, + "num_tokens": 519483408.0, + "step": 13379 + }, + { + "epoch": 1.7020735275410255, + "grad_norm": 2.5157549381256104, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8902649879455566, + "num_tokens": 519523056.0, + "step": 13380 + }, + { + "epoch": 1.7022007378196158, + "grad_norm": 2.4626989364624023, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8976951241493225, + "num_tokens": 519561705.0, + "step": 13381 + }, + { + "epoch": 1.7023279480982063, + "grad_norm": 2.5237526893615723, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8908253908157349, + "num_tokens": 519603218.0, + "step": 13382 + }, + { + "epoch": 1.7024551583767968, + "grad_norm": 2.893559217453003, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8845414519309998, + "num_tokens": 519636783.0, + "step": 13383 + }, + { + "epoch": 1.7025823686553874, + "grad_norm": 2.5748491287231445, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8854959011077881, + "num_tokens": 519676830.0, + "step": 13384 + }, + { + "epoch": 1.702709578933978, + "grad_norm": 2.8164193630218506, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8869128823280334, + "num_tokens": 519711373.0, + "step": 13385 + }, + { + "epoch": 1.7028367892125684, + "grad_norm": 2.524890899658203, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8949214220046997, + "num_tokens": 519750778.0, + "step": 13386 + }, + { + "epoch": 1.7029639994911587, + "grad_norm": 2.7565016746520996, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8908681273460388, + "num_tokens": 519785077.0, + "step": 13387 + }, + { + "epoch": 1.7030912097697493, + "grad_norm": 2.6525180339813232, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8921400308609009, + "num_tokens": 519820518.0, + "step": 13388 + }, + { + "epoch": 1.7032184200483398, + "grad_norm": 2.632814407348633, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8925811052322388, + "num_tokens": 519860777.0, + "step": 13389 + }, + { + "epoch": 1.7033456303269303, + "grad_norm": 2.753715753555298, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8884671330451965, + "num_tokens": 519895109.0, + "step": 13390 + }, + { + "epoch": 1.7034728406055208, + "grad_norm": 2.3970584869384766, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9007971286773682, + "num_tokens": 519939240.0, + "step": 13391 + }, + { + "epoch": 1.7036000508841114, + "grad_norm": 2.8043737411499023, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8734602332115173, + "num_tokens": 519980205.0, + "step": 13392 + }, + { + "epoch": 1.703727261162702, + "grad_norm": 2.5362443923950195, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.891013503074646, + "num_tokens": 520019666.0, + "step": 13393 + }, + { + "epoch": 1.7038544714412924, + "grad_norm": 2.365171194076538, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8951029777526855, + "num_tokens": 520063849.0, + "step": 13394 + }, + { + "epoch": 1.703981681719883, + "grad_norm": 2.7904438972473145, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8780108094215393, + "num_tokens": 520102134.0, + "step": 13395 + }, + { + "epoch": 1.7041088919984735, + "grad_norm": 2.870229721069336, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8903518319129944, + "num_tokens": 520134559.0, + "step": 13396 + }, + { + "epoch": 1.704236102277064, + "grad_norm": 2.572357654571533, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8937661051750183, + "num_tokens": 520173764.0, + "step": 13397 + }, + { + "epoch": 1.7043633125556545, + "grad_norm": 2.416034460067749, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8889228105545044, + "num_tokens": 520219133.0, + "step": 13398 + }, + { + "epoch": 1.704490522834245, + "grad_norm": 2.5707778930664062, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8843709230422974, + "num_tokens": 520262870.0, + "step": 13399 + }, + { + "epoch": 1.7046177331128356, + "grad_norm": 2.6854779720306396, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9015681743621826, + "num_tokens": 520296008.0, + "step": 13400 + }, + { + "epoch": 1.7047449433914261, + "grad_norm": 2.5996742248535156, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8915146589279175, + "num_tokens": 520334870.0, + "step": 13401 + }, + { + "epoch": 1.7048721536700167, + "grad_norm": 2.618307590484619, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.886086106300354, + "num_tokens": 520372449.0, + "step": 13402 + }, + { + "epoch": 1.7049993639486072, + "grad_norm": 2.3898112773895264, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9086266160011292, + "num_tokens": 520411414.0, + "step": 13403 + }, + { + "epoch": 1.7051265742271977, + "grad_norm": 2.4192166328430176, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9015370607376099, + "num_tokens": 520452590.0, + "step": 13404 + }, + { + "epoch": 1.705253784505788, + "grad_norm": 2.686829090118408, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8904086351394653, + "num_tokens": 520493826.0, + "step": 13405 + }, + { + "epoch": 1.7053809947843785, + "grad_norm": 2.5134830474853516, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8898091316223145, + "num_tokens": 520535080.0, + "step": 13406 + }, + { + "epoch": 1.705508205062969, + "grad_norm": 2.4974658489227295, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8905901908874512, + "num_tokens": 520579385.0, + "step": 13407 + }, + { + "epoch": 1.7056354153415596, + "grad_norm": 2.508559226989746, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8944530487060547, + "num_tokens": 520621215.0, + "step": 13408 + }, + { + "epoch": 1.7057626256201501, + "grad_norm": 2.5080556869506836, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9062398672103882, + "num_tokens": 520655872.0, + "step": 13409 + }, + { + "epoch": 1.7058898358987407, + "grad_norm": 2.577911853790283, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9004946351051331, + "num_tokens": 520691372.0, + "step": 13410 + }, + { + "epoch": 1.706017046177331, + "grad_norm": 2.747563123703003, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8920990228652954, + "num_tokens": 520726408.0, + "step": 13411 + }, + { + "epoch": 1.7061442564559215, + "grad_norm": 2.656006097793579, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8872483968734741, + "num_tokens": 520765546.0, + "step": 13412 + }, + { + "epoch": 1.706271466734512, + "grad_norm": 2.787984609603882, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8766888380050659, + "num_tokens": 520804829.0, + "step": 13413 + }, + { + "epoch": 1.7063986770131025, + "grad_norm": 2.6066396236419678, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8897823691368103, + "num_tokens": 520844977.0, + "step": 13414 + }, + { + "epoch": 1.706525887291693, + "grad_norm": 2.7410664558410645, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8790500164031982, + "num_tokens": 520881035.0, + "step": 13415 + }, + { + "epoch": 1.7066530975702836, + "grad_norm": 2.4254379272460938, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8951123952865601, + "num_tokens": 520921313.0, + "step": 13416 + }, + { + "epoch": 1.7067803078488741, + "grad_norm": 2.400573968887329, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8857042193412781, + "num_tokens": 520966988.0, + "step": 13417 + }, + { + "epoch": 1.7069075181274647, + "grad_norm": 2.3847506046295166, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9075464606285095, + "num_tokens": 521004777.0, + "step": 13418 + }, + { + "epoch": 1.7070347284060552, + "grad_norm": 2.9659664630889893, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8969287872314453, + "num_tokens": 521030151.0, + "step": 13419 + }, + { + "epoch": 1.7071619386846457, + "grad_norm": 2.695108413696289, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8893371820449829, + "num_tokens": 521067526.0, + "step": 13420 + }, + { + "epoch": 1.7072891489632362, + "grad_norm": 2.5145797729492188, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.894597053527832, + "num_tokens": 521107123.0, + "step": 13421 + }, + { + "epoch": 1.7074163592418268, + "grad_norm": 2.312420606613159, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8936676979064941, + "num_tokens": 521152217.0, + "step": 13422 + }, + { + "epoch": 1.7075435695204173, + "grad_norm": 2.6148524284362793, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8906205892562866, + "num_tokens": 521190146.0, + "step": 13423 + }, + { + "epoch": 1.7076707797990078, + "grad_norm": 2.8032803535461426, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8813135623931885, + "num_tokens": 521228797.0, + "step": 13424 + }, + { + "epoch": 1.7077979900775984, + "grad_norm": 2.764202356338501, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.8828506469726562, + "num_tokens": 521264691.0, + "step": 13425 + }, + { + "epoch": 1.7079252003561889, + "grad_norm": 2.6766159534454346, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8941431045532227, + "num_tokens": 521299542.0, + "step": 13426 + }, + { + "epoch": 1.7080524106347794, + "grad_norm": 2.533693790435791, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8948013782501221, + "num_tokens": 521340656.0, + "step": 13427 + }, + { + "epoch": 1.70817962091337, + "grad_norm": 3.061166524887085, + "learning_rate": 1e-06, + "loss": 0.3315, + "mean_token_accuracy": 0.8793139457702637, + "num_tokens": 521372743.0, + "step": 13428 + }, + { + "epoch": 1.7083068311919605, + "grad_norm": 2.808032751083374, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8974459171295166, + "num_tokens": 521407710.0, + "step": 13429 + }, + { + "epoch": 1.7084340414705508, + "grad_norm": 2.374742269515991, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8984191417694092, + "num_tokens": 521452074.0, + "step": 13430 + }, + { + "epoch": 1.7085612517491413, + "grad_norm": 3.1072146892547607, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8808598518371582, + "num_tokens": 521485063.0, + "step": 13431 + }, + { + "epoch": 1.7086884620277318, + "grad_norm": 2.6197097301483154, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8902674317359924, + "num_tokens": 521525220.0, + "step": 13432 + }, + { + "epoch": 1.7088156723063224, + "grad_norm": 2.709657907485962, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8881675004959106, + "num_tokens": 521563294.0, + "step": 13433 + }, + { + "epoch": 1.708942882584913, + "grad_norm": 2.687371015548706, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8934157490730286, + "num_tokens": 521598589.0, + "step": 13434 + }, + { + "epoch": 1.7090700928635034, + "grad_norm": 2.5145535469055176, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.8967103958129883, + "num_tokens": 521638665.0, + "step": 13435 + }, + { + "epoch": 1.7091973031420937, + "grad_norm": 2.428191900253296, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9009337425231934, + "num_tokens": 521683026.0, + "step": 13436 + }, + { + "epoch": 1.7093245134206843, + "grad_norm": 2.4982988834381104, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8985925316810608, + "num_tokens": 521722976.0, + "step": 13437 + }, + { + "epoch": 1.7094517236992748, + "grad_norm": 2.7025368213653564, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8906489610671997, + "num_tokens": 521758514.0, + "step": 13438 + }, + { + "epoch": 1.7095789339778653, + "grad_norm": 2.7805187702178955, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8982741832733154, + "num_tokens": 521790647.0, + "step": 13439 + }, + { + "epoch": 1.7097061442564558, + "grad_norm": 2.5439624786376953, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8945987224578857, + "num_tokens": 521831250.0, + "step": 13440 + }, + { + "epoch": 1.7098333545350464, + "grad_norm": 2.6599926948547363, + "learning_rate": 1e-06, + "loss": 0.3303, + "mean_token_accuracy": 0.8813197612762451, + "num_tokens": 521872705.0, + "step": 13441 + }, + { + "epoch": 1.709960564813637, + "grad_norm": 2.5441436767578125, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8978732228279114, + "num_tokens": 521911085.0, + "step": 13442 + }, + { + "epoch": 1.7100877750922274, + "grad_norm": 2.427884101867676, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.8967454433441162, + "num_tokens": 521950349.0, + "step": 13443 + }, + { + "epoch": 1.710214985370818, + "grad_norm": 2.4954473972320557, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8894025087356567, + "num_tokens": 521993953.0, + "step": 13444 + }, + { + "epoch": 1.7103421956494085, + "grad_norm": 2.5722618103027344, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9081341028213501, + "num_tokens": 522027681.0, + "step": 13445 + }, + { + "epoch": 1.710469405927999, + "grad_norm": 2.6554298400878906, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.8977835178375244, + "num_tokens": 522061151.0, + "step": 13446 + }, + { + "epoch": 1.7105966162065895, + "grad_norm": 2.7941672801971436, + "learning_rate": 1e-06, + "loss": 0.324, + "mean_token_accuracy": 0.8836488723754883, + "num_tokens": 522096546.0, + "step": 13447 + }, + { + "epoch": 1.71072382648518, + "grad_norm": 2.522277355194092, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8962015509605408, + "num_tokens": 522139556.0, + "step": 13448 + }, + { + "epoch": 1.7108510367637706, + "grad_norm": 2.456500768661499, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9002553224563599, + "num_tokens": 522178295.0, + "step": 13449 + }, + { + "epoch": 1.7109782470423611, + "grad_norm": 2.766610622406006, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8890335559844971, + "num_tokens": 522214991.0, + "step": 13450 + }, + { + "epoch": 1.7111054573209517, + "grad_norm": 2.544039487838745, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8993193507194519, + "num_tokens": 522253293.0, + "step": 13451 + }, + { + "epoch": 1.7112326675995422, + "grad_norm": 2.631392240524292, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8922445774078369, + "num_tokens": 522292108.0, + "step": 13452 + }, + { + "epoch": 1.7113598778781327, + "grad_norm": 2.5041277408599854, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8983371257781982, + "num_tokens": 522326216.0, + "step": 13453 + }, + { + "epoch": 1.711487088156723, + "grad_norm": 2.536430597305298, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9003440737724304, + "num_tokens": 522363976.0, + "step": 13454 + }, + { + "epoch": 1.7116142984353135, + "grad_norm": 2.33642315864563, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8898295164108276, + "num_tokens": 522409421.0, + "step": 13455 + }, + { + "epoch": 1.711741508713904, + "grad_norm": 2.6387217044830322, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8859161138534546, + "num_tokens": 522446646.0, + "step": 13456 + }, + { + "epoch": 1.7118687189924946, + "grad_norm": 2.5861191749572754, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8905338048934937, + "num_tokens": 522487763.0, + "step": 13457 + }, + { + "epoch": 1.7119959292710851, + "grad_norm": 2.694964647293091, + "learning_rate": 1e-06, + "loss": 0.3433, + "mean_token_accuracy": 0.8808563947677612, + "num_tokens": 522526057.0, + "step": 13458 + }, + { + "epoch": 1.7121231395496757, + "grad_norm": 2.613237142562866, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8915436267852783, + "num_tokens": 522565872.0, + "step": 13459 + }, + { + "epoch": 1.712250349828266, + "grad_norm": 2.6430439949035645, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8833104372024536, + "num_tokens": 522608292.0, + "step": 13460 + }, + { + "epoch": 1.7123775601068565, + "grad_norm": 2.647972583770752, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.8862748742103577, + "num_tokens": 522648498.0, + "step": 13461 + }, + { + "epoch": 1.712504770385447, + "grad_norm": 2.662294864654541, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8822307586669922, + "num_tokens": 522686048.0, + "step": 13462 + }, + { + "epoch": 1.7126319806640375, + "grad_norm": 2.6762330532073975, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8993477821350098, + "num_tokens": 522719037.0, + "step": 13463 + }, + { + "epoch": 1.712759190942628, + "grad_norm": 2.6430087089538574, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8788986206054688, + "num_tokens": 522759989.0, + "step": 13464 + }, + { + "epoch": 1.7128864012212186, + "grad_norm": 2.4282641410827637, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8958579897880554, + "num_tokens": 522801918.0, + "step": 13465 + }, + { + "epoch": 1.7130136114998091, + "grad_norm": 2.6378605365753174, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8994477987289429, + "num_tokens": 522844602.0, + "step": 13466 + }, + { + "epoch": 1.7131408217783997, + "grad_norm": 2.5802536010742188, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.8992764949798584, + "num_tokens": 522880710.0, + "step": 13467 + }, + { + "epoch": 1.7132680320569902, + "grad_norm": 2.494086742401123, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8984792232513428, + "num_tokens": 522920166.0, + "step": 13468 + }, + { + "epoch": 1.7133952423355807, + "grad_norm": 2.4566922187805176, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.892784833908081, + "num_tokens": 522961740.0, + "step": 13469 + }, + { + "epoch": 1.7135224526141712, + "grad_norm": 2.8482868671417236, + "learning_rate": 1e-06, + "loss": 0.354, + "mean_token_accuracy": 0.8738872408866882, + "num_tokens": 522997367.0, + "step": 13470 + }, + { + "epoch": 1.7136496628927618, + "grad_norm": 2.974008560180664, + "learning_rate": 1e-06, + "loss": 0.3375, + "mean_token_accuracy": 0.8786989450454712, + "num_tokens": 523033025.0, + "step": 13471 + }, + { + "epoch": 1.7137768731713523, + "grad_norm": 2.7003226280212402, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8907867074012756, + "num_tokens": 523069761.0, + "step": 13472 + }, + { + "epoch": 1.7139040834499428, + "grad_norm": 2.5508036613464355, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8855416178703308, + "num_tokens": 523108666.0, + "step": 13473 + }, + { + "epoch": 1.7140312937285334, + "grad_norm": 2.5433034896850586, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8988622426986694, + "num_tokens": 523145106.0, + "step": 13474 + }, + { + "epoch": 1.7141585040071239, + "grad_norm": 2.32800555229187, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8987926840782166, + "num_tokens": 523191042.0, + "step": 13475 + }, + { + "epoch": 1.7142857142857144, + "grad_norm": 2.808422803878784, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8877890110015869, + "num_tokens": 523227430.0, + "step": 13476 + }, + { + "epoch": 1.714412924564305, + "grad_norm": 2.5293023586273193, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8940538167953491, + "num_tokens": 523267734.0, + "step": 13477 + }, + { + "epoch": 1.7145401348428955, + "grad_norm": 2.492220163345337, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9021878242492676, + "num_tokens": 523306628.0, + "step": 13478 + }, + { + "epoch": 1.7146673451214858, + "grad_norm": 2.91869854927063, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8873643279075623, + "num_tokens": 523339481.0, + "step": 13479 + }, + { + "epoch": 1.7147945554000763, + "grad_norm": 2.4297423362731934, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8920533061027527, + "num_tokens": 523383744.0, + "step": 13480 + }, + { + "epoch": 1.7149217656786668, + "grad_norm": 2.5982131958007812, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8966448903083801, + "num_tokens": 523423639.0, + "step": 13481 + }, + { + "epoch": 1.7150489759572574, + "grad_norm": 2.7482762336730957, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8848612308502197, + "num_tokens": 523459735.0, + "step": 13482 + }, + { + "epoch": 1.7151761862358479, + "grad_norm": 2.745279550552368, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.898576021194458, + "num_tokens": 523491649.0, + "step": 13483 + }, + { + "epoch": 1.7153033965144384, + "grad_norm": 2.452255964279175, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8915954232215881, + "num_tokens": 523536410.0, + "step": 13484 + }, + { + "epoch": 1.7154306067930287, + "grad_norm": 2.901085138320923, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8931764960289001, + "num_tokens": 523570796.0, + "step": 13485 + }, + { + "epoch": 1.7155578170716193, + "grad_norm": 2.5283501148223877, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8942907452583313, + "num_tokens": 523615224.0, + "step": 13486 + }, + { + "epoch": 1.7156850273502098, + "grad_norm": 2.438765287399292, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8982875347137451, + "num_tokens": 523656293.0, + "step": 13487 + }, + { + "epoch": 1.7158122376288003, + "grad_norm": 3.014462471008301, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8828837871551514, + "num_tokens": 523690073.0, + "step": 13488 + }, + { + "epoch": 1.7159394479073908, + "grad_norm": 2.466212749481201, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8883897066116333, + "num_tokens": 523731776.0, + "step": 13489 + }, + { + "epoch": 1.7160666581859814, + "grad_norm": 2.947597026824951, + "learning_rate": 1e-06, + "loss": 0.3339, + "mean_token_accuracy": 0.8788091540336609, + "num_tokens": 523766388.0, + "step": 13490 + }, + { + "epoch": 1.716193868464572, + "grad_norm": 2.672893762588501, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9014993906021118, + "num_tokens": 523800533.0, + "step": 13491 + }, + { + "epoch": 1.7163210787431624, + "grad_norm": 2.442485809326172, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8945745229721069, + "num_tokens": 523846306.0, + "step": 13492 + }, + { + "epoch": 1.716448289021753, + "grad_norm": 2.6274423599243164, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8902398347854614, + "num_tokens": 523883818.0, + "step": 13493 + }, + { + "epoch": 1.7165754993003435, + "grad_norm": 2.643289566040039, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8900163173675537, + "num_tokens": 523920790.0, + "step": 13494 + }, + { + "epoch": 1.716702709578934, + "grad_norm": 2.673452138900757, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8911734819412231, + "num_tokens": 523957274.0, + "step": 13495 + }, + { + "epoch": 1.7168299198575245, + "grad_norm": 2.740985870361328, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8929531574249268, + "num_tokens": 523991477.0, + "step": 13496 + }, + { + "epoch": 1.716957130136115, + "grad_norm": 2.458679437637329, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.886627733707428, + "num_tokens": 524035011.0, + "step": 13497 + }, + { + "epoch": 1.7170843404147056, + "grad_norm": 2.465418815612793, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8990047574043274, + "num_tokens": 524075277.0, + "step": 13498 + }, + { + "epoch": 1.7172115506932961, + "grad_norm": 2.572021484375, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8895080089569092, + "num_tokens": 524114979.0, + "step": 13499 + }, + { + "epoch": 1.7173387609718866, + "grad_norm": 2.5677759647369385, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.893932580947876, + "num_tokens": 524153407.0, + "step": 13500 + }, + { + "epoch": 1.7174659712504772, + "grad_norm": 2.43851375579834, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8952624201774597, + "num_tokens": 524195581.0, + "step": 13501 + }, + { + "epoch": 1.7175931815290677, + "grad_norm": 2.750281810760498, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8783904314041138, + "num_tokens": 524236361.0, + "step": 13502 + }, + { + "epoch": 1.717720391807658, + "grad_norm": 2.3803629875183105, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9009029269218445, + "num_tokens": 524278064.0, + "step": 13503 + }, + { + "epoch": 1.7178476020862485, + "grad_norm": 2.528599262237549, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9059846997261047, + "num_tokens": 524316392.0, + "step": 13504 + }, + { + "epoch": 1.717974812364839, + "grad_norm": 2.263862133026123, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8936104774475098, + "num_tokens": 524364332.0, + "step": 13505 + }, + { + "epoch": 1.7181020226434296, + "grad_norm": 2.6090087890625, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8945574164390564, + "num_tokens": 524402935.0, + "step": 13506 + }, + { + "epoch": 1.7182292329220201, + "grad_norm": 2.9108293056488037, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8820962905883789, + "num_tokens": 524441034.0, + "step": 13507 + }, + { + "epoch": 1.7183564432006107, + "grad_norm": 2.445725440979004, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.9006261825561523, + "num_tokens": 524486144.0, + "step": 13508 + }, + { + "epoch": 1.718483653479201, + "grad_norm": 2.49877667427063, + "learning_rate": 1e-06, + "loss": 0.3338, + "mean_token_accuracy": 0.8824117183685303, + "num_tokens": 524531646.0, + "step": 13509 + }, + { + "epoch": 1.7186108637577915, + "grad_norm": 2.680358648300171, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8911247849464417, + "num_tokens": 524569839.0, + "step": 13510 + }, + { + "epoch": 1.718738074036382, + "grad_norm": 2.728203535079956, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8847030401229858, + "num_tokens": 524611306.0, + "step": 13511 + }, + { + "epoch": 1.7188652843149725, + "grad_norm": 2.577523708343506, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8814521431922913, + "num_tokens": 524653439.0, + "step": 13512 + }, + { + "epoch": 1.718992494593563, + "grad_norm": 2.655674934387207, + "learning_rate": 1e-06, + "loss": 0.331, + "mean_token_accuracy": 0.8817650675773621, + "num_tokens": 524694561.0, + "step": 13513 + }, + { + "epoch": 1.7191197048721536, + "grad_norm": 2.604567527770996, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.892180323600769, + "num_tokens": 524733006.0, + "step": 13514 + }, + { + "epoch": 1.7192469151507441, + "grad_norm": 2.675137758255005, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9002258777618408, + "num_tokens": 524764717.0, + "step": 13515 + }, + { + "epoch": 1.7193741254293347, + "grad_norm": 2.803126096725464, + "learning_rate": 1e-06, + "loss": 0.3697, + "mean_token_accuracy": 0.8696688413619995, + "num_tokens": 524803649.0, + "step": 13516 + }, + { + "epoch": 1.7195013357079252, + "grad_norm": 2.278791904449463, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8934430480003357, + "num_tokens": 524850488.0, + "step": 13517 + }, + { + "epoch": 1.7196285459865157, + "grad_norm": 2.652740716934204, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8896344304084778, + "num_tokens": 524887382.0, + "step": 13518 + }, + { + "epoch": 1.7197557562651062, + "grad_norm": 2.569377899169922, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8985810875892639, + "num_tokens": 524925694.0, + "step": 13519 + }, + { + "epoch": 1.7198829665436968, + "grad_norm": 2.375948905944824, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8917151093482971, + "num_tokens": 524972692.0, + "step": 13520 + }, + { + "epoch": 1.7200101768222873, + "grad_norm": 2.291961908340454, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8951783180236816, + "num_tokens": 525021386.0, + "step": 13521 + }, + { + "epoch": 1.7201373871008778, + "grad_norm": 2.6346659660339355, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8871552348136902, + "num_tokens": 525061268.0, + "step": 13522 + }, + { + "epoch": 1.7202645973794684, + "grad_norm": 2.467717409133911, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8920289874076843, + "num_tokens": 525105598.0, + "step": 13523 + }, + { + "epoch": 1.7203918076580589, + "grad_norm": 2.611755609512329, + "learning_rate": 1e-06, + "loss": 0.3571, + "mean_token_accuracy": 0.8723899126052856, + "num_tokens": 525147065.0, + "step": 13524 + }, + { + "epoch": 1.7205190179366494, + "grad_norm": 2.5301055908203125, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9072794914245605, + "num_tokens": 525183053.0, + "step": 13525 + }, + { + "epoch": 1.72064622821524, + "grad_norm": 2.4460697174072266, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8856208324432373, + "num_tokens": 525228215.0, + "step": 13526 + }, + { + "epoch": 1.7207734384938305, + "grad_norm": 2.578531503677368, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.877338171005249, + "num_tokens": 525270015.0, + "step": 13527 + }, + { + "epoch": 1.7209006487724208, + "grad_norm": 2.9649856090545654, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8927221894264221, + "num_tokens": 525303423.0, + "step": 13528 + }, + { + "epoch": 1.7210278590510113, + "grad_norm": 2.5051004886627197, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8963305950164795, + "num_tokens": 525344588.0, + "step": 13529 + }, + { + "epoch": 1.7211550693296018, + "grad_norm": 2.9918925762176514, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.888697624206543, + "num_tokens": 525376795.0, + "step": 13530 + }, + { + "epoch": 1.7212822796081924, + "grad_norm": 2.8412022590637207, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8956384658813477, + "num_tokens": 525407696.0, + "step": 13531 + }, + { + "epoch": 1.7214094898867829, + "grad_norm": 2.888385057449341, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8793888092041016, + "num_tokens": 525441379.0, + "step": 13532 + }, + { + "epoch": 1.7215367001653734, + "grad_norm": 2.55330753326416, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.897099494934082, + "num_tokens": 525486906.0, + "step": 13533 + }, + { + "epoch": 1.7216639104439637, + "grad_norm": 2.5308585166931152, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8926774263381958, + "num_tokens": 525526380.0, + "step": 13534 + }, + { + "epoch": 1.7217911207225542, + "grad_norm": 2.5169947147369385, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8856721520423889, + "num_tokens": 525566882.0, + "step": 13535 + }, + { + "epoch": 1.7219183310011448, + "grad_norm": 2.74052357673645, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8942933678627014, + "num_tokens": 525602424.0, + "step": 13536 + }, + { + "epoch": 1.7220455412797353, + "grad_norm": 2.5311756134033203, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9012888073921204, + "num_tokens": 525640078.0, + "step": 13537 + }, + { + "epoch": 1.7221727515583258, + "grad_norm": 2.4249379634857178, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8970057964324951, + "num_tokens": 525683385.0, + "step": 13538 + }, + { + "epoch": 1.7222999618369164, + "grad_norm": 2.5534110069274902, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8992037773132324, + "num_tokens": 525721811.0, + "step": 13539 + }, + { + "epoch": 1.7224271721155069, + "grad_norm": 2.882002830505371, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8810904026031494, + "num_tokens": 525758557.0, + "step": 13540 + }, + { + "epoch": 1.7225543823940974, + "grad_norm": 2.8686602115631104, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8932266235351562, + "num_tokens": 525789444.0, + "step": 13541 + }, + { + "epoch": 1.722681592672688, + "grad_norm": 2.505481481552124, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8943062424659729, + "num_tokens": 525829961.0, + "step": 13542 + }, + { + "epoch": 1.7228088029512785, + "grad_norm": 2.8910229206085205, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.883931040763855, + "num_tokens": 525866023.0, + "step": 13543 + }, + { + "epoch": 1.722936013229869, + "grad_norm": 2.7259395122528076, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.8981908559799194, + "num_tokens": 525899436.0, + "step": 13544 + }, + { + "epoch": 1.7230632235084595, + "grad_norm": 2.607393264770508, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8970081806182861, + "num_tokens": 525938811.0, + "step": 13545 + }, + { + "epoch": 1.72319043378705, + "grad_norm": 2.4406557083129883, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8995835781097412, + "num_tokens": 525980247.0, + "step": 13546 + }, + { + "epoch": 1.7233176440656406, + "grad_norm": 2.503650188446045, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9035173058509827, + "num_tokens": 526015493.0, + "step": 13547 + }, + { + "epoch": 1.7234448543442311, + "grad_norm": 2.4540302753448486, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8758764266967773, + "num_tokens": 526062755.0, + "step": 13548 + }, + { + "epoch": 1.7235720646228216, + "grad_norm": 2.6073546409606934, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8838624954223633, + "num_tokens": 526103669.0, + "step": 13549 + }, + { + "epoch": 1.7236992749014122, + "grad_norm": 2.6516523361206055, + "learning_rate": 1e-06, + "loss": 0.3273, + "mean_token_accuracy": 0.8822822570800781, + "num_tokens": 526144320.0, + "step": 13550 + }, + { + "epoch": 1.7238264851800027, + "grad_norm": 2.843449831008911, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8845398426055908, + "num_tokens": 526177381.0, + "step": 13551 + }, + { + "epoch": 1.723953695458593, + "grad_norm": 2.7055447101593018, + "learning_rate": 1e-06, + "loss": 0.3698, + "mean_token_accuracy": 0.8684552311897278, + "num_tokens": 526222657.0, + "step": 13552 + }, + { + "epoch": 1.7240809057371835, + "grad_norm": 2.661475896835327, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8884674310684204, + "num_tokens": 526259026.0, + "step": 13553 + }, + { + "epoch": 1.724208116015774, + "grad_norm": 2.4570767879486084, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.894545316696167, + "num_tokens": 526300499.0, + "step": 13554 + }, + { + "epoch": 1.7243353262943646, + "grad_norm": 2.71370267868042, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8951885104179382, + "num_tokens": 526333900.0, + "step": 13555 + }, + { + "epoch": 1.7244625365729551, + "grad_norm": 2.62127685546875, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8931866884231567, + "num_tokens": 526368585.0, + "step": 13556 + }, + { + "epoch": 1.7245897468515456, + "grad_norm": 2.893544912338257, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8928577899932861, + "num_tokens": 526402148.0, + "step": 13557 + }, + { + "epoch": 1.724716957130136, + "grad_norm": 2.770752191543579, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8919700384140015, + "num_tokens": 526438514.0, + "step": 13558 + }, + { + "epoch": 1.7248441674087265, + "grad_norm": 2.649183988571167, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8913325071334839, + "num_tokens": 526475597.0, + "step": 13559 + }, + { + "epoch": 1.724971377687317, + "grad_norm": 2.763916015625, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8803084492683411, + "num_tokens": 526514076.0, + "step": 13560 + }, + { + "epoch": 1.7250985879659075, + "grad_norm": 2.809603452682495, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8879823088645935, + "num_tokens": 526548662.0, + "step": 13561 + }, + { + "epoch": 1.725225798244498, + "grad_norm": 2.6076457500457764, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8811392784118652, + "num_tokens": 526590289.0, + "step": 13562 + }, + { + "epoch": 1.7253530085230886, + "grad_norm": 2.880794048309326, + "learning_rate": 1e-06, + "loss": 0.3388, + "mean_token_accuracy": 0.878193736076355, + "num_tokens": 526626390.0, + "step": 13563 + }, + { + "epoch": 1.7254802188016791, + "grad_norm": 2.4430301189422607, + "learning_rate": 1e-06, + "loss": 0.3184, + "mean_token_accuracy": 0.8853570222854614, + "num_tokens": 526668488.0, + "step": 13564 + }, + { + "epoch": 1.7256074290802697, + "grad_norm": 2.5791015625, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8858852386474609, + "num_tokens": 526708263.0, + "step": 13565 + }, + { + "epoch": 1.7257346393588602, + "grad_norm": 2.4079036712646484, + "learning_rate": 1e-06, + "loss": 0.3216, + "mean_token_accuracy": 0.8852365612983704, + "num_tokens": 526752876.0, + "step": 13566 + }, + { + "epoch": 1.7258618496374507, + "grad_norm": 2.7082667350769043, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8811825513839722, + "num_tokens": 526789825.0, + "step": 13567 + }, + { + "epoch": 1.7259890599160412, + "grad_norm": 2.993828296661377, + "learning_rate": 1e-06, + "loss": 0.3509, + "mean_token_accuracy": 0.87522292137146, + "num_tokens": 526822290.0, + "step": 13568 + }, + { + "epoch": 1.7261162701946318, + "grad_norm": 2.446215867996216, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8913789391517639, + "num_tokens": 526863026.0, + "step": 13569 + }, + { + "epoch": 1.7262434804732223, + "grad_norm": 2.75260591506958, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8870803713798523, + "num_tokens": 526901294.0, + "step": 13570 + }, + { + "epoch": 1.7263706907518128, + "grad_norm": 2.83005690574646, + "learning_rate": 1e-06, + "loss": 0.3467, + "mean_token_accuracy": 0.8750234842300415, + "num_tokens": 526937454.0, + "step": 13571 + }, + { + "epoch": 1.7264979010304033, + "grad_norm": 2.5834717750549316, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8901342153549194, + "num_tokens": 526977478.0, + "step": 13572 + }, + { + "epoch": 1.7266251113089939, + "grad_norm": 2.6708548069000244, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8820869326591492, + "num_tokens": 527015078.0, + "step": 13573 + }, + { + "epoch": 1.7267523215875844, + "grad_norm": 2.4841442108154297, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8911480903625488, + "num_tokens": 527056042.0, + "step": 13574 + }, + { + "epoch": 1.726879531866175, + "grad_norm": 2.5550272464752197, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8914117217063904, + "num_tokens": 527096842.0, + "step": 13575 + }, + { + "epoch": 1.7270067421447655, + "grad_norm": 2.67366099357605, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8841501474380493, + "num_tokens": 527136693.0, + "step": 13576 + }, + { + "epoch": 1.7271339524233558, + "grad_norm": 2.366558313369751, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9001649618148804, + "num_tokens": 527179220.0, + "step": 13577 + }, + { + "epoch": 1.7272611627019463, + "grad_norm": 2.76218843460083, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8857912421226501, + "num_tokens": 527215055.0, + "step": 13578 + }, + { + "epoch": 1.7273883729805368, + "grad_norm": 2.5213162899017334, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8994825482368469, + "num_tokens": 527253116.0, + "step": 13579 + }, + { + "epoch": 1.7275155832591274, + "grad_norm": 2.3684754371643066, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9023886322975159, + "num_tokens": 527292881.0, + "step": 13580 + }, + { + "epoch": 1.7276427935377179, + "grad_norm": 2.7519137859344482, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8777438998222351, + "num_tokens": 527328849.0, + "step": 13581 + }, + { + "epoch": 1.7277700038163084, + "grad_norm": 2.5689380168914795, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9024493098258972, + "num_tokens": 527363836.0, + "step": 13582 + }, + { + "epoch": 1.7278972140948987, + "grad_norm": 2.6132194995880127, + "learning_rate": 1e-06, + "loss": 0.3398, + "mean_token_accuracy": 0.881646990776062, + "num_tokens": 527405808.0, + "step": 13583 + }, + { + "epoch": 1.7280244243734892, + "grad_norm": 2.6114330291748047, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.899222195148468, + "num_tokens": 527444632.0, + "step": 13584 + }, + { + "epoch": 1.7281516346520798, + "grad_norm": 2.8042209148406982, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8899357318878174, + "num_tokens": 527482496.0, + "step": 13585 + }, + { + "epoch": 1.7282788449306703, + "grad_norm": 2.711301326751709, + "learning_rate": 1e-06, + "loss": 0.3395, + "mean_token_accuracy": 0.8800244331359863, + "num_tokens": 527521859.0, + "step": 13586 + }, + { + "epoch": 1.7284060552092608, + "grad_norm": 2.4652984142303467, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8891328573226929, + "num_tokens": 527563002.0, + "step": 13587 + }, + { + "epoch": 1.7285332654878514, + "grad_norm": 2.593303918838501, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.894416093826294, + "num_tokens": 527599907.0, + "step": 13588 + }, + { + "epoch": 1.7286604757664419, + "grad_norm": 2.395224094390869, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8853756189346313, + "num_tokens": 527648154.0, + "step": 13589 + }, + { + "epoch": 1.7287876860450324, + "grad_norm": 2.5259206295013428, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8927974700927734, + "num_tokens": 527685659.0, + "step": 13590 + }, + { + "epoch": 1.728914896323623, + "grad_norm": 2.761847496032715, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8947172164916992, + "num_tokens": 527718176.0, + "step": 13591 + }, + { + "epoch": 1.7290421066022135, + "grad_norm": 2.7719364166259766, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8949329257011414, + "num_tokens": 527752405.0, + "step": 13592 + }, + { + "epoch": 1.729169316880804, + "grad_norm": 2.5971570014953613, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8943608403205872, + "num_tokens": 527791673.0, + "step": 13593 + }, + { + "epoch": 1.7292965271593945, + "grad_norm": 2.584979295730591, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.8969290256500244, + "num_tokens": 527830802.0, + "step": 13594 + }, + { + "epoch": 1.729423737437985, + "grad_norm": 2.543856143951416, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8842343091964722, + "num_tokens": 527871653.0, + "step": 13595 + }, + { + "epoch": 1.7295509477165756, + "grad_norm": 2.6863057613372803, + "learning_rate": 1e-06, + "loss": 0.3648, + "mean_token_accuracy": 0.8735944628715515, + "num_tokens": 527918531.0, + "step": 13596 + }, + { + "epoch": 1.729678157995166, + "grad_norm": 2.5882883071899414, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8911165595054626, + "num_tokens": 527958182.0, + "step": 13597 + }, + { + "epoch": 1.7298053682737566, + "grad_norm": 2.4742610454559326, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8872268795967102, + "num_tokens": 528000573.0, + "step": 13598 + }, + { + "epoch": 1.7299325785523472, + "grad_norm": 2.517627000808716, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8898837566375732, + "num_tokens": 528041834.0, + "step": 13599 + }, + { + "epoch": 1.7300597888309377, + "grad_norm": 2.6349260807037354, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.886581301689148, + "num_tokens": 528079071.0, + "step": 13600 + }, + { + "epoch": 1.730186999109528, + "grad_norm": 2.892524242401123, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8819347023963928, + "num_tokens": 528113762.0, + "step": 13601 + }, + { + "epoch": 1.7303142093881185, + "grad_norm": 2.802321195602417, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8834632039070129, + "num_tokens": 528152865.0, + "step": 13602 + }, + { + "epoch": 1.730441419666709, + "grad_norm": 2.730665683746338, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8859204053878784, + "num_tokens": 528189491.0, + "step": 13603 + }, + { + "epoch": 1.7305686299452996, + "grad_norm": 2.472810983657837, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9059193134307861, + "num_tokens": 528229657.0, + "step": 13604 + }, + { + "epoch": 1.7306958402238901, + "grad_norm": 2.5671072006225586, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8859733939170837, + "num_tokens": 528272736.0, + "step": 13605 + }, + { + "epoch": 1.7308230505024806, + "grad_norm": 2.4349899291992188, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9020472168922424, + "num_tokens": 528310682.0, + "step": 13606 + }, + { + "epoch": 1.730950260781071, + "grad_norm": 2.746896743774414, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8795003890991211, + "num_tokens": 528346566.0, + "step": 13607 + }, + { + "epoch": 1.7310774710596615, + "grad_norm": 2.467254400253296, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9003859162330627, + "num_tokens": 528385389.0, + "step": 13608 + }, + { + "epoch": 1.731204681338252, + "grad_norm": 2.4567034244537354, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8936934471130371, + "num_tokens": 528428623.0, + "step": 13609 + }, + { + "epoch": 1.7313318916168425, + "grad_norm": 2.6903350353240967, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.8713887333869934, + "num_tokens": 528471249.0, + "step": 13610 + }, + { + "epoch": 1.731459101895433, + "grad_norm": 2.4298744201660156, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8873718976974487, + "num_tokens": 528516851.0, + "step": 13611 + }, + { + "epoch": 1.7315863121740236, + "grad_norm": 2.8022079467773438, + "learning_rate": 1e-06, + "loss": 0.3447, + "mean_token_accuracy": 0.8785614371299744, + "num_tokens": 528554856.0, + "step": 13612 + }, + { + "epoch": 1.7317135224526141, + "grad_norm": 2.7836012840270996, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8808289766311646, + "num_tokens": 528591791.0, + "step": 13613 + }, + { + "epoch": 1.7318407327312046, + "grad_norm": 2.7197165489196777, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8905360102653503, + "num_tokens": 528626669.0, + "step": 13614 + }, + { + "epoch": 1.7319679430097952, + "grad_norm": 2.616623640060425, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8905178308486938, + "num_tokens": 528666222.0, + "step": 13615 + }, + { + "epoch": 1.7320951532883857, + "grad_norm": 2.7927393913269043, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.904229998588562, + "num_tokens": 528704641.0, + "step": 13616 + }, + { + "epoch": 1.7322223635669762, + "grad_norm": 2.6535778045654297, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8851209878921509, + "num_tokens": 528744903.0, + "step": 13617 + }, + { + "epoch": 1.7323495738455668, + "grad_norm": 2.7415030002593994, + "learning_rate": 1e-06, + "loss": 0.3732, + "mean_token_accuracy": 0.8693169355392456, + "num_tokens": 528788545.0, + "step": 13618 + }, + { + "epoch": 1.7324767841241573, + "grad_norm": 2.497861623764038, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8863579630851746, + "num_tokens": 528831244.0, + "step": 13619 + }, + { + "epoch": 1.7326039944027478, + "grad_norm": 2.7621614933013916, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8862300515174866, + "num_tokens": 528868360.0, + "step": 13620 + }, + { + "epoch": 1.7327312046813383, + "grad_norm": 2.7559614181518555, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8869855403900146, + "num_tokens": 528907167.0, + "step": 13621 + }, + { + "epoch": 1.7328584149599289, + "grad_norm": 2.862287759780884, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8992140293121338, + "num_tokens": 528946081.0, + "step": 13622 + }, + { + "epoch": 1.7329856252385194, + "grad_norm": 2.528372287750244, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8888574242591858, + "num_tokens": 528989445.0, + "step": 13623 + }, + { + "epoch": 1.73311283551711, + "grad_norm": 2.5144593715667725, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8875731825828552, + "num_tokens": 529030616.0, + "step": 13624 + }, + { + "epoch": 1.7332400457957005, + "grad_norm": 2.4075255393981934, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9034306406974792, + "num_tokens": 529071190.0, + "step": 13625 + }, + { + "epoch": 1.7333672560742908, + "grad_norm": 2.791614055633545, + "learning_rate": 1e-06, + "loss": 0.3579, + "mean_token_accuracy": 0.871711254119873, + "num_tokens": 529114487.0, + "step": 13626 + }, + { + "epoch": 1.7334944663528813, + "grad_norm": 2.552013397216797, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.9007388353347778, + "num_tokens": 529153383.0, + "step": 13627 + }, + { + "epoch": 1.7336216766314718, + "grad_norm": 2.4286224842071533, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8849035501480103, + "num_tokens": 529196326.0, + "step": 13628 + }, + { + "epoch": 1.7337488869100623, + "grad_norm": 2.343904733657837, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8889084458351135, + "num_tokens": 529245027.0, + "step": 13629 + }, + { + "epoch": 1.7338760971886529, + "grad_norm": 2.7265827655792236, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8992794752120972, + "num_tokens": 529284011.0, + "step": 13630 + }, + { + "epoch": 1.7340033074672434, + "grad_norm": 2.483001232147217, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9004319310188293, + "num_tokens": 529325412.0, + "step": 13631 + }, + { + "epoch": 1.7341305177458337, + "grad_norm": 2.641136407852173, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.891362190246582, + "num_tokens": 529363941.0, + "step": 13632 + }, + { + "epoch": 1.7342577280244242, + "grad_norm": 2.537080764770508, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.89399254322052, + "num_tokens": 529404725.0, + "step": 13633 + }, + { + "epoch": 1.7343849383030148, + "grad_norm": 2.4064340591430664, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8955897092819214, + "num_tokens": 529451149.0, + "step": 13634 + }, + { + "epoch": 1.7345121485816053, + "grad_norm": 2.6580910682678223, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8914117813110352, + "num_tokens": 529487854.0, + "step": 13635 + }, + { + "epoch": 1.7346393588601958, + "grad_norm": 2.805798292160034, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8932896852493286, + "num_tokens": 529520899.0, + "step": 13636 + }, + { + "epoch": 1.7347665691387864, + "grad_norm": 2.4735782146453857, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9065654277801514, + "num_tokens": 529560186.0, + "step": 13637 + }, + { + "epoch": 1.7348937794173769, + "grad_norm": 2.6030361652374268, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8898050785064697, + "num_tokens": 529597336.0, + "step": 13638 + }, + { + "epoch": 1.7350209896959674, + "grad_norm": 2.7536118030548096, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8929786086082458, + "num_tokens": 529635549.0, + "step": 13639 + }, + { + "epoch": 1.735148199974558, + "grad_norm": 2.557561159133911, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8838135004043579, + "num_tokens": 529674620.0, + "step": 13640 + }, + { + "epoch": 1.7352754102531485, + "grad_norm": 2.683180570602417, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8927339315414429, + "num_tokens": 529712695.0, + "step": 13641 + }, + { + "epoch": 1.735402620531739, + "grad_norm": 2.8824615478515625, + "learning_rate": 1e-06, + "loss": 0.3604, + "mean_token_accuracy": 0.871992290019989, + "num_tokens": 529750452.0, + "step": 13642 + }, + { + "epoch": 1.7355298308103295, + "grad_norm": 2.691615581512451, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.889396607875824, + "num_tokens": 529788304.0, + "step": 13643 + }, + { + "epoch": 1.73565704108892, + "grad_norm": 2.7670445442199707, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8902416229248047, + "num_tokens": 529823681.0, + "step": 13644 + }, + { + "epoch": 1.7357842513675106, + "grad_norm": 2.7022342681884766, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8926080465316772, + "num_tokens": 529858715.0, + "step": 13645 + }, + { + "epoch": 1.735911461646101, + "grad_norm": 2.5831873416900635, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8901659846305847, + "num_tokens": 529897334.0, + "step": 13646 + }, + { + "epoch": 1.7360386719246916, + "grad_norm": 2.5332701206207275, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.894748330116272, + "num_tokens": 529937802.0, + "step": 13647 + }, + { + "epoch": 1.7361658822032822, + "grad_norm": 2.5222864151000977, + "learning_rate": 1e-06, + "loss": 0.3116, + "mean_token_accuracy": 0.8890212774276733, + "num_tokens": 529977768.0, + "step": 13648 + }, + { + "epoch": 1.7362930924818727, + "grad_norm": 2.6594247817993164, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8965701460838318, + "num_tokens": 530012907.0, + "step": 13649 + }, + { + "epoch": 1.736420302760463, + "grad_norm": 2.5008444786071777, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8913735151290894, + "num_tokens": 530055726.0, + "step": 13650 + }, + { + "epoch": 1.7365475130390535, + "grad_norm": 2.2561089992523193, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8919709920883179, + "num_tokens": 530104405.0, + "step": 13651 + }, + { + "epoch": 1.736674723317644, + "grad_norm": 2.4505832195281982, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.9001456499099731, + "num_tokens": 530145316.0, + "step": 13652 + }, + { + "epoch": 1.7368019335962346, + "grad_norm": 2.436436176300049, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8888646960258484, + "num_tokens": 530192067.0, + "step": 13653 + }, + { + "epoch": 1.736929143874825, + "grad_norm": 2.534729480743408, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8954177498817444, + "num_tokens": 530229743.0, + "step": 13654 + }, + { + "epoch": 1.7370563541534156, + "grad_norm": 2.684046745300293, + "learning_rate": 1e-06, + "loss": 0.3356, + "mean_token_accuracy": 0.8816008567810059, + "num_tokens": 530269095.0, + "step": 13655 + }, + { + "epoch": 1.737183564432006, + "grad_norm": 2.528555393218994, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8974119424819946, + "num_tokens": 530306867.0, + "step": 13656 + }, + { + "epoch": 1.7373107747105965, + "grad_norm": 2.7818899154663086, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.888603687286377, + "num_tokens": 530341375.0, + "step": 13657 + }, + { + "epoch": 1.737437984989187, + "grad_norm": 2.758244276046753, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.88785320520401, + "num_tokens": 530378055.0, + "step": 13658 + }, + { + "epoch": 1.7375651952677775, + "grad_norm": 2.832850217819214, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8844393491744995, + "num_tokens": 530416217.0, + "step": 13659 + }, + { + "epoch": 1.737692405546368, + "grad_norm": 2.698289155960083, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8920658230781555, + "num_tokens": 530453029.0, + "step": 13660 + }, + { + "epoch": 1.7378196158249586, + "grad_norm": 2.7101023197174072, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8882983922958374, + "num_tokens": 530488574.0, + "step": 13661 + }, + { + "epoch": 1.7379468261035491, + "grad_norm": 2.3448519706726074, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8895787000656128, + "num_tokens": 530534968.0, + "step": 13662 + }, + { + "epoch": 1.7380740363821396, + "grad_norm": 2.4629180431365967, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9001567363739014, + "num_tokens": 530571950.0, + "step": 13663 + }, + { + "epoch": 1.7382012466607302, + "grad_norm": 2.419529438018799, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8993847370147705, + "num_tokens": 530614763.0, + "step": 13664 + }, + { + "epoch": 1.7383284569393207, + "grad_norm": 2.554556369781494, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9010186195373535, + "num_tokens": 530650422.0, + "step": 13665 + }, + { + "epoch": 1.7384556672179112, + "grad_norm": 2.7403032779693604, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8888217210769653, + "num_tokens": 530684977.0, + "step": 13666 + }, + { + "epoch": 1.7385828774965018, + "grad_norm": 2.541626453399658, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8848979473114014, + "num_tokens": 530728929.0, + "step": 13667 + }, + { + "epoch": 1.7387100877750923, + "grad_norm": 2.3311634063720703, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9047229886054993, + "num_tokens": 530773345.0, + "step": 13668 + }, + { + "epoch": 1.7388372980536828, + "grad_norm": 3.0921287536621094, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8781563639640808, + "num_tokens": 530805742.0, + "step": 13669 + }, + { + "epoch": 1.7389645083322733, + "grad_norm": 2.659902334213257, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8843131065368652, + "num_tokens": 530845560.0, + "step": 13670 + }, + { + "epoch": 1.7390917186108639, + "grad_norm": 2.731074094772339, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8839459419250488, + "num_tokens": 530883633.0, + "step": 13671 + }, + { + "epoch": 1.7392189288894544, + "grad_norm": 2.8260929584503174, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8857405781745911, + "num_tokens": 530918495.0, + "step": 13672 + }, + { + "epoch": 1.739346139168045, + "grad_norm": 2.3725056648254395, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8962948322296143, + "num_tokens": 530961150.0, + "step": 13673 + }, + { + "epoch": 1.7394733494466355, + "grad_norm": 2.4914190769195557, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9024598598480225, + "num_tokens": 530998377.0, + "step": 13674 + }, + { + "epoch": 1.7396005597252258, + "grad_norm": 2.469174385070801, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9074608683586121, + "num_tokens": 531033201.0, + "step": 13675 + }, + { + "epoch": 1.7397277700038163, + "grad_norm": 2.7772538661956787, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8890799283981323, + "num_tokens": 531068858.0, + "step": 13676 + }, + { + "epoch": 1.7398549802824068, + "grad_norm": 2.4303345680236816, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9017467498779297, + "num_tokens": 531109672.0, + "step": 13677 + }, + { + "epoch": 1.7399821905609973, + "grad_norm": 2.409860372543335, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8986495733261108, + "num_tokens": 531149617.0, + "step": 13678 + }, + { + "epoch": 1.7401094008395879, + "grad_norm": 2.7274138927459717, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8902442455291748, + "num_tokens": 531185911.0, + "step": 13679 + }, + { + "epoch": 1.7402366111181784, + "grad_norm": 2.6556224822998047, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.888931393623352, + "num_tokens": 531224692.0, + "step": 13680 + }, + { + "epoch": 1.7403638213967687, + "grad_norm": 2.886287212371826, + "learning_rate": 1e-06, + "loss": 0.3486, + "mean_token_accuracy": 0.874250054359436, + "num_tokens": 531264572.0, + "step": 13681 + }, + { + "epoch": 1.7404910316753592, + "grad_norm": 2.5957865715026855, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8927531242370605, + "num_tokens": 531301585.0, + "step": 13682 + }, + { + "epoch": 1.7406182419539498, + "grad_norm": 2.786066770553589, + "learning_rate": 1e-06, + "loss": 0.3385, + "mean_token_accuracy": 0.8816247582435608, + "num_tokens": 531340921.0, + "step": 13683 + }, + { + "epoch": 1.7407454522325403, + "grad_norm": 2.5457587242126465, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9035708904266357, + "num_tokens": 531377250.0, + "step": 13684 + }, + { + "epoch": 1.7408726625111308, + "grad_norm": 2.6863787174224854, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8837324380874634, + "num_tokens": 531414802.0, + "step": 13685 + }, + { + "epoch": 1.7409998727897213, + "grad_norm": 2.5377771854400635, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8883158564567566, + "num_tokens": 531454582.0, + "step": 13686 + }, + { + "epoch": 1.7411270830683119, + "grad_norm": 2.7059197425842285, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8923557996749878, + "num_tokens": 531493405.0, + "step": 13687 + }, + { + "epoch": 1.7412542933469024, + "grad_norm": 2.529031276702881, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9045711159706116, + "num_tokens": 531529691.0, + "step": 13688 + }, + { + "epoch": 1.741381503625493, + "grad_norm": 2.620737314224243, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.894252598285675, + "num_tokens": 531565849.0, + "step": 13689 + }, + { + "epoch": 1.7415087139040835, + "grad_norm": 2.7562808990478516, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8950164318084717, + "num_tokens": 531598312.0, + "step": 13690 + }, + { + "epoch": 1.741635924182674, + "grad_norm": 2.651454210281372, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8998115658760071, + "num_tokens": 531632068.0, + "step": 13691 + }, + { + "epoch": 1.7417631344612645, + "grad_norm": 2.7532832622528076, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8961007595062256, + "num_tokens": 531664996.0, + "step": 13692 + }, + { + "epoch": 1.741890344739855, + "grad_norm": 2.829270839691162, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8843700289726257, + "num_tokens": 531700207.0, + "step": 13693 + }, + { + "epoch": 1.7420175550184456, + "grad_norm": 2.6510586738586426, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.894623875617981, + "num_tokens": 531736144.0, + "step": 13694 + }, + { + "epoch": 1.742144765297036, + "grad_norm": 2.687314510345459, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8936331272125244, + "num_tokens": 531770766.0, + "step": 13695 + }, + { + "epoch": 1.7422719755756266, + "grad_norm": 2.849745988845825, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9018855094909668, + "num_tokens": 531801364.0, + "step": 13696 + }, + { + "epoch": 1.7423991858542172, + "grad_norm": 2.6022462844848633, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8823896646499634, + "num_tokens": 531840306.0, + "step": 13697 + }, + { + "epoch": 1.7425263961328077, + "grad_norm": 2.788243055343628, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.883104681968689, + "num_tokens": 531875892.0, + "step": 13698 + }, + { + "epoch": 1.742653606411398, + "grad_norm": 2.7184269428253174, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8883706331253052, + "num_tokens": 531913113.0, + "step": 13699 + }, + { + "epoch": 1.7427808166899885, + "grad_norm": 2.7228806018829346, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8908364772796631, + "num_tokens": 531946767.0, + "step": 13700 + }, + { + "epoch": 1.742908026968579, + "grad_norm": 2.501526355743408, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.894133985042572, + "num_tokens": 531985340.0, + "step": 13701 + }, + { + "epoch": 1.7430352372471696, + "grad_norm": 2.6970365047454834, + "learning_rate": 1e-06, + "loss": 0.3721, + "mean_token_accuracy": 0.8670839667320251, + "num_tokens": 532025742.0, + "step": 13702 + }, + { + "epoch": 1.74316244752576, + "grad_norm": 2.924280881881714, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8881314396858215, + "num_tokens": 532058354.0, + "step": 13703 + }, + { + "epoch": 1.7432896578043506, + "grad_norm": 2.8135101795196533, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8870059847831726, + "num_tokens": 532092892.0, + "step": 13704 + }, + { + "epoch": 1.743416868082941, + "grad_norm": 2.897984027862549, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8807563185691833, + "num_tokens": 532128003.0, + "step": 13705 + }, + { + "epoch": 1.7435440783615315, + "grad_norm": 2.756531238555908, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8953603506088257, + "num_tokens": 532161277.0, + "step": 13706 + }, + { + "epoch": 1.743671288640122, + "grad_norm": 2.607618808746338, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8979998826980591, + "num_tokens": 532197603.0, + "step": 13707 + }, + { + "epoch": 1.7437984989187125, + "grad_norm": 2.7894113063812256, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8951612710952759, + "num_tokens": 532230375.0, + "step": 13708 + }, + { + "epoch": 1.743925709197303, + "grad_norm": 2.661364793777466, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8910121917724609, + "num_tokens": 532265925.0, + "step": 13709 + }, + { + "epoch": 1.7440529194758936, + "grad_norm": 2.522624969482422, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8964676856994629, + "num_tokens": 532304099.0, + "step": 13710 + }, + { + "epoch": 1.744180129754484, + "grad_norm": 2.5818543434143066, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8940625190734863, + "num_tokens": 532344006.0, + "step": 13711 + }, + { + "epoch": 1.7443073400330746, + "grad_norm": 2.6784005165100098, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.892716646194458, + "num_tokens": 532382883.0, + "step": 13712 + }, + { + "epoch": 1.7444345503116652, + "grad_norm": 2.6752769947052, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9032339453697205, + "num_tokens": 532421556.0, + "step": 13713 + }, + { + "epoch": 1.7445617605902557, + "grad_norm": 2.8110756874084473, + "learning_rate": 1e-06, + "loss": 0.3354, + "mean_token_accuracy": 0.8807467222213745, + "num_tokens": 532457511.0, + "step": 13714 + }, + { + "epoch": 1.7446889708688462, + "grad_norm": 2.811800241470337, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8874157667160034, + "num_tokens": 532490268.0, + "step": 13715 + }, + { + "epoch": 1.7448161811474368, + "grad_norm": 2.5733397006988525, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8889822363853455, + "num_tokens": 532531227.0, + "step": 13716 + }, + { + "epoch": 1.7449433914260273, + "grad_norm": 2.7241084575653076, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.892646074295044, + "num_tokens": 532568568.0, + "step": 13717 + }, + { + "epoch": 1.7450706017046178, + "grad_norm": 2.5657880306243896, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8847787380218506, + "num_tokens": 532608639.0, + "step": 13718 + }, + { + "epoch": 1.7451978119832083, + "grad_norm": 2.6922454833984375, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8879076838493347, + "num_tokens": 532645702.0, + "step": 13719 + }, + { + "epoch": 1.7453250222617989, + "grad_norm": 2.760004997253418, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8887835741043091, + "num_tokens": 532679632.0, + "step": 13720 + }, + { + "epoch": 1.7454522325403894, + "grad_norm": 2.7196855545043945, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8867870569229126, + "num_tokens": 532715153.0, + "step": 13721 + }, + { + "epoch": 1.74557944281898, + "grad_norm": 2.7412593364715576, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.88913893699646, + "num_tokens": 532757744.0, + "step": 13722 + }, + { + "epoch": 1.7457066530975704, + "grad_norm": 2.733083724975586, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8959430456161499, + "num_tokens": 532791269.0, + "step": 13723 + }, + { + "epoch": 1.7458338633761608, + "grad_norm": 2.6615471839904785, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8812546730041504, + "num_tokens": 532832741.0, + "step": 13724 + }, + { + "epoch": 1.7459610736547513, + "grad_norm": 2.501300573348999, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8984874486923218, + "num_tokens": 532874559.0, + "step": 13725 + }, + { + "epoch": 1.7460882839333418, + "grad_norm": 2.6805217266082764, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8926360011100769, + "num_tokens": 532912965.0, + "step": 13726 + }, + { + "epoch": 1.7462154942119323, + "grad_norm": 2.6642849445343018, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.891075849533081, + "num_tokens": 532948989.0, + "step": 13727 + }, + { + "epoch": 1.7463427044905229, + "grad_norm": 2.489335298538208, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8937903642654419, + "num_tokens": 532990986.0, + "step": 13728 + }, + { + "epoch": 1.7464699147691134, + "grad_norm": 2.6377949714660645, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8896006941795349, + "num_tokens": 533029257.0, + "step": 13729 + }, + { + "epoch": 1.7465971250477037, + "grad_norm": 2.713883399963379, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8875234127044678, + "num_tokens": 533067601.0, + "step": 13730 + }, + { + "epoch": 1.7467243353262942, + "grad_norm": 2.760176658630371, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.8855738043785095, + "num_tokens": 533109329.0, + "step": 13731 + }, + { + "epoch": 1.7468515456048848, + "grad_norm": 2.6497256755828857, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8991568684577942, + "num_tokens": 533146261.0, + "step": 13732 + }, + { + "epoch": 1.7469787558834753, + "grad_norm": 2.641000270843506, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.890841543674469, + "num_tokens": 533185502.0, + "step": 13733 + }, + { + "epoch": 1.7471059661620658, + "grad_norm": 2.5473294258117676, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.8964688777923584, + "num_tokens": 533225580.0, + "step": 13734 + }, + { + "epoch": 1.7472331764406563, + "grad_norm": 2.7260994911193848, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.883849561214447, + "num_tokens": 533265064.0, + "step": 13735 + }, + { + "epoch": 1.7473603867192469, + "grad_norm": 2.7963056564331055, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8893827199935913, + "num_tokens": 533301227.0, + "step": 13736 + }, + { + "epoch": 1.7474875969978374, + "grad_norm": 2.13181209564209, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9076222777366638, + "num_tokens": 533346019.0, + "step": 13737 + }, + { + "epoch": 1.747614807276428, + "grad_norm": 2.678708791732788, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8936812877655029, + "num_tokens": 533379817.0, + "step": 13738 + }, + { + "epoch": 1.7477420175550185, + "grad_norm": 2.300743341445923, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8935961127281189, + "num_tokens": 533426198.0, + "step": 13739 + }, + { + "epoch": 1.747869227833609, + "grad_norm": 2.5889129638671875, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8971958756446838, + "num_tokens": 533462844.0, + "step": 13740 + }, + { + "epoch": 1.7479964381121995, + "grad_norm": 2.6280386447906494, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8929879665374756, + "num_tokens": 533498878.0, + "step": 13741 + }, + { + "epoch": 1.74812364839079, + "grad_norm": 2.4848427772521973, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8975536227226257, + "num_tokens": 533538277.0, + "step": 13742 + }, + { + "epoch": 1.7482508586693806, + "grad_norm": 2.747589111328125, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.894774854183197, + "num_tokens": 533573883.0, + "step": 13743 + }, + { + "epoch": 1.748378068947971, + "grad_norm": 2.445488691329956, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8856641054153442, + "num_tokens": 533620504.0, + "step": 13744 + }, + { + "epoch": 1.7485052792265616, + "grad_norm": 2.7052602767944336, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8947494029998779, + "num_tokens": 533654442.0, + "step": 13745 + }, + { + "epoch": 1.7486324895051522, + "grad_norm": 2.6478872299194336, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8875369429588318, + "num_tokens": 533693248.0, + "step": 13746 + }, + { + "epoch": 1.7487596997837427, + "grad_norm": 2.8348538875579834, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8893225193023682, + "num_tokens": 533728652.0, + "step": 13747 + }, + { + "epoch": 1.748886910062333, + "grad_norm": 2.456204414367676, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8876553177833557, + "num_tokens": 533773269.0, + "step": 13748 + }, + { + "epoch": 1.7490141203409235, + "grad_norm": 2.364501476287842, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.906667947769165, + "num_tokens": 533813091.0, + "step": 13749 + }, + { + "epoch": 1.749141330619514, + "grad_norm": 2.4629838466644287, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9006931781768799, + "num_tokens": 533850377.0, + "step": 13750 + }, + { + "epoch": 1.7492685408981046, + "grad_norm": 2.7707462310791016, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8847445845603943, + "num_tokens": 533884478.0, + "step": 13751 + }, + { + "epoch": 1.749395751176695, + "grad_norm": 2.486630916595459, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8934227824211121, + "num_tokens": 533922972.0, + "step": 13752 + }, + { + "epoch": 1.7495229614552856, + "grad_norm": 2.547224998474121, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8952776789665222, + "num_tokens": 533962769.0, + "step": 13753 + }, + { + "epoch": 1.749650171733876, + "grad_norm": 2.635934591293335, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8952803015708923, + "num_tokens": 533999157.0, + "step": 13754 + }, + { + "epoch": 1.7497773820124665, + "grad_norm": 2.530219554901123, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8941965103149414, + "num_tokens": 534038502.0, + "step": 13755 + }, + { + "epoch": 1.749904592291057, + "grad_norm": 2.608403444290161, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8834834098815918, + "num_tokens": 534076907.0, + "step": 13756 + }, + { + "epoch": 1.7500318025696475, + "grad_norm": 2.668618679046631, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8993245363235474, + "num_tokens": 534116624.0, + "step": 13757 + }, + { + "epoch": 1.750159012848238, + "grad_norm": 2.2902467250823975, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8977005481719971, + "num_tokens": 534163753.0, + "step": 13758 + }, + { + "epoch": 1.7502862231268286, + "grad_norm": 2.6069531440734863, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8972317576408386, + "num_tokens": 534203309.0, + "step": 13759 + }, + { + "epoch": 1.750413433405419, + "grad_norm": 2.6011505126953125, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8966301679611206, + "num_tokens": 534240043.0, + "step": 13760 + }, + { + "epoch": 1.7505406436840096, + "grad_norm": 2.5150303840637207, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8943701982498169, + "num_tokens": 534279493.0, + "step": 13761 + }, + { + "epoch": 1.7506678539626002, + "grad_norm": 2.382704257965088, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8872241377830505, + "num_tokens": 534324340.0, + "step": 13762 + }, + { + "epoch": 1.7507950642411907, + "grad_norm": 2.8513870239257812, + "learning_rate": 1e-06, + "loss": 0.3646, + "mean_token_accuracy": 0.8706576824188232, + "num_tokens": 534363483.0, + "step": 13763 + }, + { + "epoch": 1.7509222745197812, + "grad_norm": 2.6737794876098633, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8836542367935181, + "num_tokens": 534403026.0, + "step": 13764 + }, + { + "epoch": 1.7510494847983717, + "grad_norm": 2.7907803058624268, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.88856041431427, + "num_tokens": 534437278.0, + "step": 13765 + }, + { + "epoch": 1.7511766950769623, + "grad_norm": 2.6954355239868164, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8825157880783081, + "num_tokens": 534474492.0, + "step": 13766 + }, + { + "epoch": 1.7513039053555528, + "grad_norm": 2.4806432723999023, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8984295129776001, + "num_tokens": 534516269.0, + "step": 13767 + }, + { + "epoch": 1.7514311156341433, + "grad_norm": 2.5613574981689453, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8821176886558533, + "num_tokens": 534558608.0, + "step": 13768 + }, + { + "epoch": 1.7515583259127339, + "grad_norm": 2.6578214168548584, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8899128437042236, + "num_tokens": 534598927.0, + "step": 13769 + }, + { + "epoch": 1.7516855361913244, + "grad_norm": 2.5563063621520996, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8865687847137451, + "num_tokens": 534642092.0, + "step": 13770 + }, + { + "epoch": 1.751812746469915, + "grad_norm": 2.743067979812622, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.8813999891281128, + "num_tokens": 534681335.0, + "step": 13771 + }, + { + "epoch": 1.7519399567485054, + "grad_norm": 2.5902817249298096, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8974212408065796, + "num_tokens": 534723262.0, + "step": 13772 + }, + { + "epoch": 1.7520671670270958, + "grad_norm": 2.6720187664031982, + "learning_rate": 1e-06, + "loss": 0.3371, + "mean_token_accuracy": 0.878626823425293, + "num_tokens": 534767313.0, + "step": 13773 + }, + { + "epoch": 1.7521943773056863, + "grad_norm": 2.667043924331665, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8838769197463989, + "num_tokens": 534807009.0, + "step": 13774 + }, + { + "epoch": 1.7523215875842768, + "grad_norm": 2.5437915325164795, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8971463441848755, + "num_tokens": 534847332.0, + "step": 13775 + }, + { + "epoch": 1.7524487978628673, + "grad_norm": 2.534714460372925, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8969647884368896, + "num_tokens": 534885586.0, + "step": 13776 + }, + { + "epoch": 1.7525760081414579, + "grad_norm": 2.635218858718872, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8942838907241821, + "num_tokens": 534922086.0, + "step": 13777 + }, + { + "epoch": 1.7527032184200484, + "grad_norm": 2.5888609886169434, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8960210680961609, + "num_tokens": 534958707.0, + "step": 13778 + }, + { + "epoch": 1.7528304286986387, + "grad_norm": 2.3981053829193115, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8836061358451843, + "num_tokens": 535004022.0, + "step": 13779 + }, + { + "epoch": 1.7529576389772292, + "grad_norm": 2.5448601245880127, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8987283706665039, + "num_tokens": 535041779.0, + "step": 13780 + }, + { + "epoch": 1.7530848492558198, + "grad_norm": 2.401139259338379, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9035486578941345, + "num_tokens": 535084827.0, + "step": 13781 + }, + { + "epoch": 1.7532120595344103, + "grad_norm": 2.330501079559326, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9024126529693604, + "num_tokens": 535127588.0, + "step": 13782 + }, + { + "epoch": 1.7533392698130008, + "grad_norm": 2.701939582824707, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8815096616744995, + "num_tokens": 535165953.0, + "step": 13783 + }, + { + "epoch": 1.7534664800915913, + "grad_norm": 2.7814245223999023, + "learning_rate": 1e-06, + "loss": 0.3678, + "mean_token_accuracy": 0.8765648603439331, + "num_tokens": 535206278.0, + "step": 13784 + }, + { + "epoch": 1.7535936903701819, + "grad_norm": 2.5315475463867188, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8841136693954468, + "num_tokens": 535246729.0, + "step": 13785 + }, + { + "epoch": 1.7537209006487724, + "grad_norm": 2.282487392425537, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8962526321411133, + "num_tokens": 535290656.0, + "step": 13786 + }, + { + "epoch": 1.753848110927363, + "grad_norm": 2.549208879470825, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8842803239822388, + "num_tokens": 535330337.0, + "step": 13787 + }, + { + "epoch": 1.7539753212059535, + "grad_norm": 2.735867500305176, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8904386758804321, + "num_tokens": 535363974.0, + "step": 13788 + }, + { + "epoch": 1.754102531484544, + "grad_norm": 2.785053253173828, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.889796257019043, + "num_tokens": 535401482.0, + "step": 13789 + }, + { + "epoch": 1.7542297417631345, + "grad_norm": 2.561725616455078, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8955810070037842, + "num_tokens": 535442252.0, + "step": 13790 + }, + { + "epoch": 1.754356952041725, + "grad_norm": 2.607696533203125, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8925914764404297, + "num_tokens": 535481518.0, + "step": 13791 + }, + { + "epoch": 1.7544841623203156, + "grad_norm": 2.959840774536133, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8942586183547974, + "num_tokens": 535511035.0, + "step": 13792 + }, + { + "epoch": 1.754611372598906, + "grad_norm": 2.5971286296844482, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8918246030807495, + "num_tokens": 535551503.0, + "step": 13793 + }, + { + "epoch": 1.7547385828774966, + "grad_norm": 2.598456859588623, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8873550891876221, + "num_tokens": 535596641.0, + "step": 13794 + }, + { + "epoch": 1.7548657931560872, + "grad_norm": 2.5812950134277344, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8868309855461121, + "num_tokens": 535637281.0, + "step": 13795 + }, + { + "epoch": 1.7549930034346777, + "grad_norm": 2.75915789604187, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8751271367073059, + "num_tokens": 535677780.0, + "step": 13796 + }, + { + "epoch": 1.755120213713268, + "grad_norm": 2.4012441635131836, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8961989879608154, + "num_tokens": 535722991.0, + "step": 13797 + }, + { + "epoch": 1.7552474239918585, + "grad_norm": 2.6885087490081787, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8836923241615295, + "num_tokens": 535762905.0, + "step": 13798 + }, + { + "epoch": 1.755374634270449, + "grad_norm": 2.2873451709747314, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.8991806507110596, + "num_tokens": 535804594.0, + "step": 13799 + }, + { + "epoch": 1.7555018445490396, + "grad_norm": 2.6464273929595947, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.896202564239502, + "num_tokens": 535841893.0, + "step": 13800 + }, + { + "epoch": 1.75562905482763, + "grad_norm": 2.525724411010742, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8947014212608337, + "num_tokens": 535880904.0, + "step": 13801 + }, + { + "epoch": 1.7557562651062206, + "grad_norm": 2.43792986869812, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8950360417366028, + "num_tokens": 535922727.0, + "step": 13802 + }, + { + "epoch": 1.755883475384811, + "grad_norm": 2.412259817123413, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8955904245376587, + "num_tokens": 535963708.0, + "step": 13803 + }, + { + "epoch": 1.7560106856634015, + "grad_norm": 2.6485683917999268, + "learning_rate": 1e-06, + "loss": 0.3355, + "mean_token_accuracy": 0.8819193243980408, + "num_tokens": 536005043.0, + "step": 13804 + }, + { + "epoch": 1.756137895941992, + "grad_norm": 2.625279426574707, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8920005559921265, + "num_tokens": 536042708.0, + "step": 13805 + }, + { + "epoch": 1.7562651062205825, + "grad_norm": 2.515728235244751, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8960467576980591, + "num_tokens": 536083467.0, + "step": 13806 + }, + { + "epoch": 1.756392316499173, + "grad_norm": 2.648160457611084, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8884245157241821, + "num_tokens": 536119827.0, + "step": 13807 + }, + { + "epoch": 1.7565195267777636, + "grad_norm": 2.638197183609009, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8941964507102966, + "num_tokens": 536158543.0, + "step": 13808 + }, + { + "epoch": 1.756646737056354, + "grad_norm": 2.6007964611053467, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8925966620445251, + "num_tokens": 536200875.0, + "step": 13809 + }, + { + "epoch": 1.7567739473349446, + "grad_norm": 2.4420182704925537, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8931249380111694, + "num_tokens": 536238440.0, + "step": 13810 + }, + { + "epoch": 1.7569011576135352, + "grad_norm": 2.8392958641052246, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8878390192985535, + "num_tokens": 536272516.0, + "step": 13811 + }, + { + "epoch": 1.7570283678921257, + "grad_norm": 2.6657097339630127, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8835604190826416, + "num_tokens": 536307616.0, + "step": 13812 + }, + { + "epoch": 1.7571555781707162, + "grad_norm": 2.521683931350708, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8948991298675537, + "num_tokens": 536344475.0, + "step": 13813 + }, + { + "epoch": 1.7572827884493067, + "grad_norm": 2.566596508026123, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.89154052734375, + "num_tokens": 536384286.0, + "step": 13814 + }, + { + "epoch": 1.7574099987278973, + "grad_norm": 2.356717348098755, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8997677564620972, + "num_tokens": 536427329.0, + "step": 13815 + }, + { + "epoch": 1.7575372090064878, + "grad_norm": 2.546969175338745, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8965646624565125, + "num_tokens": 536465941.0, + "step": 13816 + }, + { + "epoch": 1.7576644192850783, + "grad_norm": 2.460024118423462, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9047669172286987, + "num_tokens": 536506234.0, + "step": 13817 + }, + { + "epoch": 1.7577916295636689, + "grad_norm": 2.52878999710083, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9004770517349243, + "num_tokens": 536541395.0, + "step": 13818 + }, + { + "epoch": 1.7579188398422594, + "grad_norm": 2.6318750381469727, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8993810415267944, + "num_tokens": 536577117.0, + "step": 13819 + }, + { + "epoch": 1.75804605012085, + "grad_norm": 2.6859803199768066, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.897179126739502, + "num_tokens": 536615928.0, + "step": 13820 + }, + { + "epoch": 1.7581732603994404, + "grad_norm": 2.9391732215881348, + "learning_rate": 1e-06, + "loss": 0.3426, + "mean_token_accuracy": 0.8809144496917725, + "num_tokens": 536650956.0, + "step": 13821 + }, + { + "epoch": 1.7583004706780307, + "grad_norm": 2.8626255989074707, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8900312185287476, + "num_tokens": 536684410.0, + "step": 13822 + }, + { + "epoch": 1.7584276809566213, + "grad_norm": 2.815639019012451, + "learning_rate": 1e-06, + "loss": 0.3628, + "mean_token_accuracy": 0.8705990314483643, + "num_tokens": 536723986.0, + "step": 13823 + }, + { + "epoch": 1.7585548912352118, + "grad_norm": 2.588654041290283, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8839576840400696, + "num_tokens": 536766003.0, + "step": 13824 + }, + { + "epoch": 1.7586821015138023, + "grad_norm": 2.5032784938812256, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8992017507553101, + "num_tokens": 536804203.0, + "step": 13825 + }, + { + "epoch": 1.7588093117923929, + "grad_norm": 2.6608283519744873, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8924250602722168, + "num_tokens": 536839620.0, + "step": 13826 + }, + { + "epoch": 1.7589365220709834, + "grad_norm": 2.6284260749816895, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8991527557373047, + "num_tokens": 536874454.0, + "step": 13827 + }, + { + "epoch": 1.7590637323495737, + "grad_norm": 2.676483154296875, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8796879053115845, + "num_tokens": 536917919.0, + "step": 13828 + }, + { + "epoch": 1.7591909426281642, + "grad_norm": 2.5722265243530273, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8934235572814941, + "num_tokens": 536957741.0, + "step": 13829 + }, + { + "epoch": 1.7593181529067548, + "grad_norm": 2.7728967666625977, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8961137533187866, + "num_tokens": 536993878.0, + "step": 13830 + }, + { + "epoch": 1.7594453631853453, + "grad_norm": 2.4802258014678955, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.894755482673645, + "num_tokens": 537036797.0, + "step": 13831 + }, + { + "epoch": 1.7595725734639358, + "grad_norm": 2.7856264114379883, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8827636241912842, + "num_tokens": 537075221.0, + "step": 13832 + }, + { + "epoch": 1.7596997837425263, + "grad_norm": 2.7434091567993164, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8749808073043823, + "num_tokens": 537115900.0, + "step": 13833 + }, + { + "epoch": 1.7598269940211169, + "grad_norm": 2.503772735595703, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8981648683547974, + "num_tokens": 537154709.0, + "step": 13834 + }, + { + "epoch": 1.7599542042997074, + "grad_norm": 2.688272476196289, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8943467140197754, + "num_tokens": 537190350.0, + "step": 13835 + }, + { + "epoch": 1.760081414578298, + "grad_norm": 2.5739102363586426, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.879000186920166, + "num_tokens": 537233871.0, + "step": 13836 + }, + { + "epoch": 1.7602086248568884, + "grad_norm": 2.653852701187134, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9004263877868652, + "num_tokens": 537269870.0, + "step": 13837 + }, + { + "epoch": 1.760335835135479, + "grad_norm": 2.5598366260528564, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8879514932632446, + "num_tokens": 537312515.0, + "step": 13838 + }, + { + "epoch": 1.7604630454140695, + "grad_norm": 2.732027769088745, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8850746154785156, + "num_tokens": 537348200.0, + "step": 13839 + }, + { + "epoch": 1.76059025569266, + "grad_norm": 2.641479015350342, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8903511166572571, + "num_tokens": 537382916.0, + "step": 13840 + }, + { + "epoch": 1.7607174659712506, + "grad_norm": 2.383338689804077, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9000473022460938, + "num_tokens": 537424963.0, + "step": 13841 + }, + { + "epoch": 1.760844676249841, + "grad_norm": 2.695415735244751, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8950114250183105, + "num_tokens": 537460063.0, + "step": 13842 + }, + { + "epoch": 1.7609718865284316, + "grad_norm": 2.833362340927124, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8828128576278687, + "num_tokens": 537497652.0, + "step": 13843 + }, + { + "epoch": 1.7610990968070221, + "grad_norm": 2.4050724506378174, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.896600604057312, + "num_tokens": 537538671.0, + "step": 13844 + }, + { + "epoch": 1.7612263070856127, + "grad_norm": 2.6237361431121826, + "learning_rate": 1e-06, + "loss": 0.3374, + "mean_token_accuracy": 0.8793661594390869, + "num_tokens": 537580238.0, + "step": 13845 + }, + { + "epoch": 1.761353517364203, + "grad_norm": 2.4659132957458496, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9001065492630005, + "num_tokens": 537620650.0, + "step": 13846 + }, + { + "epoch": 1.7614807276427935, + "grad_norm": 2.508065938949585, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.897786021232605, + "num_tokens": 537661481.0, + "step": 13847 + }, + { + "epoch": 1.761607937921384, + "grad_norm": 2.680104970932007, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8880905508995056, + "num_tokens": 537700972.0, + "step": 13848 + }, + { + "epoch": 1.7617351481999746, + "grad_norm": 2.640843152999878, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8914358615875244, + "num_tokens": 537739325.0, + "step": 13849 + }, + { + "epoch": 1.761862358478565, + "grad_norm": 2.6167550086975098, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8839847445487976, + "num_tokens": 537779245.0, + "step": 13850 + }, + { + "epoch": 1.7619895687571556, + "grad_norm": 2.452094078063965, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8918447494506836, + "num_tokens": 537820510.0, + "step": 13851 + }, + { + "epoch": 1.762116779035746, + "grad_norm": 2.7785868644714355, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8800880908966064, + "num_tokens": 537857616.0, + "step": 13852 + }, + { + "epoch": 1.7622439893143365, + "grad_norm": 2.7310874462127686, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8790060877799988, + "num_tokens": 537895691.0, + "step": 13853 + }, + { + "epoch": 1.762371199592927, + "grad_norm": 2.4804329872131348, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8959146738052368, + "num_tokens": 537936578.0, + "step": 13854 + }, + { + "epoch": 1.7624984098715175, + "grad_norm": 2.7211813926696777, + "learning_rate": 1e-06, + "loss": 0.329, + "mean_token_accuracy": 0.882425844669342, + "num_tokens": 537976767.0, + "step": 13855 + }, + { + "epoch": 1.762625620150108, + "grad_norm": 2.5431203842163086, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8925550580024719, + "num_tokens": 538017724.0, + "step": 13856 + }, + { + "epoch": 1.7627528304286986, + "grad_norm": 2.60957407951355, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8919494152069092, + "num_tokens": 538057012.0, + "step": 13857 + }, + { + "epoch": 1.762880040707289, + "grad_norm": 2.7435150146484375, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8725428581237793, + "num_tokens": 538095352.0, + "step": 13858 + }, + { + "epoch": 1.7630072509858796, + "grad_norm": 2.4913811683654785, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8975993394851685, + "num_tokens": 538135783.0, + "step": 13859 + }, + { + "epoch": 1.7631344612644702, + "grad_norm": 2.47398042678833, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8873742818832397, + "num_tokens": 538177472.0, + "step": 13860 + }, + { + "epoch": 1.7632616715430607, + "grad_norm": 2.6213951110839844, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8906993865966797, + "num_tokens": 538217075.0, + "step": 13861 + }, + { + "epoch": 1.7633888818216512, + "grad_norm": 2.7069411277770996, + "learning_rate": 1e-06, + "loss": 0.3569, + "mean_token_accuracy": 0.8692517280578613, + "num_tokens": 538258334.0, + "step": 13862 + }, + { + "epoch": 1.7635160921002417, + "grad_norm": 2.7248222827911377, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9017108082771301, + "num_tokens": 538293229.0, + "step": 13863 + }, + { + "epoch": 1.7636433023788323, + "grad_norm": 2.795988082885742, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.898926854133606, + "num_tokens": 538328993.0, + "step": 13864 + }, + { + "epoch": 1.7637705126574228, + "grad_norm": 2.369302988052368, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8993346095085144, + "num_tokens": 538373184.0, + "step": 13865 + }, + { + "epoch": 1.7638977229360133, + "grad_norm": 2.642510175704956, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8961048126220703, + "num_tokens": 538412264.0, + "step": 13866 + }, + { + "epoch": 1.7640249332146039, + "grad_norm": 2.613693952560425, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8823632001876831, + "num_tokens": 538451129.0, + "step": 13867 + }, + { + "epoch": 1.7641521434931944, + "grad_norm": 2.643505334854126, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8938923478126526, + "num_tokens": 538488533.0, + "step": 13868 + }, + { + "epoch": 1.764279353771785, + "grad_norm": 2.7121031284332275, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8781277537345886, + "num_tokens": 538528256.0, + "step": 13869 + }, + { + "epoch": 1.7644065640503754, + "grad_norm": 2.59318208694458, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8844031095504761, + "num_tokens": 538568640.0, + "step": 13870 + }, + { + "epoch": 1.7645337743289657, + "grad_norm": 2.5945489406585693, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8895866870880127, + "num_tokens": 538608941.0, + "step": 13871 + }, + { + "epoch": 1.7646609846075563, + "grad_norm": 2.361328363418579, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8892632126808167, + "num_tokens": 538652847.0, + "step": 13872 + }, + { + "epoch": 1.7647881948861468, + "grad_norm": 2.5356462001800537, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8940668106079102, + "num_tokens": 538691003.0, + "step": 13873 + }, + { + "epoch": 1.7649154051647373, + "grad_norm": 2.7115869522094727, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8869638442993164, + "num_tokens": 538730016.0, + "step": 13874 + }, + { + "epoch": 1.7650426154433279, + "grad_norm": 2.742429733276367, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9024761915206909, + "num_tokens": 538775056.0, + "step": 13875 + }, + { + "epoch": 1.7651698257219184, + "grad_norm": 2.729837417602539, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8809561133384705, + "num_tokens": 538812633.0, + "step": 13876 + }, + { + "epoch": 1.7652970360005087, + "grad_norm": 2.640734910964966, + "learning_rate": 1e-06, + "loss": 0.3333, + "mean_token_accuracy": 0.8809354305267334, + "num_tokens": 538853978.0, + "step": 13877 + }, + { + "epoch": 1.7654242462790992, + "grad_norm": 2.6412179470062256, + "learning_rate": 1e-06, + "loss": 0.3168, + "mean_token_accuracy": 0.8872864842414856, + "num_tokens": 538894353.0, + "step": 13878 + }, + { + "epoch": 1.7655514565576897, + "grad_norm": 2.5691895484924316, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8823869228363037, + "num_tokens": 538935412.0, + "step": 13879 + }, + { + "epoch": 1.7656786668362803, + "grad_norm": 2.771557569503784, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8872693777084351, + "num_tokens": 538969628.0, + "step": 13880 + }, + { + "epoch": 1.7658058771148708, + "grad_norm": 2.397325038909912, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8826459646224976, + "num_tokens": 539017538.0, + "step": 13881 + }, + { + "epoch": 1.7659330873934613, + "grad_norm": 2.57430100440979, + "learning_rate": 1e-06, + "loss": 0.3155, + "mean_token_accuracy": 0.8876161575317383, + "num_tokens": 539059567.0, + "step": 13882 + }, + { + "epoch": 1.7660602976720519, + "grad_norm": 2.472140073776245, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8973182439804077, + "num_tokens": 539099153.0, + "step": 13883 + }, + { + "epoch": 1.7661875079506424, + "grad_norm": 2.454197645187378, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8915186524391174, + "num_tokens": 539140908.0, + "step": 13884 + }, + { + "epoch": 1.766314718229233, + "grad_norm": 2.4678313732147217, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8961433172225952, + "num_tokens": 539181493.0, + "step": 13885 + }, + { + "epoch": 1.7664419285078234, + "grad_norm": 2.677030324935913, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8910704851150513, + "num_tokens": 539218018.0, + "step": 13886 + }, + { + "epoch": 1.766569138786414, + "grad_norm": 2.517625570297241, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8925038576126099, + "num_tokens": 539257530.0, + "step": 13887 + }, + { + "epoch": 1.7666963490650045, + "grad_norm": 2.7100067138671875, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9022303223609924, + "num_tokens": 539288995.0, + "step": 13888 + }, + { + "epoch": 1.766823559343595, + "grad_norm": 2.6388397216796875, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8937076926231384, + "num_tokens": 539325975.0, + "step": 13889 + }, + { + "epoch": 1.7669507696221856, + "grad_norm": 2.600508689880371, + "learning_rate": 1e-06, + "loss": 0.3652, + "mean_token_accuracy": 0.8705164194107056, + "num_tokens": 539369509.0, + "step": 13890 + }, + { + "epoch": 1.767077979900776, + "grad_norm": 2.598405122756958, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8887572288513184, + "num_tokens": 539409191.0, + "step": 13891 + }, + { + "epoch": 1.7672051901793666, + "grad_norm": 2.693303346633911, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.901497483253479, + "num_tokens": 539445092.0, + "step": 13892 + }, + { + "epoch": 1.7673324004579571, + "grad_norm": 2.5482735633850098, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8864164352416992, + "num_tokens": 539487914.0, + "step": 13893 + }, + { + "epoch": 1.7674596107365477, + "grad_norm": 2.832566022872925, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.886233925819397, + "num_tokens": 539524753.0, + "step": 13894 + }, + { + "epoch": 1.767586821015138, + "grad_norm": 2.657071352005005, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9026577472686768, + "num_tokens": 539559556.0, + "step": 13895 + }, + { + "epoch": 1.7677140312937285, + "grad_norm": 2.6781563758850098, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8855957388877869, + "num_tokens": 539596838.0, + "step": 13896 + }, + { + "epoch": 1.767841241572319, + "grad_norm": 2.4934866428375244, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8966107964515686, + "num_tokens": 539638125.0, + "step": 13897 + }, + { + "epoch": 1.7679684518509096, + "grad_norm": 2.6732606887817383, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9082980751991272, + "num_tokens": 539675273.0, + "step": 13898 + }, + { + "epoch": 1.7680956621295, + "grad_norm": 2.578883647918701, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8932135701179504, + "num_tokens": 539718179.0, + "step": 13899 + }, + { + "epoch": 1.7682228724080906, + "grad_norm": 2.5735931396484375, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.881751537322998, + "num_tokens": 539760428.0, + "step": 13900 + }, + { + "epoch": 1.768350082686681, + "grad_norm": 2.69986629486084, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8864293098449707, + "num_tokens": 539798892.0, + "step": 13901 + }, + { + "epoch": 1.7684772929652715, + "grad_norm": 2.57875394821167, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8885276913642883, + "num_tokens": 539838194.0, + "step": 13902 + }, + { + "epoch": 1.768604503243862, + "grad_norm": 2.3947594165802, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9053651094436646, + "num_tokens": 539877230.0, + "step": 13903 + }, + { + "epoch": 1.7687317135224525, + "grad_norm": 2.3926098346710205, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9043091535568237, + "num_tokens": 539916402.0, + "step": 13904 + }, + { + "epoch": 1.768858923801043, + "grad_norm": 2.2858471870422363, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9025750160217285, + "num_tokens": 539962129.0, + "step": 13905 + }, + { + "epoch": 1.7689861340796336, + "grad_norm": 2.3927254676818848, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9004342555999756, + "num_tokens": 540002158.0, + "step": 13906 + }, + { + "epoch": 1.769113344358224, + "grad_norm": 2.6030638217926025, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8856855630874634, + "num_tokens": 540042366.0, + "step": 13907 + }, + { + "epoch": 1.7692405546368146, + "grad_norm": 2.600076198577881, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.892264187335968, + "num_tokens": 540081309.0, + "step": 13908 + }, + { + "epoch": 1.7693677649154052, + "grad_norm": 2.651561737060547, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8935948610305786, + "num_tokens": 540119882.0, + "step": 13909 + }, + { + "epoch": 1.7694949751939957, + "grad_norm": 2.5091445446014404, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.891049861907959, + "num_tokens": 540160898.0, + "step": 13910 + }, + { + "epoch": 1.7696221854725862, + "grad_norm": 2.7836062908172607, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.8974785804748535, + "num_tokens": 540193414.0, + "step": 13911 + }, + { + "epoch": 1.7697493957511767, + "grad_norm": 2.5860812664031982, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8876422643661499, + "num_tokens": 540233679.0, + "step": 13912 + }, + { + "epoch": 1.7698766060297673, + "grad_norm": 2.6201112270355225, + "learning_rate": 1e-06, + "loss": 0.3321, + "mean_token_accuracy": 0.8806031346321106, + "num_tokens": 540278740.0, + "step": 13913 + }, + { + "epoch": 1.7700038163083578, + "grad_norm": 2.561884880065918, + "learning_rate": 1e-06, + "loss": 0.343, + "mean_token_accuracy": 0.8782603740692139, + "num_tokens": 540323853.0, + "step": 13914 + }, + { + "epoch": 1.7701310265869483, + "grad_norm": 2.5930428504943848, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8943109512329102, + "num_tokens": 540360327.0, + "step": 13915 + }, + { + "epoch": 1.7702582368655388, + "grad_norm": 2.7266595363616943, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.885380208492279, + "num_tokens": 540398665.0, + "step": 13916 + }, + { + "epoch": 1.7703854471441294, + "grad_norm": 2.5863914489746094, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8926267623901367, + "num_tokens": 540434938.0, + "step": 13917 + }, + { + "epoch": 1.77051265742272, + "grad_norm": 2.8757972717285156, + "learning_rate": 1e-06, + "loss": 0.3568, + "mean_token_accuracy": 0.8724637031555176, + "num_tokens": 540472308.0, + "step": 13918 + }, + { + "epoch": 1.7706398677013102, + "grad_norm": 2.6796743869781494, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.879584789276123, + "num_tokens": 540510193.0, + "step": 13919 + }, + { + "epoch": 1.7707670779799007, + "grad_norm": 2.736049175262451, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.881225049495697, + "num_tokens": 540549694.0, + "step": 13920 + }, + { + "epoch": 1.7708942882584913, + "grad_norm": 2.7264387607574463, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8859730958938599, + "num_tokens": 540592100.0, + "step": 13921 + }, + { + "epoch": 1.7710214985370818, + "grad_norm": 2.4941139221191406, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.892388105392456, + "num_tokens": 540634095.0, + "step": 13922 + }, + { + "epoch": 1.7711487088156723, + "grad_norm": 2.7481164932250977, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8913270831108093, + "num_tokens": 540667966.0, + "step": 13923 + }, + { + "epoch": 1.7712759190942629, + "grad_norm": 2.921438694000244, + "learning_rate": 1e-06, + "loss": 0.3448, + "mean_token_accuracy": 0.877713680267334, + "num_tokens": 540701060.0, + "step": 13924 + }, + { + "epoch": 1.7714031293728534, + "grad_norm": 2.7089006900787354, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8882575631141663, + "num_tokens": 540737711.0, + "step": 13925 + }, + { + "epoch": 1.7715303396514437, + "grad_norm": 2.7279176712036133, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8854781985282898, + "num_tokens": 540772392.0, + "step": 13926 + }, + { + "epoch": 1.7716575499300342, + "grad_norm": 2.5700948238372803, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8870797157287598, + "num_tokens": 540812839.0, + "step": 13927 + }, + { + "epoch": 1.7717847602086247, + "grad_norm": 2.809870958328247, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8906949758529663, + "num_tokens": 540847636.0, + "step": 13928 + }, + { + "epoch": 1.7719119704872153, + "grad_norm": 2.5527379512786865, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8980125784873962, + "num_tokens": 540888883.0, + "step": 13929 + }, + { + "epoch": 1.7720391807658058, + "grad_norm": 2.5961763858795166, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8878275156021118, + "num_tokens": 540932439.0, + "step": 13930 + }, + { + "epoch": 1.7721663910443963, + "grad_norm": 2.5943360328674316, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.888789713382721, + "num_tokens": 540971840.0, + "step": 13931 + }, + { + "epoch": 1.7722936013229869, + "grad_norm": 2.5429317951202393, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8842355608940125, + "num_tokens": 541012823.0, + "step": 13932 + }, + { + "epoch": 1.7724208116015774, + "grad_norm": 2.5058088302612305, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8893920183181763, + "num_tokens": 541053284.0, + "step": 13933 + }, + { + "epoch": 1.772548021880168, + "grad_norm": 2.3956685066223145, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9015862941741943, + "num_tokens": 541094348.0, + "step": 13934 + }, + { + "epoch": 1.7726752321587584, + "grad_norm": 2.743128538131714, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8825986981391907, + "num_tokens": 541132289.0, + "step": 13935 + }, + { + "epoch": 1.772802442437349, + "grad_norm": 2.7300524711608887, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8836276531219482, + "num_tokens": 541172112.0, + "step": 13936 + }, + { + "epoch": 1.7729296527159395, + "grad_norm": 2.6920950412750244, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8893135786056519, + "num_tokens": 541211200.0, + "step": 13937 + }, + { + "epoch": 1.77305686299453, + "grad_norm": 2.491403102874756, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9015722274780273, + "num_tokens": 541250797.0, + "step": 13938 + }, + { + "epoch": 1.7731840732731206, + "grad_norm": 2.7866501808166504, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8869236707687378, + "num_tokens": 541287664.0, + "step": 13939 + }, + { + "epoch": 1.773311283551711, + "grad_norm": 2.666828155517578, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9050096273422241, + "num_tokens": 541320952.0, + "step": 13940 + }, + { + "epoch": 1.7734384938303016, + "grad_norm": 2.698429584503174, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8913670778274536, + "num_tokens": 541353847.0, + "step": 13941 + }, + { + "epoch": 1.7735657041088921, + "grad_norm": 2.5842533111572266, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8902109861373901, + "num_tokens": 541391706.0, + "step": 13942 + }, + { + "epoch": 1.7736929143874827, + "grad_norm": 2.5317633152008057, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.8998164534568787, + "num_tokens": 541429289.0, + "step": 13943 + }, + { + "epoch": 1.773820124666073, + "grad_norm": 2.4727203845977783, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8859671354293823, + "num_tokens": 541468599.0, + "step": 13944 + }, + { + "epoch": 1.7739473349446635, + "grad_norm": 2.8124759197235107, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8865334987640381, + "num_tokens": 541507754.0, + "step": 13945 + }, + { + "epoch": 1.774074545223254, + "grad_norm": 3.0050342082977295, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8820028901100159, + "num_tokens": 541540872.0, + "step": 13946 + }, + { + "epoch": 1.7742017555018446, + "grad_norm": 2.6443769931793213, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8941699266433716, + "num_tokens": 541577336.0, + "step": 13947 + }, + { + "epoch": 1.774328965780435, + "grad_norm": 2.6730778217315674, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8915973901748657, + "num_tokens": 541610175.0, + "step": 13948 + }, + { + "epoch": 1.7744561760590256, + "grad_norm": 2.694003105163574, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8954271674156189, + "num_tokens": 541644275.0, + "step": 13949 + }, + { + "epoch": 1.774583386337616, + "grad_norm": 2.5691628456115723, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8931187391281128, + "num_tokens": 541683151.0, + "step": 13950 + }, + { + "epoch": 1.7747105966162064, + "grad_norm": 2.5806570053100586, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9011970162391663, + "num_tokens": 541719052.0, + "step": 13951 + }, + { + "epoch": 1.774837806894797, + "grad_norm": 2.487745761871338, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9000469446182251, + "num_tokens": 541759237.0, + "step": 13952 + }, + { + "epoch": 1.7749650171733875, + "grad_norm": 2.7860429286956787, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.887005090713501, + "num_tokens": 541796346.0, + "step": 13953 + }, + { + "epoch": 1.775092227451978, + "grad_norm": 2.6453983783721924, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8912440538406372, + "num_tokens": 541836852.0, + "step": 13954 + }, + { + "epoch": 1.7752194377305686, + "grad_norm": 2.837312936782837, + "learning_rate": 1e-06, + "loss": 0.3414, + "mean_token_accuracy": 0.8779237270355225, + "num_tokens": 541871920.0, + "step": 13955 + }, + { + "epoch": 1.775346648009159, + "grad_norm": 2.511361598968506, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.9012734889984131, + "num_tokens": 541914126.0, + "step": 13956 + }, + { + "epoch": 1.7754738582877496, + "grad_norm": 2.562561511993408, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8897874355316162, + "num_tokens": 541954120.0, + "step": 13957 + }, + { + "epoch": 1.7756010685663401, + "grad_norm": 2.568009376525879, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8964700698852539, + "num_tokens": 541993912.0, + "step": 13958 + }, + { + "epoch": 1.7757282788449307, + "grad_norm": 2.6883645057678223, + "learning_rate": 1e-06, + "loss": 0.3506, + "mean_token_accuracy": 0.8818609714508057, + "num_tokens": 542032993.0, + "step": 13959 + }, + { + "epoch": 1.7758554891235212, + "grad_norm": 2.7293810844421387, + "learning_rate": 1e-06, + "loss": 0.3452, + "mean_token_accuracy": 0.8780646920204163, + "num_tokens": 542075332.0, + "step": 13960 + }, + { + "epoch": 1.7759826994021117, + "grad_norm": 2.6827502250671387, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9019049406051636, + "num_tokens": 542110001.0, + "step": 13961 + }, + { + "epoch": 1.7761099096807023, + "grad_norm": 2.8997318744659424, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8931464552879333, + "num_tokens": 542141883.0, + "step": 13962 + }, + { + "epoch": 1.7762371199592928, + "grad_norm": 2.5645172595977783, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8942899703979492, + "num_tokens": 542182730.0, + "step": 13963 + }, + { + "epoch": 1.7763643302378833, + "grad_norm": 2.5167436599731445, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8840049505233765, + "num_tokens": 542227962.0, + "step": 13964 + }, + { + "epoch": 1.7764915405164738, + "grad_norm": 2.6422131061553955, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8965193629264832, + "num_tokens": 542261739.0, + "step": 13965 + }, + { + "epoch": 1.7766187507950644, + "grad_norm": 2.8812520503997803, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8890553712844849, + "num_tokens": 542298660.0, + "step": 13966 + }, + { + "epoch": 1.776745961073655, + "grad_norm": 2.7868218421936035, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8927852511405945, + "num_tokens": 542334221.0, + "step": 13967 + }, + { + "epoch": 1.7768731713522452, + "grad_norm": 2.7151691913604736, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8896547555923462, + "num_tokens": 542371048.0, + "step": 13968 + }, + { + "epoch": 1.7770003816308357, + "grad_norm": 2.679044485092163, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8963085412979126, + "num_tokens": 542407666.0, + "step": 13969 + }, + { + "epoch": 1.7771275919094263, + "grad_norm": 2.324509620666504, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8923970460891724, + "num_tokens": 542453259.0, + "step": 13970 + }, + { + "epoch": 1.7772548021880168, + "grad_norm": 2.61428165435791, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8860644102096558, + "num_tokens": 542491373.0, + "step": 13971 + }, + { + "epoch": 1.7773820124666073, + "grad_norm": 2.8758602142333984, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.88630211353302, + "num_tokens": 542532689.0, + "step": 13972 + }, + { + "epoch": 1.7775092227451978, + "grad_norm": 2.578709602355957, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8870132565498352, + "num_tokens": 542575646.0, + "step": 13973 + }, + { + "epoch": 1.7776364330237884, + "grad_norm": 2.8727364540100098, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.886891782283783, + "num_tokens": 542607114.0, + "step": 13974 + }, + { + "epoch": 1.7777636433023787, + "grad_norm": 2.4499170780181885, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8938054442405701, + "num_tokens": 542651943.0, + "step": 13975 + }, + { + "epoch": 1.7778908535809692, + "grad_norm": 2.4102559089660645, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9050726890563965, + "num_tokens": 542691053.0, + "step": 13976 + }, + { + "epoch": 1.7780180638595597, + "grad_norm": 2.8681187629699707, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8875603079795837, + "num_tokens": 542727592.0, + "step": 13977 + }, + { + "epoch": 1.7781452741381503, + "grad_norm": 2.600938320159912, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.898600697517395, + "num_tokens": 542766712.0, + "step": 13978 + }, + { + "epoch": 1.7782724844167408, + "grad_norm": 2.582092523574829, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8875428438186646, + "num_tokens": 542811389.0, + "step": 13979 + }, + { + "epoch": 1.7783996946953313, + "grad_norm": 2.719203472137451, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.887083888053894, + "num_tokens": 542852159.0, + "step": 13980 + }, + { + "epoch": 1.7785269049739219, + "grad_norm": 2.6870100498199463, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8839337825775146, + "num_tokens": 542893654.0, + "step": 13981 + }, + { + "epoch": 1.7786541152525124, + "grad_norm": 2.6211049556732178, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8951692581176758, + "num_tokens": 542931065.0, + "step": 13982 + }, + { + "epoch": 1.778781325531103, + "grad_norm": 2.6873645782470703, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8849113583564758, + "num_tokens": 542965570.0, + "step": 13983 + }, + { + "epoch": 1.7789085358096934, + "grad_norm": 2.7013602256774902, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8844813704490662, + "num_tokens": 543005163.0, + "step": 13984 + }, + { + "epoch": 1.779035746088284, + "grad_norm": 2.7789196968078613, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8963230848312378, + "num_tokens": 543039117.0, + "step": 13985 + }, + { + "epoch": 1.7791629563668745, + "grad_norm": 2.4909024238586426, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8923760056495667, + "num_tokens": 543079877.0, + "step": 13986 + }, + { + "epoch": 1.779290166645465, + "grad_norm": 2.54025936126709, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8848980665206909, + "num_tokens": 543122384.0, + "step": 13987 + }, + { + "epoch": 1.7794173769240555, + "grad_norm": 2.6134228706359863, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8987189531326294, + "num_tokens": 543159216.0, + "step": 13988 + }, + { + "epoch": 1.779544587202646, + "grad_norm": 2.83247971534729, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8857412338256836, + "num_tokens": 543192666.0, + "step": 13989 + }, + { + "epoch": 1.7796717974812366, + "grad_norm": 2.3774473667144775, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8903907537460327, + "num_tokens": 543239747.0, + "step": 13990 + }, + { + "epoch": 1.7797990077598271, + "grad_norm": 2.637479066848755, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8956577777862549, + "num_tokens": 543277816.0, + "step": 13991 + }, + { + "epoch": 1.7799262180384177, + "grad_norm": 2.8742668628692627, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8846383690834045, + "num_tokens": 543313218.0, + "step": 13992 + }, + { + "epoch": 1.780053428317008, + "grad_norm": 2.844212770462036, + "learning_rate": 1e-06, + "loss": 0.3324, + "mean_token_accuracy": 0.8819395303726196, + "num_tokens": 543348400.0, + "step": 13993 + }, + { + "epoch": 1.7801806385955985, + "grad_norm": 2.642824172973633, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8968714475631714, + "num_tokens": 543383038.0, + "step": 13994 + }, + { + "epoch": 1.780307848874189, + "grad_norm": 2.591296672821045, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8976091742515564, + "num_tokens": 543419871.0, + "step": 13995 + }, + { + "epoch": 1.7804350591527796, + "grad_norm": 2.7408597469329834, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8875128030776978, + "num_tokens": 543456284.0, + "step": 13996 + }, + { + "epoch": 1.78056226943137, + "grad_norm": 2.506009578704834, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8848524689674377, + "num_tokens": 543498310.0, + "step": 13997 + }, + { + "epoch": 1.7806894797099606, + "grad_norm": 2.7324845790863037, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8889440298080444, + "num_tokens": 543531981.0, + "step": 13998 + }, + { + "epoch": 1.780816689988551, + "grad_norm": 2.400634288787842, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9040192365646362, + "num_tokens": 543569590.0, + "step": 13999 + }, + { + "epoch": 1.7809439002671414, + "grad_norm": 2.8216562271118164, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8868570327758789, + "num_tokens": 543606389.0, + "step": 14000 + }, + { + "epoch": 1.781071110545732, + "grad_norm": 2.4490914344787598, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8894520998001099, + "num_tokens": 543648946.0, + "step": 14001 + }, + { + "epoch": 1.7811983208243225, + "grad_norm": 2.4834063053131104, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.881013035774231, + "num_tokens": 543690659.0, + "step": 14002 + }, + { + "epoch": 1.781325531102913, + "grad_norm": 2.9949591159820557, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8798007369041443, + "num_tokens": 543723327.0, + "step": 14003 + }, + { + "epoch": 1.7814527413815036, + "grad_norm": 2.5982112884521484, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9001756906509399, + "num_tokens": 543755820.0, + "step": 14004 + }, + { + "epoch": 1.781579951660094, + "grad_norm": 2.684844732284546, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8954481482505798, + "num_tokens": 543789974.0, + "step": 14005 + }, + { + "epoch": 1.7817071619386846, + "grad_norm": 2.64982533454895, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8876663446426392, + "num_tokens": 543827434.0, + "step": 14006 + }, + { + "epoch": 1.7818343722172751, + "grad_norm": 2.5769150257110596, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8956881165504456, + "num_tokens": 543866338.0, + "step": 14007 + }, + { + "epoch": 1.7819615824958657, + "grad_norm": 2.7208259105682373, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8934019804000854, + "num_tokens": 543901592.0, + "step": 14008 + }, + { + "epoch": 1.7820887927744562, + "grad_norm": 2.8979737758636475, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.892715573310852, + "num_tokens": 543932859.0, + "step": 14009 + }, + { + "epoch": 1.7822160030530467, + "grad_norm": 2.5978524684906006, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.895492672920227, + "num_tokens": 543969010.0, + "step": 14010 + }, + { + "epoch": 1.7823432133316373, + "grad_norm": 2.5833582878112793, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8895471096038818, + "num_tokens": 544005951.0, + "step": 14011 + }, + { + "epoch": 1.7824704236102278, + "grad_norm": 2.553006410598755, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8900730013847351, + "num_tokens": 544046067.0, + "step": 14012 + }, + { + "epoch": 1.7825976338888183, + "grad_norm": 2.914076805114746, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.8864390850067139, + "num_tokens": 544078453.0, + "step": 14013 + }, + { + "epoch": 1.7827248441674088, + "grad_norm": 2.8120124340057373, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8971707820892334, + "num_tokens": 544114263.0, + "step": 14014 + }, + { + "epoch": 1.7828520544459994, + "grad_norm": 2.7463767528533936, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8873299360275269, + "num_tokens": 544149943.0, + "step": 14015 + }, + { + "epoch": 1.78297926472459, + "grad_norm": 2.4975996017456055, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8931032419204712, + "num_tokens": 544190089.0, + "step": 14016 + }, + { + "epoch": 1.7831064750031802, + "grad_norm": 2.7334227561950684, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8902510404586792, + "num_tokens": 544225749.0, + "step": 14017 + }, + { + "epoch": 1.7832336852817707, + "grad_norm": 2.5153307914733887, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8846544623374939, + "num_tokens": 544268231.0, + "step": 14018 + }, + { + "epoch": 1.7833608955603613, + "grad_norm": 2.848510980606079, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8895025849342346, + "num_tokens": 544302832.0, + "step": 14019 + }, + { + "epoch": 1.7834881058389518, + "grad_norm": 2.7934939861297607, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8830031156539917, + "num_tokens": 544339411.0, + "step": 14020 + }, + { + "epoch": 1.7836153161175423, + "grad_norm": 2.5460362434387207, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.881962776184082, + "num_tokens": 544385657.0, + "step": 14021 + }, + { + "epoch": 1.7837425263961328, + "grad_norm": 2.330169677734375, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9058642387390137, + "num_tokens": 544427213.0, + "step": 14022 + }, + { + "epoch": 1.7838697366747234, + "grad_norm": 2.817978858947754, + "learning_rate": 1e-06, + "loss": 0.3508, + "mean_token_accuracy": 0.8767253160476685, + "num_tokens": 544467834.0, + "step": 14023 + }, + { + "epoch": 1.7839969469533137, + "grad_norm": 2.6878342628479004, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8783693909645081, + "num_tokens": 544508911.0, + "step": 14024 + }, + { + "epoch": 1.7841241572319042, + "grad_norm": 2.5747954845428467, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9057071208953857, + "num_tokens": 544547634.0, + "step": 14025 + }, + { + "epoch": 1.7842513675104947, + "grad_norm": 2.708559274673462, + "learning_rate": 1e-06, + "loss": 0.371, + "mean_token_accuracy": 0.8663003444671631, + "num_tokens": 544591492.0, + "step": 14026 + }, + { + "epoch": 1.7843785777890853, + "grad_norm": 2.5601038932800293, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8896989822387695, + "num_tokens": 544631433.0, + "step": 14027 + }, + { + "epoch": 1.7845057880676758, + "grad_norm": 2.9292361736297607, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.888786792755127, + "num_tokens": 544662169.0, + "step": 14028 + }, + { + "epoch": 1.7846329983462663, + "grad_norm": 2.481708288192749, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8821582794189453, + "num_tokens": 544706645.0, + "step": 14029 + }, + { + "epoch": 1.7847602086248568, + "grad_norm": 2.749586820602417, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8794000744819641, + "num_tokens": 544745244.0, + "step": 14030 + }, + { + "epoch": 1.7848874189034474, + "grad_norm": 2.6919748783111572, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.885404109954834, + "num_tokens": 544779174.0, + "step": 14031 + }, + { + "epoch": 1.785014629182038, + "grad_norm": 2.7499074935913086, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9007992148399353, + "num_tokens": 544810832.0, + "step": 14032 + }, + { + "epoch": 1.7851418394606284, + "grad_norm": 2.710179567337036, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8872599601745605, + "num_tokens": 544848482.0, + "step": 14033 + }, + { + "epoch": 1.785269049739219, + "grad_norm": 2.7384040355682373, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8951029777526855, + "num_tokens": 544882684.0, + "step": 14034 + }, + { + "epoch": 1.7853962600178095, + "grad_norm": 2.4053521156311035, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8896114826202393, + "num_tokens": 544925830.0, + "step": 14035 + }, + { + "epoch": 1.7855234702964, + "grad_norm": 2.5204684734344482, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.8840094804763794, + "num_tokens": 544968317.0, + "step": 14036 + }, + { + "epoch": 1.7856506805749905, + "grad_norm": 2.5349481105804443, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8964973092079163, + "num_tokens": 545007096.0, + "step": 14037 + }, + { + "epoch": 1.785777890853581, + "grad_norm": 2.443673849105835, + "learning_rate": 1e-06, + "loss": 0.3127, + "mean_token_accuracy": 0.8896120190620422, + "num_tokens": 545048919.0, + "step": 14038 + }, + { + "epoch": 1.7859051011321716, + "grad_norm": 2.468341588973999, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8960405588150024, + "num_tokens": 545087700.0, + "step": 14039 + }, + { + "epoch": 1.7860323114107621, + "grad_norm": 2.5357577800750732, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8977826833724976, + "num_tokens": 545126528.0, + "step": 14040 + }, + { + "epoch": 1.7861595216893527, + "grad_norm": 2.6131205558776855, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8870925903320312, + "num_tokens": 545167548.0, + "step": 14041 + }, + { + "epoch": 1.786286731967943, + "grad_norm": 2.8868422508239746, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.882076621055603, + "num_tokens": 545201111.0, + "step": 14042 + }, + { + "epoch": 1.7864139422465335, + "grad_norm": 2.4463002681732178, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8900393843650818, + "num_tokens": 545245280.0, + "step": 14043 + }, + { + "epoch": 1.786541152525124, + "grad_norm": 2.53314208984375, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8972415924072266, + "num_tokens": 545285929.0, + "step": 14044 + }, + { + "epoch": 1.7866683628037145, + "grad_norm": 2.423476219177246, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8958468437194824, + "num_tokens": 545327609.0, + "step": 14045 + }, + { + "epoch": 1.786795573082305, + "grad_norm": 2.5228147506713867, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9022283554077148, + "num_tokens": 545365988.0, + "step": 14046 + }, + { + "epoch": 1.7869227833608956, + "grad_norm": 2.9256906509399414, + "learning_rate": 1e-06, + "loss": 0.3773, + "mean_token_accuracy": 0.8655803203582764, + "num_tokens": 545404707.0, + "step": 14047 + }, + { + "epoch": 1.787049993639486, + "grad_norm": 2.4498279094696045, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9086428880691528, + "num_tokens": 545446551.0, + "step": 14048 + }, + { + "epoch": 1.7871772039180764, + "grad_norm": 2.9025402069091797, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8854325413703918, + "num_tokens": 545480641.0, + "step": 14049 + }, + { + "epoch": 1.787304414196667, + "grad_norm": 2.467355728149414, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8931264877319336, + "num_tokens": 545521590.0, + "step": 14050 + }, + { + "epoch": 1.7874316244752575, + "grad_norm": 2.4926886558532715, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8948422074317932, + "num_tokens": 545561221.0, + "step": 14051 + }, + { + "epoch": 1.787558834753848, + "grad_norm": 2.5224449634552, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8937856554985046, + "num_tokens": 545600573.0, + "step": 14052 + }, + { + "epoch": 1.7876860450324386, + "grad_norm": 2.6196372509002686, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8818362355232239, + "num_tokens": 545639590.0, + "step": 14053 + }, + { + "epoch": 1.787813255311029, + "grad_norm": 2.555532693862915, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.898811936378479, + "num_tokens": 545677103.0, + "step": 14054 + }, + { + "epoch": 1.7879404655896196, + "grad_norm": 2.433884859085083, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8933076858520508, + "num_tokens": 545720350.0, + "step": 14055 + }, + { + "epoch": 1.7880676758682101, + "grad_norm": 2.584855318069458, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8990881443023682, + "num_tokens": 545753138.0, + "step": 14056 + }, + { + "epoch": 1.7881948861468007, + "grad_norm": 2.65570330619812, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8912197351455688, + "num_tokens": 545787256.0, + "step": 14057 + }, + { + "epoch": 1.7883220964253912, + "grad_norm": 2.8110718727111816, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8850181102752686, + "num_tokens": 545826196.0, + "step": 14058 + }, + { + "epoch": 1.7884493067039817, + "grad_norm": 2.6365246772766113, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8961225748062134, + "num_tokens": 545864371.0, + "step": 14059 + }, + { + "epoch": 1.7885765169825723, + "grad_norm": 2.4905805587768555, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8848236799240112, + "num_tokens": 545908078.0, + "step": 14060 + }, + { + "epoch": 1.7887037272611628, + "grad_norm": 2.9668538570404053, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8855316638946533, + "num_tokens": 545941245.0, + "step": 14061 + }, + { + "epoch": 1.7888309375397533, + "grad_norm": 2.360060930252075, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.8992840647697449, + "num_tokens": 545984666.0, + "step": 14062 + }, + { + "epoch": 1.7889581478183438, + "grad_norm": 2.6083643436431885, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8923897743225098, + "num_tokens": 546024073.0, + "step": 14063 + }, + { + "epoch": 1.7890853580969344, + "grad_norm": 2.5339574813842773, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8925948143005371, + "num_tokens": 546061930.0, + "step": 14064 + }, + { + "epoch": 1.789212568375525, + "grad_norm": 2.478553056716919, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8913666605949402, + "num_tokens": 546106602.0, + "step": 14065 + }, + { + "epoch": 1.7893397786541152, + "grad_norm": 2.760094404220581, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.876427412033081, + "num_tokens": 546148049.0, + "step": 14066 + }, + { + "epoch": 1.7894669889327057, + "grad_norm": 2.6307899951934814, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8978091478347778, + "num_tokens": 546182351.0, + "step": 14067 + }, + { + "epoch": 1.7895941992112963, + "grad_norm": 2.5492448806762695, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8934763073921204, + "num_tokens": 546218905.0, + "step": 14068 + }, + { + "epoch": 1.7897214094898868, + "grad_norm": 3.09255051612854, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8861649036407471, + "num_tokens": 546248085.0, + "step": 14069 + }, + { + "epoch": 1.7898486197684773, + "grad_norm": 2.5518767833709717, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.89313805103302, + "num_tokens": 546286556.0, + "step": 14070 + }, + { + "epoch": 1.7899758300470678, + "grad_norm": 2.7372958660125732, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8825021982192993, + "num_tokens": 546323703.0, + "step": 14071 + }, + { + "epoch": 1.7901030403256584, + "grad_norm": 2.7264716625213623, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8942369222640991, + "num_tokens": 546358542.0, + "step": 14072 + }, + { + "epoch": 1.7902302506042487, + "grad_norm": 2.648747682571411, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.8967965841293335, + "num_tokens": 546390907.0, + "step": 14073 + }, + { + "epoch": 1.7903574608828392, + "grad_norm": 3.1240737438201904, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8758710622787476, + "num_tokens": 546421450.0, + "step": 14074 + }, + { + "epoch": 1.7904846711614297, + "grad_norm": 2.8397765159606934, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8890831470489502, + "num_tokens": 546454019.0, + "step": 14075 + }, + { + "epoch": 1.7906118814400203, + "grad_norm": 2.691195249557495, + "learning_rate": 1e-06, + "loss": 0.3623, + "mean_token_accuracy": 0.874363899230957, + "num_tokens": 546496416.0, + "step": 14076 + }, + { + "epoch": 1.7907390917186108, + "grad_norm": 2.6448123455047607, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8918973207473755, + "num_tokens": 546535101.0, + "step": 14077 + }, + { + "epoch": 1.7908663019972013, + "grad_norm": 2.804325580596924, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.884870707988739, + "num_tokens": 546570366.0, + "step": 14078 + }, + { + "epoch": 1.7909935122757918, + "grad_norm": 2.802175760269165, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8918031454086304, + "num_tokens": 546607436.0, + "step": 14079 + }, + { + "epoch": 1.7911207225543824, + "grad_norm": 2.7439417839050293, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8863283395767212, + "num_tokens": 546640501.0, + "step": 14080 + }, + { + "epoch": 1.791247932832973, + "grad_norm": 2.605470895767212, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8875124454498291, + "num_tokens": 546680060.0, + "step": 14081 + }, + { + "epoch": 1.7913751431115634, + "grad_norm": 2.638275623321533, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.8992544412612915, + "num_tokens": 546712279.0, + "step": 14082 + }, + { + "epoch": 1.791502353390154, + "grad_norm": 2.898912191390991, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8889443278312683, + "num_tokens": 546742880.0, + "step": 14083 + }, + { + "epoch": 1.7916295636687445, + "grad_norm": 2.5239720344543457, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.885572075843811, + "num_tokens": 546785376.0, + "step": 14084 + }, + { + "epoch": 1.791756773947335, + "grad_norm": 2.5787572860717773, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8950251936912537, + "num_tokens": 546820726.0, + "step": 14085 + }, + { + "epoch": 1.7918839842259255, + "grad_norm": 2.511434316635132, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8917800784111023, + "num_tokens": 546859451.0, + "step": 14086 + }, + { + "epoch": 1.792011194504516, + "grad_norm": 2.550888776779175, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8899151682853699, + "num_tokens": 546898663.0, + "step": 14087 + }, + { + "epoch": 1.7921384047831066, + "grad_norm": 2.567107915878296, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8933653831481934, + "num_tokens": 546939392.0, + "step": 14088 + }, + { + "epoch": 1.7922656150616971, + "grad_norm": 2.6144838333129883, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8884444832801819, + "num_tokens": 546977225.0, + "step": 14089 + }, + { + "epoch": 1.7923928253402877, + "grad_norm": 2.582590103149414, + "learning_rate": 1e-06, + "loss": 0.3233, + "mean_token_accuracy": 0.8821216225624084, + "num_tokens": 547017585.0, + "step": 14090 + }, + { + "epoch": 1.792520035618878, + "grad_norm": 2.4331393241882324, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9003704786300659, + "num_tokens": 547062187.0, + "step": 14091 + }, + { + "epoch": 1.7926472458974685, + "grad_norm": 2.717224597930908, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8978298306465149, + "num_tokens": 547097568.0, + "step": 14092 + }, + { + "epoch": 1.792774456176059, + "grad_norm": 2.6801223754882812, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8879003524780273, + "num_tokens": 547133777.0, + "step": 14093 + }, + { + "epoch": 1.7929016664546495, + "grad_norm": 2.690298080444336, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8854635953903198, + "num_tokens": 547172058.0, + "step": 14094 + }, + { + "epoch": 1.79302887673324, + "grad_norm": 2.623041868209839, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8952294588088989, + "num_tokens": 547213795.0, + "step": 14095 + }, + { + "epoch": 1.7931560870118306, + "grad_norm": 2.8046977519989014, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8954929709434509, + "num_tokens": 547245228.0, + "step": 14096 + }, + { + "epoch": 1.793283297290421, + "grad_norm": 2.699855089187622, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8961912393569946, + "num_tokens": 547280522.0, + "step": 14097 + }, + { + "epoch": 1.7934105075690114, + "grad_norm": 2.5883116722106934, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8798791170120239, + "num_tokens": 547320244.0, + "step": 14098 + }, + { + "epoch": 1.793537717847602, + "grad_norm": 2.7362117767333984, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8921805620193481, + "num_tokens": 547356540.0, + "step": 14099 + }, + { + "epoch": 1.7936649281261925, + "grad_norm": 2.538022041320801, + "learning_rate": 1e-06, + "loss": 0.3601, + "mean_token_accuracy": 0.8722473978996277, + "num_tokens": 547404026.0, + "step": 14100 + }, + { + "epoch": 1.793792138404783, + "grad_norm": 2.542520761489868, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8864383101463318, + "num_tokens": 547448875.0, + "step": 14101 + }, + { + "epoch": 1.7939193486833735, + "grad_norm": 2.605645179748535, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8932198882102966, + "num_tokens": 547487062.0, + "step": 14102 + }, + { + "epoch": 1.794046558961964, + "grad_norm": 2.83587384223938, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8902925252914429, + "num_tokens": 547520555.0, + "step": 14103 + }, + { + "epoch": 1.7941737692405546, + "grad_norm": 2.979900360107422, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8788092136383057, + "num_tokens": 547555242.0, + "step": 14104 + }, + { + "epoch": 1.7943009795191451, + "grad_norm": 2.6391844749450684, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8879382610321045, + "num_tokens": 547596174.0, + "step": 14105 + }, + { + "epoch": 1.7944281897977357, + "grad_norm": 2.549330949783325, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8860970139503479, + "num_tokens": 547635678.0, + "step": 14106 + }, + { + "epoch": 1.7945554000763262, + "grad_norm": 2.6335065364837646, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8853219747543335, + "num_tokens": 547675043.0, + "step": 14107 + }, + { + "epoch": 1.7946826103549167, + "grad_norm": 2.758779764175415, + "learning_rate": 1e-06, + "loss": 0.3584, + "mean_token_accuracy": 0.87427818775177, + "num_tokens": 547715265.0, + "step": 14108 + }, + { + "epoch": 1.7948098206335072, + "grad_norm": 2.530515670776367, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8997517824172974, + "num_tokens": 547752098.0, + "step": 14109 + }, + { + "epoch": 1.7949370309120978, + "grad_norm": 2.6444590091705322, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.888100802898407, + "num_tokens": 547792829.0, + "step": 14110 + }, + { + "epoch": 1.7950642411906883, + "grad_norm": 2.399661064147949, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8933433890342712, + "num_tokens": 547837541.0, + "step": 14111 + }, + { + "epoch": 1.7951914514692788, + "grad_norm": 2.4595329761505127, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.90163254737854, + "num_tokens": 547876563.0, + "step": 14112 + }, + { + "epoch": 1.7953186617478694, + "grad_norm": 2.5288984775543213, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8960482478141785, + "num_tokens": 547913004.0, + "step": 14113 + }, + { + "epoch": 1.7954458720264599, + "grad_norm": 2.4561145305633545, + "learning_rate": 1e-06, + "loss": 0.3529, + "mean_token_accuracy": 0.8759335279464722, + "num_tokens": 547961354.0, + "step": 14114 + }, + { + "epoch": 1.7955730823050502, + "grad_norm": 2.7117624282836914, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8936892747879028, + "num_tokens": 547997621.0, + "step": 14115 + }, + { + "epoch": 1.7957002925836407, + "grad_norm": 2.5903940200805664, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8840506076812744, + "num_tokens": 548039330.0, + "step": 14116 + }, + { + "epoch": 1.7958275028622313, + "grad_norm": 2.5637781620025635, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8964638710021973, + "num_tokens": 548077462.0, + "step": 14117 + }, + { + "epoch": 1.7959547131408218, + "grad_norm": 2.5610833168029785, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8872700333595276, + "num_tokens": 548120103.0, + "step": 14118 + }, + { + "epoch": 1.7960819234194123, + "grad_norm": 2.898379325866699, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8963301181793213, + "num_tokens": 548152169.0, + "step": 14119 + }, + { + "epoch": 1.7962091336980028, + "grad_norm": 2.43947172164917, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8894504904747009, + "num_tokens": 548192133.0, + "step": 14120 + }, + { + "epoch": 1.7963363439765934, + "grad_norm": 2.7184906005859375, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8862829804420471, + "num_tokens": 548228729.0, + "step": 14121 + }, + { + "epoch": 1.7964635542551837, + "grad_norm": 2.4492671489715576, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8856148719787598, + "num_tokens": 548270444.0, + "step": 14122 + }, + { + "epoch": 1.7965907645337742, + "grad_norm": 2.378772735595703, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8921395540237427, + "num_tokens": 548314687.0, + "step": 14123 + }, + { + "epoch": 1.7967179748123647, + "grad_norm": 2.5692596435546875, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8743486404418945, + "num_tokens": 548360023.0, + "step": 14124 + }, + { + "epoch": 1.7968451850909553, + "grad_norm": 2.6225223541259766, + "learning_rate": 1e-06, + "loss": 0.3201, + "mean_token_accuracy": 0.8865479826927185, + "num_tokens": 548401440.0, + "step": 14125 + }, + { + "epoch": 1.7969723953695458, + "grad_norm": 2.564073324203491, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.900136411190033, + "num_tokens": 548437158.0, + "step": 14126 + }, + { + "epoch": 1.7970996056481363, + "grad_norm": 2.882571220397949, + "learning_rate": 1e-06, + "loss": 0.3743, + "mean_token_accuracy": 0.863858163356781, + "num_tokens": 548475253.0, + "step": 14127 + }, + { + "epoch": 1.7972268159267268, + "grad_norm": 2.7339015007019043, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9058793783187866, + "num_tokens": 548506485.0, + "step": 14128 + }, + { + "epoch": 1.7973540262053174, + "grad_norm": 2.700556755065918, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8843675851821899, + "num_tokens": 548543418.0, + "step": 14129 + }, + { + "epoch": 1.797481236483908, + "grad_norm": 2.54624342918396, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8914026021957397, + "num_tokens": 548582161.0, + "step": 14130 + }, + { + "epoch": 1.7976084467624984, + "grad_norm": 2.7241153717041016, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8865522742271423, + "num_tokens": 548616352.0, + "step": 14131 + }, + { + "epoch": 1.797735657041089, + "grad_norm": 2.4560835361480713, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8948553204536438, + "num_tokens": 548655986.0, + "step": 14132 + }, + { + "epoch": 1.7978628673196795, + "grad_norm": 3.004256248474121, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8962904810905457, + "num_tokens": 548687595.0, + "step": 14133 + }, + { + "epoch": 1.79799007759827, + "grad_norm": 2.6348307132720947, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.889812707901001, + "num_tokens": 548726992.0, + "step": 14134 + }, + { + "epoch": 1.7981172878768605, + "grad_norm": 2.5269834995269775, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8935134410858154, + "num_tokens": 548770022.0, + "step": 14135 + }, + { + "epoch": 1.798244498155451, + "grad_norm": 2.419718027114868, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8932344913482666, + "num_tokens": 548810718.0, + "step": 14136 + }, + { + "epoch": 1.7983717084340416, + "grad_norm": 2.4409213066101074, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.8992973566055298, + "num_tokens": 548850657.0, + "step": 14137 + }, + { + "epoch": 1.7984989187126321, + "grad_norm": 2.658169984817505, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.8924039602279663, + "num_tokens": 548888438.0, + "step": 14138 + }, + { + "epoch": 1.7986261289912227, + "grad_norm": 2.7156214714050293, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8863842487335205, + "num_tokens": 548926135.0, + "step": 14139 + }, + { + "epoch": 1.798753339269813, + "grad_norm": 2.564897298812866, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8958032727241516, + "num_tokens": 548965328.0, + "step": 14140 + }, + { + "epoch": 1.7988805495484035, + "grad_norm": 2.682589530944824, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8943849205970764, + "num_tokens": 549001205.0, + "step": 14141 + }, + { + "epoch": 1.799007759826994, + "grad_norm": 2.6532132625579834, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8903669118881226, + "num_tokens": 549039219.0, + "step": 14142 + }, + { + "epoch": 1.7991349701055845, + "grad_norm": 2.626253128051758, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.909130334854126, + "num_tokens": 549076484.0, + "step": 14143 + }, + { + "epoch": 1.799262180384175, + "grad_norm": 2.806788921356201, + "learning_rate": 1e-06, + "loss": 0.3263, + "mean_token_accuracy": 0.8839135766029358, + "num_tokens": 549113418.0, + "step": 14144 + }, + { + "epoch": 1.7993893906627656, + "grad_norm": 2.5068416595458984, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8966563940048218, + "num_tokens": 549153297.0, + "step": 14145 + }, + { + "epoch": 1.799516600941356, + "grad_norm": 2.578016996383667, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9036329984664917, + "num_tokens": 549190329.0, + "step": 14146 + }, + { + "epoch": 1.7996438112199464, + "grad_norm": 2.4506778717041016, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8966554403305054, + "num_tokens": 549231784.0, + "step": 14147 + }, + { + "epoch": 1.799771021498537, + "grad_norm": 2.99076771736145, + "learning_rate": 1e-06, + "loss": 0.3412, + "mean_token_accuracy": 0.8748756647109985, + "num_tokens": 549268692.0, + "step": 14148 + }, + { + "epoch": 1.7998982317771275, + "grad_norm": 2.5352065563201904, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8949971199035645, + "num_tokens": 549307579.0, + "step": 14149 + }, + { + "epoch": 1.800025442055718, + "grad_norm": 2.6469061374664307, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8973355293273926, + "num_tokens": 549342996.0, + "step": 14150 + }, + { + "epoch": 1.8001526523343085, + "grad_norm": 2.9714412689208984, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8765848875045776, + "num_tokens": 549376387.0, + "step": 14151 + }, + { + "epoch": 1.800279862612899, + "grad_norm": 2.807694435119629, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8842868804931641, + "num_tokens": 549412722.0, + "step": 14152 + }, + { + "epoch": 1.8004070728914896, + "grad_norm": 2.6978957653045654, + "learning_rate": 1e-06, + "loss": 0.3503, + "mean_token_accuracy": 0.8780745267868042, + "num_tokens": 549451673.0, + "step": 14153 + }, + { + "epoch": 1.8005342831700801, + "grad_norm": 2.520021438598633, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8863606452941895, + "num_tokens": 549495078.0, + "step": 14154 + }, + { + "epoch": 1.8006614934486707, + "grad_norm": 2.6927573680877686, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8918217420578003, + "num_tokens": 549536035.0, + "step": 14155 + }, + { + "epoch": 1.8007887037272612, + "grad_norm": 2.694739818572998, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8998346328735352, + "num_tokens": 549569140.0, + "step": 14156 + }, + { + "epoch": 1.8009159140058517, + "grad_norm": 2.8963632583618164, + "learning_rate": 1e-06, + "loss": 0.3632, + "mean_token_accuracy": 0.8813701868057251, + "num_tokens": 549606384.0, + "step": 14157 + }, + { + "epoch": 1.8010431242844422, + "grad_norm": 2.3000829219818115, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8957436084747314, + "num_tokens": 549649748.0, + "step": 14158 + }, + { + "epoch": 1.8011703345630328, + "grad_norm": 2.373364210128784, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8874728679656982, + "num_tokens": 549692421.0, + "step": 14159 + }, + { + "epoch": 1.8012975448416233, + "grad_norm": 2.6754934787750244, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.889404833316803, + "num_tokens": 549728621.0, + "step": 14160 + }, + { + "epoch": 1.8014247551202138, + "grad_norm": 2.5970261096954346, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.894007682800293, + "num_tokens": 549766526.0, + "step": 14161 + }, + { + "epoch": 1.8015519653988044, + "grad_norm": 2.350471019744873, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.889691948890686, + "num_tokens": 549812494.0, + "step": 14162 + }, + { + "epoch": 1.8016791756773949, + "grad_norm": 2.68778657913208, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8812854290008545, + "num_tokens": 549851932.0, + "step": 14163 + }, + { + "epoch": 1.8018063859559852, + "grad_norm": 2.373450517654419, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9001628160476685, + "num_tokens": 549895730.0, + "step": 14164 + }, + { + "epoch": 1.8019335962345757, + "grad_norm": 2.5061206817626953, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8926602602005005, + "num_tokens": 549939812.0, + "step": 14165 + }, + { + "epoch": 1.8020608065131662, + "grad_norm": 2.629673480987549, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8903160095214844, + "num_tokens": 549977857.0, + "step": 14166 + }, + { + "epoch": 1.8021880167917568, + "grad_norm": 2.5525214672088623, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8901138305664062, + "num_tokens": 550018645.0, + "step": 14167 + }, + { + "epoch": 1.8023152270703473, + "grad_norm": 2.576627731323242, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8885320425033569, + "num_tokens": 550059199.0, + "step": 14168 + }, + { + "epoch": 1.8024424373489378, + "grad_norm": 2.820315361022949, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8821852207183838, + "num_tokens": 550094418.0, + "step": 14169 + }, + { + "epoch": 1.8025696476275284, + "grad_norm": 2.7792670726776123, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8933943510055542, + "num_tokens": 550131717.0, + "step": 14170 + }, + { + "epoch": 1.8026968579061187, + "grad_norm": 2.6908352375030518, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8867462873458862, + "num_tokens": 550169305.0, + "step": 14171 + }, + { + "epoch": 1.8028240681847092, + "grad_norm": 2.597252607345581, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8877640962600708, + "num_tokens": 550209081.0, + "step": 14172 + }, + { + "epoch": 1.8029512784632997, + "grad_norm": 2.4904744625091553, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9052267074584961, + "num_tokens": 550248486.0, + "step": 14173 + }, + { + "epoch": 1.8030784887418903, + "grad_norm": 2.8436286449432373, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8853083848953247, + "num_tokens": 550285039.0, + "step": 14174 + }, + { + "epoch": 1.8032056990204808, + "grad_norm": 2.5754799842834473, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8878412246704102, + "num_tokens": 550326936.0, + "step": 14175 + }, + { + "epoch": 1.8033329092990713, + "grad_norm": 2.515669345855713, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8956719636917114, + "num_tokens": 550364846.0, + "step": 14176 + }, + { + "epoch": 1.8034601195776618, + "grad_norm": 2.4010562896728516, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8993081450462341, + "num_tokens": 550406457.0, + "step": 14177 + }, + { + "epoch": 1.8035873298562524, + "grad_norm": 2.5916895866394043, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9003806114196777, + "num_tokens": 550440781.0, + "step": 14178 + }, + { + "epoch": 1.803714540134843, + "grad_norm": 2.895648717880249, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8878471255302429, + "num_tokens": 550471783.0, + "step": 14179 + }, + { + "epoch": 1.8038417504134334, + "grad_norm": 2.353187084197998, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9105970859527588, + "num_tokens": 550510275.0, + "step": 14180 + }, + { + "epoch": 1.803968960692024, + "grad_norm": 2.3150718212127686, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9105032682418823, + "num_tokens": 550551134.0, + "step": 14181 + }, + { + "epoch": 1.8040961709706145, + "grad_norm": 2.5049946308135986, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8915367126464844, + "num_tokens": 550593195.0, + "step": 14182 + }, + { + "epoch": 1.804223381249205, + "grad_norm": 2.77846360206604, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8941277265548706, + "num_tokens": 550628560.0, + "step": 14183 + }, + { + "epoch": 1.8043505915277955, + "grad_norm": 2.620201826095581, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8995997905731201, + "num_tokens": 550662929.0, + "step": 14184 + }, + { + "epoch": 1.804477801806386, + "grad_norm": 2.630972385406494, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8928079605102539, + "num_tokens": 550700774.0, + "step": 14185 + }, + { + "epoch": 1.8046050120849766, + "grad_norm": 2.7796084880828857, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8816357851028442, + "num_tokens": 550738026.0, + "step": 14186 + }, + { + "epoch": 1.8047322223635671, + "grad_norm": 2.6092135906219482, + "learning_rate": 1e-06, + "loss": 0.3154, + "mean_token_accuracy": 0.8889736533164978, + "num_tokens": 550778464.0, + "step": 14187 + }, + { + "epoch": 1.8048594326421576, + "grad_norm": 2.986922025680542, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8900713920593262, + "num_tokens": 550809023.0, + "step": 14188 + }, + { + "epoch": 1.804986642920748, + "grad_norm": 2.870452880859375, + "learning_rate": 1e-06, + "loss": 0.3394, + "mean_token_accuracy": 0.8818804621696472, + "num_tokens": 550844339.0, + "step": 14189 + }, + { + "epoch": 1.8051138531993385, + "grad_norm": 2.4327280521392822, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.8987038135528564, + "num_tokens": 550885183.0, + "step": 14190 + }, + { + "epoch": 1.805241063477929, + "grad_norm": 2.6235296726226807, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8860677480697632, + "num_tokens": 550924198.0, + "step": 14191 + }, + { + "epoch": 1.8053682737565195, + "grad_norm": 2.5894970893859863, + "learning_rate": 1e-06, + "loss": 0.3172, + "mean_token_accuracy": 0.8860961198806763, + "num_tokens": 550965326.0, + "step": 14192 + }, + { + "epoch": 1.80549548403511, + "grad_norm": 2.810818910598755, + "learning_rate": 1e-06, + "loss": 0.3525, + "mean_token_accuracy": 0.8771121501922607, + "num_tokens": 551001820.0, + "step": 14193 + }, + { + "epoch": 1.8056226943137006, + "grad_norm": 2.6892073154449463, + "learning_rate": 1e-06, + "loss": 0.3212, + "mean_token_accuracy": 0.8836013674736023, + "num_tokens": 551038713.0, + "step": 14194 + }, + { + "epoch": 1.805749904592291, + "grad_norm": 2.6659440994262695, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9023216962814331, + "num_tokens": 551071786.0, + "step": 14195 + }, + { + "epoch": 1.8058771148708814, + "grad_norm": 2.452085256576538, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8937245607376099, + "num_tokens": 551114138.0, + "step": 14196 + }, + { + "epoch": 1.806004325149472, + "grad_norm": 2.6902921199798584, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8917552828788757, + "num_tokens": 551156111.0, + "step": 14197 + }, + { + "epoch": 1.8061315354280625, + "grad_norm": 2.457943916320801, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.893151044845581, + "num_tokens": 551202128.0, + "step": 14198 + }, + { + "epoch": 1.806258745706653, + "grad_norm": 2.6397669315338135, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8959603905677795, + "num_tokens": 551244455.0, + "step": 14199 + }, + { + "epoch": 1.8063859559852435, + "grad_norm": 2.4343645572662354, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8927293419837952, + "num_tokens": 551289388.0, + "step": 14200 + }, + { + "epoch": 1.806513166263834, + "grad_norm": 2.2843239307403564, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9085828065872192, + "num_tokens": 551331568.0, + "step": 14201 + }, + { + "epoch": 1.8066403765424246, + "grad_norm": 2.3039474487304688, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8991494178771973, + "num_tokens": 551375986.0, + "step": 14202 + }, + { + "epoch": 1.8067675868210151, + "grad_norm": 2.573806047439575, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8954797983169556, + "num_tokens": 551414946.0, + "step": 14203 + }, + { + "epoch": 1.8068947970996057, + "grad_norm": 2.4203779697418213, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.9000520706176758, + "num_tokens": 551452296.0, + "step": 14204 + }, + { + "epoch": 1.8070220073781962, + "grad_norm": 2.567551374435425, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8800784349441528, + "num_tokens": 551495826.0, + "step": 14205 + }, + { + "epoch": 1.8071492176567867, + "grad_norm": 2.599771022796631, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.891433596611023, + "num_tokens": 551535547.0, + "step": 14206 + }, + { + "epoch": 1.8072764279353772, + "grad_norm": 2.4601356983184814, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8957360982894897, + "num_tokens": 551574402.0, + "step": 14207 + }, + { + "epoch": 1.8074036382139678, + "grad_norm": 2.794494867324829, + "learning_rate": 1e-06, + "loss": 0.3407, + "mean_token_accuracy": 0.8801594972610474, + "num_tokens": 551611153.0, + "step": 14208 + }, + { + "epoch": 1.8075308484925583, + "grad_norm": 2.5936644077301025, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8932356834411621, + "num_tokens": 551651345.0, + "step": 14209 + }, + { + "epoch": 1.8076580587711488, + "grad_norm": 2.5971875190734863, + "learning_rate": 1e-06, + "loss": 0.3295, + "mean_token_accuracy": 0.8820270895957947, + "num_tokens": 551691419.0, + "step": 14210 + }, + { + "epoch": 1.8077852690497394, + "grad_norm": 2.7721962928771973, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8890301585197449, + "num_tokens": 551730810.0, + "step": 14211 + }, + { + "epoch": 1.8079124793283299, + "grad_norm": 2.7446396350860596, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9000484347343445, + "num_tokens": 551767771.0, + "step": 14212 + }, + { + "epoch": 1.8080396896069202, + "grad_norm": 2.6628503799438477, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8959734439849854, + "num_tokens": 551805843.0, + "step": 14213 + }, + { + "epoch": 1.8081668998855107, + "grad_norm": 2.599478006362915, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8799885511398315, + "num_tokens": 551845401.0, + "step": 14214 + }, + { + "epoch": 1.8082941101641012, + "grad_norm": 2.662405252456665, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9042626023292542, + "num_tokens": 551880101.0, + "step": 14215 + }, + { + "epoch": 1.8084213204426918, + "grad_norm": 2.6235883235931396, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.897314190864563, + "num_tokens": 551920276.0, + "step": 14216 + }, + { + "epoch": 1.8085485307212823, + "grad_norm": 2.4697322845458984, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8900305032730103, + "num_tokens": 551967969.0, + "step": 14217 + }, + { + "epoch": 1.8086757409998728, + "grad_norm": 2.829890489578247, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8879360556602478, + "num_tokens": 552001747.0, + "step": 14218 + }, + { + "epoch": 1.8088029512784631, + "grad_norm": 2.4803507328033447, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8927037715911865, + "num_tokens": 552046752.0, + "step": 14219 + }, + { + "epoch": 1.8089301615570537, + "grad_norm": 2.7199463844299316, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8852483034133911, + "num_tokens": 552085189.0, + "step": 14220 + }, + { + "epoch": 1.8090573718356442, + "grad_norm": 2.7408053874969482, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8957285284996033, + "num_tokens": 552119579.0, + "step": 14221 + }, + { + "epoch": 1.8091845821142347, + "grad_norm": 2.797879695892334, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8911754488945007, + "num_tokens": 552154031.0, + "step": 14222 + }, + { + "epoch": 1.8093117923928252, + "grad_norm": 2.8678085803985596, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8900610208511353, + "num_tokens": 552188857.0, + "step": 14223 + }, + { + "epoch": 1.8094390026714158, + "grad_norm": 2.770221471786499, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8860535621643066, + "num_tokens": 552224030.0, + "step": 14224 + }, + { + "epoch": 1.8095662129500063, + "grad_norm": 2.687431812286377, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8932318091392517, + "num_tokens": 552259972.0, + "step": 14225 + }, + { + "epoch": 1.8096934232285968, + "grad_norm": 2.6855287551879883, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8896320462226868, + "num_tokens": 552298254.0, + "step": 14226 + }, + { + "epoch": 1.8098206335071874, + "grad_norm": 2.8441672325134277, + "learning_rate": 1e-06, + "loss": 0.3527, + "mean_token_accuracy": 0.8734732270240784, + "num_tokens": 552338951.0, + "step": 14227 + }, + { + "epoch": 1.8099478437857779, + "grad_norm": 2.6266016960144043, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8935943841934204, + "num_tokens": 552378983.0, + "step": 14228 + }, + { + "epoch": 1.8100750540643684, + "grad_norm": 2.8191819190979004, + "learning_rate": 1e-06, + "loss": 0.3408, + "mean_token_accuracy": 0.8785058259963989, + "num_tokens": 552416476.0, + "step": 14229 + }, + { + "epoch": 1.810202264342959, + "grad_norm": 2.4948813915252686, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8873010277748108, + "num_tokens": 552461120.0, + "step": 14230 + }, + { + "epoch": 1.8103294746215495, + "grad_norm": 2.7084410190582275, + "learning_rate": 1e-06, + "loss": 0.3224, + "mean_token_accuracy": 0.8813590407371521, + "num_tokens": 552498947.0, + "step": 14231 + }, + { + "epoch": 1.81045668490014, + "grad_norm": 2.510774850845337, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.8986963033676147, + "num_tokens": 552540504.0, + "step": 14232 + }, + { + "epoch": 1.8105838951787305, + "grad_norm": 2.880809783935547, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8883224725723267, + "num_tokens": 552572346.0, + "step": 14233 + }, + { + "epoch": 1.810711105457321, + "grad_norm": 2.7862133979797363, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8827147483825684, + "num_tokens": 552608186.0, + "step": 14234 + }, + { + "epoch": 1.8108383157359116, + "grad_norm": 2.762699842453003, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8762751817703247, + "num_tokens": 552647434.0, + "step": 14235 + }, + { + "epoch": 1.8109655260145021, + "grad_norm": 2.77211332321167, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.892092227935791, + "num_tokens": 552686258.0, + "step": 14236 + }, + { + "epoch": 1.8110927362930926, + "grad_norm": 2.5885143280029297, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8960279226303101, + "num_tokens": 552724135.0, + "step": 14237 + }, + { + "epoch": 1.811219946571683, + "grad_norm": 2.907097578048706, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8955889940261841, + "num_tokens": 552757090.0, + "step": 14238 + }, + { + "epoch": 1.8113471568502735, + "grad_norm": 2.6841461658477783, + "learning_rate": 1e-06, + "loss": 0.342, + "mean_token_accuracy": 0.8790680170059204, + "num_tokens": 552795918.0, + "step": 14239 + }, + { + "epoch": 1.811474367128864, + "grad_norm": 2.51159930229187, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8914980888366699, + "num_tokens": 552835298.0, + "step": 14240 + }, + { + "epoch": 1.8116015774074545, + "grad_norm": 2.8989181518554688, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8784735202789307, + "num_tokens": 552871349.0, + "step": 14241 + }, + { + "epoch": 1.811728787686045, + "grad_norm": 2.551267385482788, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9054807424545288, + "num_tokens": 552907812.0, + "step": 14242 + }, + { + "epoch": 1.8118559979646356, + "grad_norm": 2.6552553176879883, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8870707750320435, + "num_tokens": 552946453.0, + "step": 14243 + }, + { + "epoch": 1.811983208243226, + "grad_norm": 2.8378100395202637, + "learning_rate": 1e-06, + "loss": 0.3422, + "mean_token_accuracy": 0.8755764961242676, + "num_tokens": 552983035.0, + "step": 14244 + }, + { + "epoch": 1.8121104185218164, + "grad_norm": 2.7458713054656982, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8912878036499023, + "num_tokens": 553017777.0, + "step": 14245 + }, + { + "epoch": 1.812237628800407, + "grad_norm": 2.5601344108581543, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.889560878276825, + "num_tokens": 553057686.0, + "step": 14246 + }, + { + "epoch": 1.8123648390789975, + "grad_norm": 2.41988205909729, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8968216776847839, + "num_tokens": 553103223.0, + "step": 14247 + }, + { + "epoch": 1.812492049357588, + "grad_norm": 2.5667853355407715, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8781293630599976, + "num_tokens": 553147857.0, + "step": 14248 + }, + { + "epoch": 1.8126192596361785, + "grad_norm": 2.49596905708313, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8883901238441467, + "num_tokens": 553190722.0, + "step": 14249 + }, + { + "epoch": 1.812746469914769, + "grad_norm": 2.7944753170013428, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8879427909851074, + "num_tokens": 553223428.0, + "step": 14250 + }, + { + "epoch": 1.8128736801933596, + "grad_norm": 2.8028337955474854, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8828617334365845, + "num_tokens": 553262016.0, + "step": 14251 + }, + { + "epoch": 1.8130008904719501, + "grad_norm": 2.6963138580322266, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8908822536468506, + "num_tokens": 553299221.0, + "step": 14252 + }, + { + "epoch": 1.8131281007505406, + "grad_norm": 2.509036064147949, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8790470957756042, + "num_tokens": 553346211.0, + "step": 14253 + }, + { + "epoch": 1.8132553110291312, + "grad_norm": 3.0027105808258057, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.8773906230926514, + "num_tokens": 553378765.0, + "step": 14254 + }, + { + "epoch": 1.8133825213077217, + "grad_norm": 2.3691134452819824, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8934515714645386, + "num_tokens": 553420292.0, + "step": 14255 + }, + { + "epoch": 1.8135097315863122, + "grad_norm": 2.74821400642395, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8897131681442261, + "num_tokens": 553457458.0, + "step": 14256 + }, + { + "epoch": 1.8136369418649028, + "grad_norm": 2.9065797328948975, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.893013060092926, + "num_tokens": 553487248.0, + "step": 14257 + }, + { + "epoch": 1.8137641521434933, + "grad_norm": 2.9862935543060303, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.86761873960495, + "num_tokens": 553524209.0, + "step": 14258 + }, + { + "epoch": 1.8138913624220838, + "grad_norm": 2.5263829231262207, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.893687903881073, + "num_tokens": 553565494.0, + "step": 14259 + }, + { + "epoch": 1.8140185727006743, + "grad_norm": 2.354556083679199, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8905830979347229, + "num_tokens": 553610649.0, + "step": 14260 + }, + { + "epoch": 1.8141457829792649, + "grad_norm": 2.580225706100464, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8885787129402161, + "num_tokens": 553652457.0, + "step": 14261 + }, + { + "epoch": 1.8142729932578552, + "grad_norm": 2.5741641521453857, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8954266309738159, + "num_tokens": 553688646.0, + "step": 14262 + }, + { + "epoch": 1.8144002035364457, + "grad_norm": 2.502110719680786, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8949974775314331, + "num_tokens": 553729057.0, + "step": 14263 + }, + { + "epoch": 1.8145274138150362, + "grad_norm": 2.454528331756592, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9027822017669678, + "num_tokens": 553767437.0, + "step": 14264 + }, + { + "epoch": 1.8146546240936268, + "grad_norm": 2.4732627868652344, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8953751921653748, + "num_tokens": 553808087.0, + "step": 14265 + }, + { + "epoch": 1.8147818343722173, + "grad_norm": 2.5167770385742188, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8928230404853821, + "num_tokens": 553845243.0, + "step": 14266 + }, + { + "epoch": 1.8149090446508078, + "grad_norm": 2.6636178493499756, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8891148567199707, + "num_tokens": 553880810.0, + "step": 14267 + }, + { + "epoch": 1.8150362549293981, + "grad_norm": 2.442553758621216, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.8999242782592773, + "num_tokens": 553921931.0, + "step": 14268 + }, + { + "epoch": 1.8151634652079887, + "grad_norm": 2.5802853107452393, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8885726928710938, + "num_tokens": 553963446.0, + "step": 14269 + }, + { + "epoch": 1.8152906754865792, + "grad_norm": 2.7216200828552246, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8878536224365234, + "num_tokens": 553999235.0, + "step": 14270 + }, + { + "epoch": 1.8154178857651697, + "grad_norm": 2.551565408706665, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9019293785095215, + "num_tokens": 554036166.0, + "step": 14271 + }, + { + "epoch": 1.8155450960437602, + "grad_norm": 2.69708251953125, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8897526264190674, + "num_tokens": 554071170.0, + "step": 14272 + }, + { + "epoch": 1.8156723063223508, + "grad_norm": 2.4628725051879883, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9032028913497925, + "num_tokens": 554110572.0, + "step": 14273 + }, + { + "epoch": 1.8157995166009413, + "grad_norm": 2.6860766410827637, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8839080929756165, + "num_tokens": 554150292.0, + "step": 14274 + }, + { + "epoch": 1.8159267268795318, + "grad_norm": 2.3510310649871826, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9037346839904785, + "num_tokens": 554189970.0, + "step": 14275 + }, + { + "epoch": 1.8160539371581224, + "grad_norm": 2.9967055320739746, + "learning_rate": 1e-06, + "loss": 0.3734, + "mean_token_accuracy": 0.86940598487854, + "num_tokens": 554225749.0, + "step": 14276 + }, + { + "epoch": 1.8161811474367129, + "grad_norm": 2.8861019611358643, + "learning_rate": 1e-06, + "loss": 0.3468, + "mean_token_accuracy": 0.876008152961731, + "num_tokens": 554261310.0, + "step": 14277 + }, + { + "epoch": 1.8163083577153034, + "grad_norm": 2.8849411010742188, + "learning_rate": 1e-06, + "loss": 0.327, + "mean_token_accuracy": 0.8836175203323364, + "num_tokens": 554297436.0, + "step": 14278 + }, + { + "epoch": 1.816435567993894, + "grad_norm": 2.5640926361083984, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9011824131011963, + "num_tokens": 554334380.0, + "step": 14279 + }, + { + "epoch": 1.8165627782724845, + "grad_norm": 2.5645408630371094, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8960909843444824, + "num_tokens": 554375475.0, + "step": 14280 + }, + { + "epoch": 1.816689988551075, + "grad_norm": 2.592529535293579, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8954452872276306, + "num_tokens": 554413769.0, + "step": 14281 + }, + { + "epoch": 1.8168171988296655, + "grad_norm": 2.508425712585449, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8954168558120728, + "num_tokens": 554452247.0, + "step": 14282 + }, + { + "epoch": 1.816944409108256, + "grad_norm": 2.56966233253479, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8925907611846924, + "num_tokens": 554492700.0, + "step": 14283 + }, + { + "epoch": 1.8170716193868466, + "grad_norm": 2.5323545932769775, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8906019330024719, + "num_tokens": 554536309.0, + "step": 14284 + }, + { + "epoch": 1.817198829665437, + "grad_norm": 2.5958118438720703, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.900375485420227, + "num_tokens": 554571919.0, + "step": 14285 + }, + { + "epoch": 1.8173260399440276, + "grad_norm": 2.720296859741211, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8959835767745972, + "num_tokens": 554609213.0, + "step": 14286 + }, + { + "epoch": 1.817453250222618, + "grad_norm": 2.8304803371429443, + "learning_rate": 1e-06, + "loss": 0.3434, + "mean_token_accuracy": 0.8800060749053955, + "num_tokens": 554644891.0, + "step": 14287 + }, + { + "epoch": 1.8175804605012085, + "grad_norm": 2.639356851577759, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8954780101776123, + "num_tokens": 554679294.0, + "step": 14288 + }, + { + "epoch": 1.817707670779799, + "grad_norm": 2.5557262897491455, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8997910022735596, + "num_tokens": 554714158.0, + "step": 14289 + }, + { + "epoch": 1.8178348810583895, + "grad_norm": 2.4873251914978027, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8857946991920471, + "num_tokens": 554755646.0, + "step": 14290 + }, + { + "epoch": 1.81796209133698, + "grad_norm": 2.4022464752197266, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9054156541824341, + "num_tokens": 554792899.0, + "step": 14291 + }, + { + "epoch": 1.8180893016155706, + "grad_norm": 2.558720588684082, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8991157412528992, + "num_tokens": 554827001.0, + "step": 14292 + }, + { + "epoch": 1.818216511894161, + "grad_norm": 2.7752718925476074, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8890770673751831, + "num_tokens": 554863472.0, + "step": 14293 + }, + { + "epoch": 1.8183437221727514, + "grad_norm": 2.669914484024048, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.896875262260437, + "num_tokens": 554896715.0, + "step": 14294 + }, + { + "epoch": 1.818470932451342, + "grad_norm": 2.7731385231018066, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.889925479888916, + "num_tokens": 554931155.0, + "step": 14295 + }, + { + "epoch": 1.8185981427299325, + "grad_norm": 2.7045652866363525, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.897680401802063, + "num_tokens": 554963581.0, + "step": 14296 + }, + { + "epoch": 1.818725353008523, + "grad_norm": 2.8165090084075928, + "learning_rate": 1e-06, + "loss": 0.3442, + "mean_token_accuracy": 0.8802636861801147, + "num_tokens": 555001287.0, + "step": 14297 + }, + { + "epoch": 1.8188525632871135, + "grad_norm": 2.395256519317627, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8834556341171265, + "num_tokens": 555046733.0, + "step": 14298 + }, + { + "epoch": 1.818979773565704, + "grad_norm": 2.6602578163146973, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8912781476974487, + "num_tokens": 555083796.0, + "step": 14299 + }, + { + "epoch": 1.8191069838442946, + "grad_norm": 2.698975086212158, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8869181275367737, + "num_tokens": 555123657.0, + "step": 14300 + }, + { + "epoch": 1.8192341941228851, + "grad_norm": 2.7376391887664795, + "learning_rate": 1e-06, + "loss": 0.3364, + "mean_token_accuracy": 0.8801697492599487, + "num_tokens": 555162602.0, + "step": 14301 + }, + { + "epoch": 1.8193614044014756, + "grad_norm": 2.617640495300293, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8893941640853882, + "num_tokens": 555201384.0, + "step": 14302 + }, + { + "epoch": 1.8194886146800662, + "grad_norm": 2.781341791152954, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8753647804260254, + "num_tokens": 555242580.0, + "step": 14303 + }, + { + "epoch": 1.8196158249586567, + "grad_norm": 2.65838360786438, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8956340551376343, + "num_tokens": 555279659.0, + "step": 14304 + }, + { + "epoch": 1.8197430352372472, + "grad_norm": 2.5649173259735107, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.8953231573104858, + "num_tokens": 555317068.0, + "step": 14305 + }, + { + "epoch": 1.8198702455158378, + "grad_norm": 2.6625125408172607, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8914029598236084, + "num_tokens": 555357270.0, + "step": 14306 + }, + { + "epoch": 1.8199974557944283, + "grad_norm": 2.7407660484313965, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8904495239257812, + "num_tokens": 555390397.0, + "step": 14307 + }, + { + "epoch": 1.8201246660730188, + "grad_norm": 2.3514513969421387, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9011443853378296, + "num_tokens": 555433418.0, + "step": 14308 + }, + { + "epoch": 1.8202518763516093, + "grad_norm": 2.652341365814209, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9048495292663574, + "num_tokens": 555466200.0, + "step": 14309 + }, + { + "epoch": 1.8203790866301999, + "grad_norm": 2.3581466674804688, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8902459144592285, + "num_tokens": 555512493.0, + "step": 14310 + }, + { + "epoch": 1.8205062969087902, + "grad_norm": 2.688815116882324, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8825671672821045, + "num_tokens": 555552656.0, + "step": 14311 + }, + { + "epoch": 1.8206335071873807, + "grad_norm": 2.5764641761779785, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8982117772102356, + "num_tokens": 555590263.0, + "step": 14312 + }, + { + "epoch": 1.8207607174659712, + "grad_norm": 2.576002597808838, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8862659931182861, + "num_tokens": 555630013.0, + "step": 14313 + }, + { + "epoch": 1.8208879277445618, + "grad_norm": 2.4120380878448486, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.898993730545044, + "num_tokens": 555674281.0, + "step": 14314 + }, + { + "epoch": 1.8210151380231523, + "grad_norm": 2.698908567428589, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8916406631469727, + "num_tokens": 555710307.0, + "step": 14315 + }, + { + "epoch": 1.8211423483017428, + "grad_norm": 2.524728775024414, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8879396915435791, + "num_tokens": 555752213.0, + "step": 14316 + }, + { + "epoch": 1.8212695585803331, + "grad_norm": 2.8104281425476074, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8984866142272949, + "num_tokens": 555784383.0, + "step": 14317 + }, + { + "epoch": 1.8213967688589237, + "grad_norm": 2.5822954177856445, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8933612704277039, + "num_tokens": 555825138.0, + "step": 14318 + }, + { + "epoch": 1.8215239791375142, + "grad_norm": 2.5780017375946045, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.893685519695282, + "num_tokens": 555865642.0, + "step": 14319 + }, + { + "epoch": 1.8216511894161047, + "grad_norm": 2.28356671333313, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.904381275177002, + "num_tokens": 555908532.0, + "step": 14320 + }, + { + "epoch": 1.8217783996946952, + "grad_norm": 2.5487489700317383, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8933405876159668, + "num_tokens": 555948313.0, + "step": 14321 + }, + { + "epoch": 1.8219056099732858, + "grad_norm": 2.609616279602051, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8974220156669617, + "num_tokens": 555987054.0, + "step": 14322 + }, + { + "epoch": 1.8220328202518763, + "grad_norm": 2.8603763580322266, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8888546228408813, + "num_tokens": 556021665.0, + "step": 14323 + }, + { + "epoch": 1.8221600305304668, + "grad_norm": 2.872072458267212, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8938722014427185, + "num_tokens": 556052639.0, + "step": 14324 + }, + { + "epoch": 1.8222872408090574, + "grad_norm": 2.536728858947754, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9020772576332092, + "num_tokens": 556091654.0, + "step": 14325 + }, + { + "epoch": 1.8224144510876479, + "grad_norm": 2.6958041191101074, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8888444900512695, + "num_tokens": 556126810.0, + "step": 14326 + }, + { + "epoch": 1.8225416613662384, + "grad_norm": 2.4869697093963623, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8984322547912598, + "num_tokens": 556164594.0, + "step": 14327 + }, + { + "epoch": 1.822668871644829, + "grad_norm": 2.416163444519043, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9026057124137878, + "num_tokens": 556203639.0, + "step": 14328 + }, + { + "epoch": 1.8227960819234195, + "grad_norm": 2.5741519927978516, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8954086303710938, + "num_tokens": 556243810.0, + "step": 14329 + }, + { + "epoch": 1.82292329220201, + "grad_norm": 2.597241163253784, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9028339982032776, + "num_tokens": 556284194.0, + "step": 14330 + }, + { + "epoch": 1.8230505024806005, + "grad_norm": 2.61702036857605, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8934626579284668, + "num_tokens": 556321570.0, + "step": 14331 + }, + { + "epoch": 1.823177712759191, + "grad_norm": 2.7178561687469482, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8943426012992859, + "num_tokens": 556355557.0, + "step": 14332 + }, + { + "epoch": 1.8233049230377816, + "grad_norm": 2.4475533962249756, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8939309120178223, + "num_tokens": 556397328.0, + "step": 14333 + }, + { + "epoch": 1.823432133316372, + "grad_norm": 2.6037817001342773, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8793291449546814, + "num_tokens": 556437747.0, + "step": 14334 + }, + { + "epoch": 1.8235593435949626, + "grad_norm": 2.368886709213257, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.892553985118866, + "num_tokens": 556479194.0, + "step": 14335 + }, + { + "epoch": 1.823686553873553, + "grad_norm": 2.4645018577575684, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8899041414260864, + "num_tokens": 556520892.0, + "step": 14336 + }, + { + "epoch": 1.8238137641521435, + "grad_norm": 2.9890787601470947, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.888001561164856, + "num_tokens": 556551385.0, + "step": 14337 + }, + { + "epoch": 1.823940974430734, + "grad_norm": 2.6732521057128906, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8786506056785583, + "num_tokens": 556590288.0, + "step": 14338 + }, + { + "epoch": 1.8240681847093245, + "grad_norm": 2.7524564266204834, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8928744792938232, + "num_tokens": 556628885.0, + "step": 14339 + }, + { + "epoch": 1.824195394987915, + "grad_norm": 2.6034533977508545, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8978101015090942, + "num_tokens": 556665900.0, + "step": 14340 + }, + { + "epoch": 1.8243226052665056, + "grad_norm": 2.6093997955322266, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9040164947509766, + "num_tokens": 556698183.0, + "step": 14341 + }, + { + "epoch": 1.8244498155450959, + "grad_norm": 2.4933645725250244, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9029985666275024, + "num_tokens": 556735705.0, + "step": 14342 + }, + { + "epoch": 1.8245770258236864, + "grad_norm": 2.4941916465759277, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9027608633041382, + "num_tokens": 556776927.0, + "step": 14343 + }, + { + "epoch": 1.824704236102277, + "grad_norm": 2.7443161010742188, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8953256607055664, + "num_tokens": 556817074.0, + "step": 14344 + }, + { + "epoch": 1.8248314463808675, + "grad_norm": 2.8465702533721924, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8908658027648926, + "num_tokens": 556851718.0, + "step": 14345 + }, + { + "epoch": 1.824958656659458, + "grad_norm": 2.4560706615448, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9033642411231995, + "num_tokens": 556887036.0, + "step": 14346 + }, + { + "epoch": 1.8250858669380485, + "grad_norm": 2.6068005561828613, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8946958780288696, + "num_tokens": 556927895.0, + "step": 14347 + }, + { + "epoch": 1.825213077216639, + "grad_norm": 2.654879331588745, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8836556673049927, + "num_tokens": 556969810.0, + "step": 14348 + }, + { + "epoch": 1.8253402874952296, + "grad_norm": 2.668485641479492, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8950490951538086, + "num_tokens": 557007421.0, + "step": 14349 + }, + { + "epoch": 1.8254674977738201, + "grad_norm": 2.3628432750701904, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8930315375328064, + "num_tokens": 557051071.0, + "step": 14350 + }, + { + "epoch": 1.8255947080524106, + "grad_norm": 2.6346447467803955, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8944913148880005, + "num_tokens": 557091525.0, + "step": 14351 + }, + { + "epoch": 1.8257219183310012, + "grad_norm": 2.4665863513946533, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.89512038230896, + "num_tokens": 557135429.0, + "step": 14352 + }, + { + "epoch": 1.8258491286095917, + "grad_norm": 2.469343662261963, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8872816562652588, + "num_tokens": 557177326.0, + "step": 14353 + }, + { + "epoch": 1.8259763388881822, + "grad_norm": 2.4083683490753174, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8944202661514282, + "num_tokens": 557218118.0, + "step": 14354 + }, + { + "epoch": 1.8261035491667728, + "grad_norm": 2.8227245807647705, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8845895528793335, + "num_tokens": 557253020.0, + "step": 14355 + }, + { + "epoch": 1.8262307594453633, + "grad_norm": 2.438070058822632, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8989920616149902, + "num_tokens": 557295064.0, + "step": 14356 + }, + { + "epoch": 1.8263579697239538, + "grad_norm": 2.635972738265991, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8911881446838379, + "num_tokens": 557333043.0, + "step": 14357 + }, + { + "epoch": 1.8264851800025443, + "grad_norm": 2.6908035278320312, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8860909938812256, + "num_tokens": 557373051.0, + "step": 14358 + }, + { + "epoch": 1.8266123902811349, + "grad_norm": 2.3743181228637695, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8982980251312256, + "num_tokens": 557416201.0, + "step": 14359 + }, + { + "epoch": 1.8267396005597252, + "grad_norm": 2.5127227306365967, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8987891674041748, + "num_tokens": 557456051.0, + "step": 14360 + }, + { + "epoch": 1.8268668108383157, + "grad_norm": 2.6580593585968018, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8948094844818115, + "num_tokens": 557491866.0, + "step": 14361 + }, + { + "epoch": 1.8269940211169062, + "grad_norm": 2.5908217430114746, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9041315317153931, + "num_tokens": 557528145.0, + "step": 14362 + }, + { + "epoch": 1.8271212313954968, + "grad_norm": 2.42475962638855, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8928831815719604, + "num_tokens": 557568923.0, + "step": 14363 + }, + { + "epoch": 1.8272484416740873, + "grad_norm": 2.814716100692749, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8933297395706177, + "num_tokens": 557604565.0, + "step": 14364 + }, + { + "epoch": 1.8273756519526778, + "grad_norm": 2.6955666542053223, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.9011064767837524, + "num_tokens": 557637842.0, + "step": 14365 + }, + { + "epoch": 1.8275028622312681, + "grad_norm": 2.515767812728882, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8961164951324463, + "num_tokens": 557674688.0, + "step": 14366 + }, + { + "epoch": 1.8276300725098586, + "grad_norm": 2.35115122795105, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8960254192352295, + "num_tokens": 557719715.0, + "step": 14367 + }, + { + "epoch": 1.8277572827884492, + "grad_norm": 2.559692144393921, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8942017555236816, + "num_tokens": 557757857.0, + "step": 14368 + }, + { + "epoch": 1.8278844930670397, + "grad_norm": 2.556015729904175, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8920930027961731, + "num_tokens": 557796874.0, + "step": 14369 + }, + { + "epoch": 1.8280117033456302, + "grad_norm": 2.6678528785705566, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.886868417263031, + "num_tokens": 557837469.0, + "step": 14370 + }, + { + "epoch": 1.8281389136242208, + "grad_norm": 2.683370351791382, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.887138843536377, + "num_tokens": 557874948.0, + "step": 14371 + }, + { + "epoch": 1.8282661239028113, + "grad_norm": 2.6751410961151123, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9052959084510803, + "num_tokens": 557904326.0, + "step": 14372 + }, + { + "epoch": 1.8283933341814018, + "grad_norm": 2.5932388305664062, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.886047899723053, + "num_tokens": 557944466.0, + "step": 14373 + }, + { + "epoch": 1.8285205444599923, + "grad_norm": 2.4090545177459717, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9065028429031372, + "num_tokens": 557983744.0, + "step": 14374 + }, + { + "epoch": 1.8286477547385829, + "grad_norm": 2.628100633621216, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8904284834861755, + "num_tokens": 558021711.0, + "step": 14375 + }, + { + "epoch": 1.8287749650171734, + "grad_norm": 2.7602176666259766, + "learning_rate": 1e-06, + "loss": 0.3347, + "mean_token_accuracy": 0.8826731443405151, + "num_tokens": 558062535.0, + "step": 14376 + }, + { + "epoch": 1.828902175295764, + "grad_norm": 2.677443742752075, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.8999930024147034, + "num_tokens": 558096490.0, + "step": 14377 + }, + { + "epoch": 1.8290293855743545, + "grad_norm": 2.6314737796783447, + "learning_rate": 1e-06, + "loss": 0.3301, + "mean_token_accuracy": 0.8817664384841919, + "num_tokens": 558138228.0, + "step": 14378 + }, + { + "epoch": 1.829156595852945, + "grad_norm": 2.548757314682007, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8947051763534546, + "num_tokens": 558179827.0, + "step": 14379 + }, + { + "epoch": 1.8292838061315355, + "grad_norm": 2.799595832824707, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8858115077018738, + "num_tokens": 558215217.0, + "step": 14380 + }, + { + "epoch": 1.829411016410126, + "grad_norm": 2.39081072807312, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9042987823486328, + "num_tokens": 558256151.0, + "step": 14381 + }, + { + "epoch": 1.8295382266887166, + "grad_norm": 2.533911943435669, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8965141177177429, + "num_tokens": 558292989.0, + "step": 14382 + }, + { + "epoch": 1.829665436967307, + "grad_norm": 2.7691752910614014, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8982892632484436, + "num_tokens": 558326449.0, + "step": 14383 + }, + { + "epoch": 1.8297926472458976, + "grad_norm": 2.5454444885253906, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8982279300689697, + "num_tokens": 558364818.0, + "step": 14384 + }, + { + "epoch": 1.829919857524488, + "grad_norm": 2.3822596073150635, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9034579992294312, + "num_tokens": 558405277.0, + "step": 14385 + }, + { + "epoch": 1.8300470678030785, + "grad_norm": 2.5915462970733643, + "learning_rate": 1e-06, + "loss": 0.3146, + "mean_token_accuracy": 0.8892313241958618, + "num_tokens": 558450749.0, + "step": 14386 + }, + { + "epoch": 1.830174278081669, + "grad_norm": 2.7442080974578857, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.889799952507019, + "num_tokens": 558486067.0, + "step": 14387 + }, + { + "epoch": 1.8303014883602595, + "grad_norm": 2.609283924102783, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8865389227867126, + "num_tokens": 558525136.0, + "step": 14388 + }, + { + "epoch": 1.83042869863885, + "grad_norm": 2.427581787109375, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.8917250633239746, + "num_tokens": 558567512.0, + "step": 14389 + }, + { + "epoch": 1.8305559089174406, + "grad_norm": 2.683563232421875, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8888233304023743, + "num_tokens": 558604487.0, + "step": 14390 + }, + { + "epoch": 1.8306831191960309, + "grad_norm": 2.6999895572662354, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8961979746818542, + "num_tokens": 558639096.0, + "step": 14391 + }, + { + "epoch": 1.8308103294746214, + "grad_norm": 2.6543960571289062, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8770595192909241, + "num_tokens": 558678018.0, + "step": 14392 + }, + { + "epoch": 1.830937539753212, + "grad_norm": 2.7281506061553955, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8913814425468445, + "num_tokens": 558713427.0, + "step": 14393 + }, + { + "epoch": 1.8310647500318025, + "grad_norm": 2.5423030853271484, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.890911340713501, + "num_tokens": 558754513.0, + "step": 14394 + }, + { + "epoch": 1.831191960310393, + "grad_norm": 2.590956687927246, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9046579599380493, + "num_tokens": 558791274.0, + "step": 14395 + }, + { + "epoch": 1.8313191705889835, + "grad_norm": 2.502732038497925, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8944095373153687, + "num_tokens": 558829813.0, + "step": 14396 + }, + { + "epoch": 1.831446380867574, + "grad_norm": 2.678290367126465, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8989315032958984, + "num_tokens": 558865870.0, + "step": 14397 + }, + { + "epoch": 1.8315735911461646, + "grad_norm": 2.7035012245178223, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8884596824645996, + "num_tokens": 558901925.0, + "step": 14398 + }, + { + "epoch": 1.831700801424755, + "grad_norm": 2.4570326805114746, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8894587755203247, + "num_tokens": 558946505.0, + "step": 14399 + }, + { + "epoch": 1.8318280117033456, + "grad_norm": 2.489455461502075, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8972029685974121, + "num_tokens": 558987943.0, + "step": 14400 + }, + { + "epoch": 1.8319552219819362, + "grad_norm": 2.5443644523620605, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9025185108184814, + "num_tokens": 559024236.0, + "step": 14401 + }, + { + "epoch": 1.8320824322605267, + "grad_norm": 2.754610538482666, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8962404727935791, + "num_tokens": 559057865.0, + "step": 14402 + }, + { + "epoch": 1.8322096425391172, + "grad_norm": 2.512722969055176, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.887985110282898, + "num_tokens": 559099230.0, + "step": 14403 + }, + { + "epoch": 1.8323368528177078, + "grad_norm": 2.5281243324279785, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9043600559234619, + "num_tokens": 559135276.0, + "step": 14404 + }, + { + "epoch": 1.8324640630962983, + "grad_norm": 2.793409585952759, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8835509419441223, + "num_tokens": 559172010.0, + "step": 14405 + }, + { + "epoch": 1.8325912733748888, + "grad_norm": 2.546255588531494, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.886840283870697, + "num_tokens": 559211359.0, + "step": 14406 + }, + { + "epoch": 1.8327184836534793, + "grad_norm": 2.578735828399658, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8987522125244141, + "num_tokens": 559249671.0, + "step": 14407 + }, + { + "epoch": 1.8328456939320699, + "grad_norm": 2.4005565643310547, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9084200263023376, + "num_tokens": 559287762.0, + "step": 14408 + }, + { + "epoch": 1.8329729042106602, + "grad_norm": 2.86308217048645, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.880596399307251, + "num_tokens": 559322731.0, + "step": 14409 + }, + { + "epoch": 1.8331001144892507, + "grad_norm": 2.8328914642333984, + "learning_rate": 1e-06, + "loss": 0.3782, + "mean_token_accuracy": 0.8672405481338501, + "num_tokens": 559362023.0, + "step": 14410 + }, + { + "epoch": 1.8332273247678412, + "grad_norm": 2.6714189052581787, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.8791594505310059, + "num_tokens": 559403201.0, + "step": 14411 + }, + { + "epoch": 1.8333545350464318, + "grad_norm": 2.4879395961761475, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8951636552810669, + "num_tokens": 559445120.0, + "step": 14412 + }, + { + "epoch": 1.8334817453250223, + "grad_norm": 2.549496650695801, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8870075345039368, + "num_tokens": 559486268.0, + "step": 14413 + }, + { + "epoch": 1.8336089556036128, + "grad_norm": 2.4975621700286865, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8761250972747803, + "num_tokens": 559531684.0, + "step": 14414 + }, + { + "epoch": 1.8337361658822031, + "grad_norm": 2.856673002243042, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8891239166259766, + "num_tokens": 559562334.0, + "step": 14415 + }, + { + "epoch": 1.8338633761607936, + "grad_norm": 2.5075738430023193, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.891559362411499, + "num_tokens": 559604922.0, + "step": 14416 + }, + { + "epoch": 1.8339905864393842, + "grad_norm": 2.8031113147735596, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8912326097488403, + "num_tokens": 559640507.0, + "step": 14417 + }, + { + "epoch": 1.8341177967179747, + "grad_norm": 2.7524948120117188, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.891523003578186, + "num_tokens": 559675806.0, + "step": 14418 + }, + { + "epoch": 1.8342450069965652, + "grad_norm": 2.6056549549102783, + "learning_rate": 1e-06, + "loss": 0.3228, + "mean_token_accuracy": 0.8859854340553284, + "num_tokens": 559715777.0, + "step": 14419 + }, + { + "epoch": 1.8343722172751558, + "grad_norm": 2.61000919342041, + "learning_rate": 1e-06, + "loss": 0.3196, + "mean_token_accuracy": 0.8872048854827881, + "num_tokens": 559754981.0, + "step": 14420 + }, + { + "epoch": 1.8344994275537463, + "grad_norm": 2.6902072429656982, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.89989173412323, + "num_tokens": 559787056.0, + "step": 14421 + }, + { + "epoch": 1.8346266378323368, + "grad_norm": 2.6836965084075928, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8977867364883423, + "num_tokens": 559822048.0, + "step": 14422 + }, + { + "epoch": 1.8347538481109273, + "grad_norm": 2.335972547531128, + "learning_rate": 1e-06, + "loss": 0.3392, + "mean_token_accuracy": 0.8760195970535278, + "num_tokens": 559870064.0, + "step": 14423 + }, + { + "epoch": 1.8348810583895179, + "grad_norm": 2.239333391189575, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8963468074798584, + "num_tokens": 559916797.0, + "step": 14424 + }, + { + "epoch": 1.8350082686681084, + "grad_norm": 2.247356653213501, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9078746438026428, + "num_tokens": 559960745.0, + "step": 14425 + }, + { + "epoch": 1.835135478946699, + "grad_norm": 2.8953113555908203, + "learning_rate": 1e-06, + "loss": 0.3209, + "mean_token_accuracy": 0.8852207660675049, + "num_tokens": 559994100.0, + "step": 14426 + }, + { + "epoch": 1.8352626892252895, + "grad_norm": 2.3785276412963867, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.892377495765686, + "num_tokens": 560038547.0, + "step": 14427 + }, + { + "epoch": 1.83538989950388, + "grad_norm": 2.4159107208251953, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8957975506782532, + "num_tokens": 560080868.0, + "step": 14428 + }, + { + "epoch": 1.8355171097824705, + "grad_norm": 2.629018545150757, + "learning_rate": 1e-06, + "loss": 0.3027, + "mean_token_accuracy": 0.8910250663757324, + "num_tokens": 560119652.0, + "step": 14429 + }, + { + "epoch": 1.835644320061061, + "grad_norm": 2.4280219078063965, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8992735147476196, + "num_tokens": 560160520.0, + "step": 14430 + }, + { + "epoch": 1.8357715303396516, + "grad_norm": 2.614002227783203, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9019507169723511, + "num_tokens": 560193994.0, + "step": 14431 + }, + { + "epoch": 1.835898740618242, + "grad_norm": 2.6087682247161865, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9001322984695435, + "num_tokens": 560230087.0, + "step": 14432 + }, + { + "epoch": 1.8360259508968326, + "grad_norm": 2.738067150115967, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8954257369041443, + "num_tokens": 560262711.0, + "step": 14433 + }, + { + "epoch": 1.836153161175423, + "grad_norm": 2.603501081466675, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8880893588066101, + "num_tokens": 560301592.0, + "step": 14434 + }, + { + "epoch": 1.8362803714540135, + "grad_norm": 2.669750928878784, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8880139589309692, + "num_tokens": 560340181.0, + "step": 14435 + }, + { + "epoch": 1.836407581732604, + "grad_norm": 2.7897346019744873, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8981965780258179, + "num_tokens": 560376607.0, + "step": 14436 + }, + { + "epoch": 1.8365347920111945, + "grad_norm": 2.6706955432891846, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.895378589630127, + "num_tokens": 560411011.0, + "step": 14437 + }, + { + "epoch": 1.836662002289785, + "grad_norm": 2.909982681274414, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8835654258728027, + "num_tokens": 560446689.0, + "step": 14438 + }, + { + "epoch": 1.8367892125683756, + "grad_norm": 2.6524291038513184, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8936465978622437, + "num_tokens": 560483040.0, + "step": 14439 + }, + { + "epoch": 1.8369164228469659, + "grad_norm": 2.832099437713623, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8825091123580933, + "num_tokens": 560519289.0, + "step": 14440 + }, + { + "epoch": 1.8370436331255564, + "grad_norm": 2.4577293395996094, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.898213803768158, + "num_tokens": 560562323.0, + "step": 14441 + }, + { + "epoch": 1.837170843404147, + "grad_norm": 2.898301839828491, + "learning_rate": 1e-06, + "loss": 0.3739, + "mean_token_accuracy": 0.871408224105835, + "num_tokens": 560602946.0, + "step": 14442 + }, + { + "epoch": 1.8372980536827375, + "grad_norm": 2.823211669921875, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8910437822341919, + "num_tokens": 560634572.0, + "step": 14443 + }, + { + "epoch": 1.837425263961328, + "grad_norm": 2.618133783340454, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8931121826171875, + "num_tokens": 560674813.0, + "step": 14444 + }, + { + "epoch": 1.8375524742399185, + "grad_norm": 2.516759157180786, + "learning_rate": 1e-06, + "loss": 0.3222, + "mean_token_accuracy": 0.8835986852645874, + "num_tokens": 560721001.0, + "step": 14445 + }, + { + "epoch": 1.837679684518509, + "grad_norm": 2.9223413467407227, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8852429986000061, + "num_tokens": 560754920.0, + "step": 14446 + }, + { + "epoch": 1.8378068947970996, + "grad_norm": 2.5797321796417236, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8903927803039551, + "num_tokens": 560793673.0, + "step": 14447 + }, + { + "epoch": 1.83793410507569, + "grad_norm": 2.8114593029022217, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8913366794586182, + "num_tokens": 560825967.0, + "step": 14448 + }, + { + "epoch": 1.8380613153542806, + "grad_norm": 2.662752866744995, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8888146877288818, + "num_tokens": 560862564.0, + "step": 14449 + }, + { + "epoch": 1.8381885256328712, + "grad_norm": 2.4834556579589844, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.8994187116622925, + "num_tokens": 560902539.0, + "step": 14450 + }, + { + "epoch": 1.8383157359114617, + "grad_norm": 2.5920515060424805, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8912322521209717, + "num_tokens": 560940364.0, + "step": 14451 + }, + { + "epoch": 1.8384429461900522, + "grad_norm": 2.5076565742492676, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.899761974811554, + "num_tokens": 560977568.0, + "step": 14452 + }, + { + "epoch": 1.8385701564686427, + "grad_norm": 2.5051112174987793, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8977991938591003, + "num_tokens": 561013950.0, + "step": 14453 + }, + { + "epoch": 1.8386973667472333, + "grad_norm": 2.445319414138794, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8812690377235413, + "num_tokens": 561062472.0, + "step": 14454 + }, + { + "epoch": 1.8388245770258238, + "grad_norm": 2.5269525051116943, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8886662721633911, + "num_tokens": 561103511.0, + "step": 14455 + }, + { + "epoch": 1.8389517873044143, + "grad_norm": 2.8867294788360596, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8764747381210327, + "num_tokens": 561138458.0, + "step": 14456 + }, + { + "epoch": 1.8390789975830049, + "grad_norm": 2.722404956817627, + "learning_rate": 1e-06, + "loss": 0.3294, + "mean_token_accuracy": 0.882897138595581, + "num_tokens": 561179870.0, + "step": 14457 + }, + { + "epoch": 1.8392062078615952, + "grad_norm": 2.4446849822998047, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.89414381980896, + "num_tokens": 561219370.0, + "step": 14458 + }, + { + "epoch": 1.8393334181401857, + "grad_norm": 2.8621599674224854, + "learning_rate": 1e-06, + "loss": 0.3255, + "mean_token_accuracy": 0.882458508014679, + "num_tokens": 561250862.0, + "step": 14459 + }, + { + "epoch": 1.8394606284187762, + "grad_norm": 2.5722248554229736, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8884930610656738, + "num_tokens": 561289191.0, + "step": 14460 + }, + { + "epoch": 1.8395878386973668, + "grad_norm": 2.2819297313690186, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9004703760147095, + "num_tokens": 561332603.0, + "step": 14461 + }, + { + "epoch": 1.8397150489759573, + "grad_norm": 2.860128402709961, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8804978728294373, + "num_tokens": 561368997.0, + "step": 14462 + }, + { + "epoch": 1.8398422592545478, + "grad_norm": 2.447171688079834, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8942389488220215, + "num_tokens": 561408208.0, + "step": 14463 + }, + { + "epoch": 1.8399694695331381, + "grad_norm": 2.323721170425415, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9015742540359497, + "num_tokens": 561450678.0, + "step": 14464 + }, + { + "epoch": 1.8400966798117286, + "grad_norm": 2.793144941329956, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.891968846321106, + "num_tokens": 561485971.0, + "step": 14465 + }, + { + "epoch": 1.8402238900903192, + "grad_norm": 2.4780054092407227, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.8983603715896606, + "num_tokens": 561524942.0, + "step": 14466 + }, + { + "epoch": 1.8403511003689097, + "grad_norm": 2.7832987308502197, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8946503400802612, + "num_tokens": 561559530.0, + "step": 14467 + }, + { + "epoch": 1.8404783106475002, + "grad_norm": 2.737732410430908, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8828630447387695, + "num_tokens": 561598153.0, + "step": 14468 + }, + { + "epoch": 1.8406055209260908, + "grad_norm": 2.5858166217803955, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8889800310134888, + "num_tokens": 561639030.0, + "step": 14469 + }, + { + "epoch": 1.8407327312046813, + "grad_norm": 2.5506021976470947, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8890982270240784, + "num_tokens": 561680727.0, + "step": 14470 + }, + { + "epoch": 1.8408599414832718, + "grad_norm": 2.6529529094696045, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8936569690704346, + "num_tokens": 561717704.0, + "step": 14471 + }, + { + "epoch": 1.8409871517618623, + "grad_norm": 2.9296858310699463, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.882272481918335, + "num_tokens": 561753614.0, + "step": 14472 + }, + { + "epoch": 1.8411143620404529, + "grad_norm": 2.5725982189178467, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8942787647247314, + "num_tokens": 561792608.0, + "step": 14473 + }, + { + "epoch": 1.8412415723190434, + "grad_norm": 2.5592331886291504, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8870083689689636, + "num_tokens": 561832248.0, + "step": 14474 + }, + { + "epoch": 1.841368782597634, + "grad_norm": 2.6714301109313965, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.895102322101593, + "num_tokens": 561866182.0, + "step": 14475 + }, + { + "epoch": 1.8414959928762245, + "grad_norm": 2.7363436222076416, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8794815540313721, + "num_tokens": 561906094.0, + "step": 14476 + }, + { + "epoch": 1.841623203154815, + "grad_norm": 2.669776201248169, + "learning_rate": 1e-06, + "loss": 0.35, + "mean_token_accuracy": 0.8752003312110901, + "num_tokens": 561949983.0, + "step": 14477 + }, + { + "epoch": 1.8417504134334055, + "grad_norm": 2.8770713806152344, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8781948685646057, + "num_tokens": 561988038.0, + "step": 14478 + }, + { + "epoch": 1.841877623711996, + "grad_norm": 2.655632734298706, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8855549097061157, + "num_tokens": 562024642.0, + "step": 14479 + }, + { + "epoch": 1.8420048339905866, + "grad_norm": 2.5500526428222656, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9049335718154907, + "num_tokens": 562062447.0, + "step": 14480 + }, + { + "epoch": 1.842132044269177, + "grad_norm": 2.49979305267334, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8920960426330566, + "num_tokens": 562103985.0, + "step": 14481 + }, + { + "epoch": 1.8422592545477676, + "grad_norm": 2.6704442501068115, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8917179107666016, + "num_tokens": 562144673.0, + "step": 14482 + }, + { + "epoch": 1.842386464826358, + "grad_norm": 2.346979856491089, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9020761847496033, + "num_tokens": 562184176.0, + "step": 14483 + }, + { + "epoch": 1.8425136751049485, + "grad_norm": 2.4947125911712646, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8883049488067627, + "num_tokens": 562224946.0, + "step": 14484 + }, + { + "epoch": 1.842640885383539, + "grad_norm": 2.6412556171417236, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8807846307754517, + "num_tokens": 562263302.0, + "step": 14485 + }, + { + "epoch": 1.8427680956621295, + "grad_norm": 2.6652231216430664, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.890892505645752, + "num_tokens": 562299484.0, + "step": 14486 + }, + { + "epoch": 1.84289530594072, + "grad_norm": 2.6292412281036377, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8883394598960876, + "num_tokens": 562337476.0, + "step": 14487 + }, + { + "epoch": 1.8430225162193106, + "grad_norm": 2.403017520904541, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9021861553192139, + "num_tokens": 562380166.0, + "step": 14488 + }, + { + "epoch": 1.8431497264979009, + "grad_norm": 2.458951711654663, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8880124688148499, + "num_tokens": 562423917.0, + "step": 14489 + }, + { + "epoch": 1.8432769367764914, + "grad_norm": 2.708510637283325, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8791719675064087, + "num_tokens": 562461642.0, + "step": 14490 + }, + { + "epoch": 1.843404147055082, + "grad_norm": 2.5279455184936523, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8930466175079346, + "num_tokens": 562503870.0, + "step": 14491 + }, + { + "epoch": 1.8435313573336725, + "grad_norm": 2.465395212173462, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8915450572967529, + "num_tokens": 562546051.0, + "step": 14492 + }, + { + "epoch": 1.843658567612263, + "grad_norm": 2.5253286361694336, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8877471089363098, + "num_tokens": 562587342.0, + "step": 14493 + }, + { + "epoch": 1.8437857778908535, + "grad_norm": 2.5976099967956543, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8828878998756409, + "num_tokens": 562627198.0, + "step": 14494 + }, + { + "epoch": 1.843912988169444, + "grad_norm": 2.6411023139953613, + "learning_rate": 1e-06, + "loss": 0.3314, + "mean_token_accuracy": 0.8787177801132202, + "num_tokens": 562664743.0, + "step": 14495 + }, + { + "epoch": 1.8440401984480346, + "grad_norm": 2.55009388923645, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8859054446220398, + "num_tokens": 562708382.0, + "step": 14496 + }, + { + "epoch": 1.844167408726625, + "grad_norm": 2.4515411853790283, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9085390567779541, + "num_tokens": 562743878.0, + "step": 14497 + }, + { + "epoch": 1.8442946190052156, + "grad_norm": 2.722134828567505, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8825168609619141, + "num_tokens": 562781570.0, + "step": 14498 + }, + { + "epoch": 1.8444218292838062, + "grad_norm": 2.4759483337402344, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9083877801895142, + "num_tokens": 562819915.0, + "step": 14499 + }, + { + "epoch": 1.8445490395623967, + "grad_norm": 3.1066408157348633, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8762900233268738, + "num_tokens": 562849270.0, + "step": 14500 + }, + { + "epoch": 1.8446762498409872, + "grad_norm": 2.5148870944976807, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8898898959159851, + "num_tokens": 562893655.0, + "step": 14501 + }, + { + "epoch": 1.8448034601195777, + "grad_norm": 2.895124673843384, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8783278465270996, + "num_tokens": 562929440.0, + "step": 14502 + }, + { + "epoch": 1.8449306703981683, + "grad_norm": 2.3572165966033936, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9062870144844055, + "num_tokens": 562967925.0, + "step": 14503 + }, + { + "epoch": 1.8450578806767588, + "grad_norm": 2.7423346042633057, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8920188546180725, + "num_tokens": 563001691.0, + "step": 14504 + }, + { + "epoch": 1.8451850909553493, + "grad_norm": 2.467292070388794, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8869627118110657, + "num_tokens": 563045128.0, + "step": 14505 + }, + { + "epoch": 1.8453123012339399, + "grad_norm": 2.4498822689056396, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8892002701759338, + "num_tokens": 563090787.0, + "step": 14506 + }, + { + "epoch": 1.8454395115125302, + "grad_norm": 2.3714895248413086, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8949729204177856, + "num_tokens": 563132265.0, + "step": 14507 + }, + { + "epoch": 1.8455667217911207, + "grad_norm": 2.4983673095703125, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8955860137939453, + "num_tokens": 563173757.0, + "step": 14508 + }, + { + "epoch": 1.8456939320697112, + "grad_norm": 2.587108612060547, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8893950581550598, + "num_tokens": 563214267.0, + "step": 14509 + }, + { + "epoch": 1.8458211423483017, + "grad_norm": 2.506072521209717, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8800817131996155, + "num_tokens": 563258662.0, + "step": 14510 + }, + { + "epoch": 1.8459483526268923, + "grad_norm": 2.516934871673584, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8927422761917114, + "num_tokens": 563299590.0, + "step": 14511 + }, + { + "epoch": 1.8460755629054828, + "grad_norm": 2.9197092056274414, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8901169300079346, + "num_tokens": 563331174.0, + "step": 14512 + }, + { + "epoch": 1.846202773184073, + "grad_norm": 2.323847770690918, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9100962281227112, + "num_tokens": 563370532.0, + "step": 14513 + }, + { + "epoch": 1.8463299834626636, + "grad_norm": 2.6840548515319824, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8852927088737488, + "num_tokens": 563412103.0, + "step": 14514 + }, + { + "epoch": 1.8464571937412542, + "grad_norm": 2.6844794750213623, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8956734538078308, + "num_tokens": 563447135.0, + "step": 14515 + }, + { + "epoch": 1.8465844040198447, + "grad_norm": 2.8377633094787598, + "learning_rate": 1e-06, + "loss": 0.3286, + "mean_token_accuracy": 0.88469398021698, + "num_tokens": 563485527.0, + "step": 14516 + }, + { + "epoch": 1.8467116142984352, + "grad_norm": 2.5782907009124756, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8902238607406616, + "num_tokens": 563524231.0, + "step": 14517 + }, + { + "epoch": 1.8468388245770258, + "grad_norm": 2.7103772163391113, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.8770485520362854, + "num_tokens": 563568149.0, + "step": 14518 + }, + { + "epoch": 1.8469660348556163, + "grad_norm": 2.4381000995635986, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8842796087265015, + "num_tokens": 563611408.0, + "step": 14519 + }, + { + "epoch": 1.8470932451342068, + "grad_norm": 2.7727668285369873, + "learning_rate": 1e-06, + "loss": 0.3513, + "mean_token_accuracy": 0.873478889465332, + "num_tokens": 563649485.0, + "step": 14520 + }, + { + "epoch": 1.8472204554127973, + "grad_norm": 2.6412301063537598, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8883198499679565, + "num_tokens": 563687016.0, + "step": 14521 + }, + { + "epoch": 1.8473476656913879, + "grad_norm": 2.87104868888855, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8898249268531799, + "num_tokens": 563719244.0, + "step": 14522 + }, + { + "epoch": 1.8474748759699784, + "grad_norm": 2.6291189193725586, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8792792558670044, + "num_tokens": 563762452.0, + "step": 14523 + }, + { + "epoch": 1.847602086248569, + "grad_norm": 2.4664485454559326, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8941336870193481, + "num_tokens": 563803310.0, + "step": 14524 + }, + { + "epoch": 1.8477292965271594, + "grad_norm": 2.509540319442749, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8906875848770142, + "num_tokens": 563845559.0, + "step": 14525 + }, + { + "epoch": 1.84785650680575, + "grad_norm": 2.572268009185791, + "learning_rate": 1e-06, + "loss": 0.3235, + "mean_token_accuracy": 0.8846879005432129, + "num_tokens": 563886630.0, + "step": 14526 + }, + { + "epoch": 1.8479837170843405, + "grad_norm": 2.900331974029541, + "learning_rate": 1e-06, + "loss": 0.3389, + "mean_token_accuracy": 0.8772404193878174, + "num_tokens": 563923013.0, + "step": 14527 + }, + { + "epoch": 1.848110927362931, + "grad_norm": 2.6059768199920654, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8926116228103638, + "num_tokens": 563958608.0, + "step": 14528 + }, + { + "epoch": 1.8482381376415216, + "grad_norm": 2.45975923538208, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8919785022735596, + "num_tokens": 564000475.0, + "step": 14529 + }, + { + "epoch": 1.848365347920112, + "grad_norm": 2.6205005645751953, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8947017192840576, + "num_tokens": 564037254.0, + "step": 14530 + }, + { + "epoch": 1.8484925581987026, + "grad_norm": 2.8293731212615967, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8991184830665588, + "num_tokens": 564069072.0, + "step": 14531 + }, + { + "epoch": 1.848619768477293, + "grad_norm": 2.5708000659942627, + "learning_rate": 1e-06, + "loss": 0.3419, + "mean_token_accuracy": 0.8779553174972534, + "num_tokens": 564114127.0, + "step": 14532 + }, + { + "epoch": 1.8487469787558835, + "grad_norm": 2.6195929050445557, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8885238170623779, + "num_tokens": 564153199.0, + "step": 14533 + }, + { + "epoch": 1.848874189034474, + "grad_norm": 2.5867884159088135, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8879254460334778, + "num_tokens": 564195407.0, + "step": 14534 + }, + { + "epoch": 1.8490013993130645, + "grad_norm": 2.7532622814178467, + "learning_rate": 1e-06, + "loss": 0.3614, + "mean_token_accuracy": 0.8703163862228394, + "num_tokens": 564234829.0, + "step": 14535 + }, + { + "epoch": 1.849128609591655, + "grad_norm": 2.4319229125976562, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8966586589813232, + "num_tokens": 564272260.0, + "step": 14536 + }, + { + "epoch": 1.8492558198702456, + "grad_norm": 2.502338171005249, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.8994489312171936, + "num_tokens": 564311555.0, + "step": 14537 + }, + { + "epoch": 1.8493830301488359, + "grad_norm": 2.5809483528137207, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8930407762527466, + "num_tokens": 564348422.0, + "step": 14538 + }, + { + "epoch": 1.8495102404274264, + "grad_norm": 2.5425374507904053, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8867599368095398, + "num_tokens": 564389173.0, + "step": 14539 + }, + { + "epoch": 1.849637450706017, + "grad_norm": 2.4932477474212646, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8808677196502686, + "num_tokens": 564434042.0, + "step": 14540 + }, + { + "epoch": 1.8497646609846075, + "grad_norm": 2.8558056354522705, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.893404483795166, + "num_tokens": 564466371.0, + "step": 14541 + }, + { + "epoch": 1.849891871263198, + "grad_norm": 2.526595115661621, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8850305080413818, + "num_tokens": 564510164.0, + "step": 14542 + }, + { + "epoch": 1.8500190815417885, + "grad_norm": 2.3830647468566895, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8951528072357178, + "num_tokens": 564554271.0, + "step": 14543 + }, + { + "epoch": 1.850146291820379, + "grad_norm": 2.4975039958953857, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8925600051879883, + "num_tokens": 564596347.0, + "step": 14544 + }, + { + "epoch": 1.8502735020989696, + "grad_norm": 2.7956278324127197, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8867538571357727, + "num_tokens": 564630712.0, + "step": 14545 + }, + { + "epoch": 1.85040071237756, + "grad_norm": 2.9089667797088623, + "learning_rate": 1e-06, + "loss": 0.3519, + "mean_token_accuracy": 0.8744213581085205, + "num_tokens": 564668619.0, + "step": 14546 + }, + { + "epoch": 1.8505279226561506, + "grad_norm": 2.661935806274414, + "learning_rate": 1e-06, + "loss": 0.3353, + "mean_token_accuracy": 0.8796098232269287, + "num_tokens": 564709647.0, + "step": 14547 + }, + { + "epoch": 1.8506551329347412, + "grad_norm": 2.5373470783233643, + "learning_rate": 1e-06, + "loss": 0.3252, + "mean_token_accuracy": 0.8850482106208801, + "num_tokens": 564752632.0, + "step": 14548 + }, + { + "epoch": 1.8507823432133317, + "grad_norm": 2.551081418991089, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8786236047744751, + "num_tokens": 564796805.0, + "step": 14549 + }, + { + "epoch": 1.8509095534919222, + "grad_norm": 2.4381823539733887, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8976858854293823, + "num_tokens": 564838822.0, + "step": 14550 + }, + { + "epoch": 1.8510367637705127, + "grad_norm": 2.687366485595703, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8913229703903198, + "num_tokens": 564874330.0, + "step": 14551 + }, + { + "epoch": 1.8511639740491033, + "grad_norm": 2.7891995906829834, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8920710682868958, + "num_tokens": 564907098.0, + "step": 14552 + }, + { + "epoch": 1.8512911843276938, + "grad_norm": 2.7042479515075684, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8887107372283936, + "num_tokens": 564943958.0, + "step": 14553 + }, + { + "epoch": 1.8514183946062843, + "grad_norm": 2.3843307495117188, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9041956067085266, + "num_tokens": 564985333.0, + "step": 14554 + }, + { + "epoch": 1.8515456048848749, + "grad_norm": 2.567314624786377, + "learning_rate": 1e-06, + "loss": 0.2986, + "mean_token_accuracy": 0.8944990038871765, + "num_tokens": 565021842.0, + "step": 14555 + }, + { + "epoch": 1.8516728151634652, + "grad_norm": 2.3959624767303467, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8959821462631226, + "num_tokens": 565067062.0, + "step": 14556 + }, + { + "epoch": 1.8518000254420557, + "grad_norm": 2.5592336654663086, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.9014761447906494, + "num_tokens": 565102397.0, + "step": 14557 + }, + { + "epoch": 1.8519272357206462, + "grad_norm": 2.6894776821136475, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8880796432495117, + "num_tokens": 565135731.0, + "step": 14558 + }, + { + "epoch": 1.8520544459992367, + "grad_norm": 2.505858898162842, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.886407732963562, + "num_tokens": 565176769.0, + "step": 14559 + }, + { + "epoch": 1.8521816562778273, + "grad_norm": 2.5826408863067627, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8842623233795166, + "num_tokens": 565215516.0, + "step": 14560 + }, + { + "epoch": 1.8523088665564178, + "grad_norm": 2.5990195274353027, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.885174572467804, + "num_tokens": 565255015.0, + "step": 14561 + }, + { + "epoch": 1.852436076835008, + "grad_norm": 2.506281614303589, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8930671215057373, + "num_tokens": 565294028.0, + "step": 14562 + }, + { + "epoch": 1.8525632871135986, + "grad_norm": 2.730494976043701, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8881049156188965, + "num_tokens": 565332104.0, + "step": 14563 + }, + { + "epoch": 1.8526904973921892, + "grad_norm": 2.707345485687256, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.9002615213394165, + "num_tokens": 565367225.0, + "step": 14564 + }, + { + "epoch": 1.8528177076707797, + "grad_norm": 2.523488998413086, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8950487971305847, + "num_tokens": 565405826.0, + "step": 14565 + }, + { + "epoch": 1.8529449179493702, + "grad_norm": 2.758906364440918, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8755943775177002, + "num_tokens": 565448495.0, + "step": 14566 + }, + { + "epoch": 1.8530721282279607, + "grad_norm": 2.4806785583496094, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.8834404945373535, + "num_tokens": 565493096.0, + "step": 14567 + }, + { + "epoch": 1.8531993385065513, + "grad_norm": 2.5014638900756836, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.901847779750824, + "num_tokens": 565530826.0, + "step": 14568 + }, + { + "epoch": 1.8533265487851418, + "grad_norm": 2.4442670345306396, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8910936713218689, + "num_tokens": 565573244.0, + "step": 14569 + }, + { + "epoch": 1.8534537590637323, + "grad_norm": 2.403069257736206, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.8994684815406799, + "num_tokens": 565616898.0, + "step": 14570 + }, + { + "epoch": 1.8535809693423229, + "grad_norm": 2.600217342376709, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8981341123580933, + "num_tokens": 565652439.0, + "step": 14571 + }, + { + "epoch": 1.8537081796209134, + "grad_norm": 2.582799196243286, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8961952924728394, + "num_tokens": 565690211.0, + "step": 14572 + }, + { + "epoch": 1.853835389899504, + "grad_norm": 2.310068130493164, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9046274423599243, + "num_tokens": 565731793.0, + "step": 14573 + }, + { + "epoch": 1.8539626001780944, + "grad_norm": 2.7207138538360596, + "learning_rate": 1e-06, + "loss": 0.3317, + "mean_token_accuracy": 0.8804662823677063, + "num_tokens": 565770044.0, + "step": 14574 + }, + { + "epoch": 1.854089810456685, + "grad_norm": 2.578758716583252, + "learning_rate": 1e-06, + "loss": 0.3306, + "mean_token_accuracy": 0.8809618353843689, + "num_tokens": 565811036.0, + "step": 14575 + }, + { + "epoch": 1.8542170207352755, + "grad_norm": 2.6391751766204834, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8959108591079712, + "num_tokens": 565847339.0, + "step": 14576 + }, + { + "epoch": 1.854344231013866, + "grad_norm": 2.614140510559082, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.8872496485710144, + "num_tokens": 565888001.0, + "step": 14577 + }, + { + "epoch": 1.8544714412924566, + "grad_norm": 2.6729962825775146, + "learning_rate": 1e-06, + "loss": 0.3042, + "mean_token_accuracy": 0.8928569555282593, + "num_tokens": 565925527.0, + "step": 14578 + }, + { + "epoch": 1.854598651571047, + "grad_norm": 2.747833251953125, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8927946090698242, + "num_tokens": 565960716.0, + "step": 14579 + }, + { + "epoch": 1.8547258618496376, + "grad_norm": 2.716801404953003, + "learning_rate": 1e-06, + "loss": 0.3455, + "mean_token_accuracy": 0.8771070241928101, + "num_tokens": 566000571.0, + "step": 14580 + }, + { + "epoch": 1.854853072128228, + "grad_norm": 2.731715202331543, + "learning_rate": 1e-06, + "loss": 0.3349, + "mean_token_accuracy": 0.8814720511436462, + "num_tokens": 566043755.0, + "step": 14581 + }, + { + "epoch": 1.8549802824068184, + "grad_norm": 2.602037191390991, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8938226103782654, + "num_tokens": 566082770.0, + "step": 14582 + }, + { + "epoch": 1.855107492685409, + "grad_norm": 2.569918632507324, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9003269076347351, + "num_tokens": 566120671.0, + "step": 14583 + }, + { + "epoch": 1.8552347029639995, + "grad_norm": 2.6246848106384277, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8860025405883789, + "num_tokens": 566164065.0, + "step": 14584 + }, + { + "epoch": 1.85536191324259, + "grad_norm": 2.5533180236816406, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8947098255157471, + "num_tokens": 566204029.0, + "step": 14585 + }, + { + "epoch": 1.8554891235211806, + "grad_norm": 2.5258193016052246, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8896570205688477, + "num_tokens": 566245225.0, + "step": 14586 + }, + { + "epoch": 1.8556163337997709, + "grad_norm": 2.465003252029419, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.8999674916267395, + "num_tokens": 566283165.0, + "step": 14587 + }, + { + "epoch": 1.8557435440783614, + "grad_norm": 2.5619852542877197, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8936935663223267, + "num_tokens": 566322903.0, + "step": 14588 + }, + { + "epoch": 1.855870754356952, + "grad_norm": 2.688866376876831, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.876178503036499, + "num_tokens": 566365822.0, + "step": 14589 + }, + { + "epoch": 1.8559979646355425, + "grad_norm": 2.434251546859741, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9113609194755554, + "num_tokens": 566403734.0, + "step": 14590 + }, + { + "epoch": 1.856125174914133, + "grad_norm": 2.851471424102783, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8930704593658447, + "num_tokens": 566435980.0, + "step": 14591 + }, + { + "epoch": 1.8562523851927235, + "grad_norm": 2.6883418560028076, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9021912813186646, + "num_tokens": 566472156.0, + "step": 14592 + }, + { + "epoch": 1.856379595471314, + "grad_norm": 2.5902161598205566, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.899167537689209, + "num_tokens": 566509970.0, + "step": 14593 + }, + { + "epoch": 1.8565068057499046, + "grad_norm": 2.578984022140503, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8900955319404602, + "num_tokens": 566549605.0, + "step": 14594 + }, + { + "epoch": 1.856634016028495, + "grad_norm": 2.702244520187378, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8927757143974304, + "num_tokens": 566583785.0, + "step": 14595 + }, + { + "epoch": 1.8567612263070856, + "grad_norm": 2.6358814239501953, + "learning_rate": 1e-06, + "loss": 0.3538, + "mean_token_accuracy": 0.8721057772636414, + "num_tokens": 566627281.0, + "step": 14596 + }, + { + "epoch": 1.8568884365856761, + "grad_norm": 2.6821277141571045, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8866149187088013, + "num_tokens": 566664948.0, + "step": 14597 + }, + { + "epoch": 1.8570156468642667, + "grad_norm": 2.621873378753662, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9013664722442627, + "num_tokens": 566701377.0, + "step": 14598 + }, + { + "epoch": 1.8571428571428572, + "grad_norm": 2.4050328731536865, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8929744958877563, + "num_tokens": 566749157.0, + "step": 14599 + }, + { + "epoch": 1.8572700674214477, + "grad_norm": 2.410113573074341, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8997716903686523, + "num_tokens": 566791562.0, + "step": 14600 + }, + { + "epoch": 1.8573972777000383, + "grad_norm": 2.5159528255462646, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8826985359191895, + "num_tokens": 566832113.0, + "step": 14601 + }, + { + "epoch": 1.8575244879786288, + "grad_norm": 2.8554956912994385, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8956475257873535, + "num_tokens": 566871384.0, + "step": 14602 + }, + { + "epoch": 1.8576516982572193, + "grad_norm": 2.782008647918701, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.885877788066864, + "num_tokens": 566907967.0, + "step": 14603 + }, + { + "epoch": 1.8577789085358098, + "grad_norm": 2.6470704078674316, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8851704597473145, + "num_tokens": 566951063.0, + "step": 14604 + }, + { + "epoch": 1.8579061188144002, + "grad_norm": 2.3846936225891113, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.895999550819397, + "num_tokens": 566992237.0, + "step": 14605 + }, + { + "epoch": 1.8580333290929907, + "grad_norm": 2.562551975250244, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8955434560775757, + "num_tokens": 567031644.0, + "step": 14606 + }, + { + "epoch": 1.8581605393715812, + "grad_norm": 2.5568504333496094, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.897097110748291, + "num_tokens": 567071903.0, + "step": 14607 + }, + { + "epoch": 1.8582877496501717, + "grad_norm": 2.562093734741211, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9024641513824463, + "num_tokens": 567112107.0, + "step": 14608 + }, + { + "epoch": 1.8584149599287623, + "grad_norm": 2.437312602996826, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8919126987457275, + "num_tokens": 567153186.0, + "step": 14609 + }, + { + "epoch": 1.8585421702073528, + "grad_norm": 2.5097994804382324, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8909133076667786, + "num_tokens": 567197008.0, + "step": 14610 + }, + { + "epoch": 1.858669380485943, + "grad_norm": 2.64127254486084, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8968334794044495, + "num_tokens": 567234234.0, + "step": 14611 + }, + { + "epoch": 1.8587965907645336, + "grad_norm": 2.672009229660034, + "learning_rate": 1e-06, + "loss": 0.3304, + "mean_token_accuracy": 0.8816171884536743, + "num_tokens": 567275820.0, + "step": 14612 + }, + { + "epoch": 1.8589238010431242, + "grad_norm": 2.4823617935180664, + "learning_rate": 1e-06, + "loss": 0.3352, + "mean_token_accuracy": 0.8792672157287598, + "num_tokens": 567320457.0, + "step": 14613 + }, + { + "epoch": 1.8590510113217147, + "grad_norm": 2.6338298320770264, + "learning_rate": 1e-06, + "loss": 0.3118, + "mean_token_accuracy": 0.8882079124450684, + "num_tokens": 567359747.0, + "step": 14614 + }, + { + "epoch": 1.8591782216003052, + "grad_norm": 2.7790000438690186, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8773670196533203, + "num_tokens": 567399406.0, + "step": 14615 + }, + { + "epoch": 1.8593054318788957, + "grad_norm": 2.5763258934020996, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8950459957122803, + "num_tokens": 567440734.0, + "step": 14616 + }, + { + "epoch": 1.8594326421574863, + "grad_norm": 2.5612640380859375, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9037003517150879, + "num_tokens": 567477069.0, + "step": 14617 + }, + { + "epoch": 1.8595598524360768, + "grad_norm": 2.573803424835205, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8985539078712463, + "num_tokens": 567513299.0, + "step": 14618 + }, + { + "epoch": 1.8596870627146673, + "grad_norm": 2.3774352073669434, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9006052017211914, + "num_tokens": 567553091.0, + "step": 14619 + }, + { + "epoch": 1.8598142729932579, + "grad_norm": 2.699720621109009, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8876028060913086, + "num_tokens": 567592433.0, + "step": 14620 + }, + { + "epoch": 1.8599414832718484, + "grad_norm": 2.6992886066436768, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8827289342880249, + "num_tokens": 567631894.0, + "step": 14621 + }, + { + "epoch": 1.860068693550439, + "grad_norm": 2.7144737243652344, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.893215000629425, + "num_tokens": 567670244.0, + "step": 14622 + }, + { + "epoch": 1.8601959038290294, + "grad_norm": 2.497072696685791, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8885611295700073, + "num_tokens": 567713979.0, + "step": 14623 + }, + { + "epoch": 1.86032311410762, + "grad_norm": 2.6682209968566895, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9009661078453064, + "num_tokens": 567745944.0, + "step": 14624 + }, + { + "epoch": 1.8604503243862105, + "grad_norm": 2.4053895473480225, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.903887152671814, + "num_tokens": 567786868.0, + "step": 14625 + }, + { + "epoch": 1.860577534664801, + "grad_norm": 2.4247641563415527, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8977136611938477, + "num_tokens": 567828199.0, + "step": 14626 + }, + { + "epoch": 1.8607047449433916, + "grad_norm": 2.8752403259277344, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8920176029205322, + "num_tokens": 567862298.0, + "step": 14627 + }, + { + "epoch": 1.860831955221982, + "grad_norm": 2.5456089973449707, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9088135957717896, + "num_tokens": 567900473.0, + "step": 14628 + }, + { + "epoch": 1.8609591655005726, + "grad_norm": 2.9139418601989746, + "learning_rate": 1e-06, + "loss": 0.3582, + "mean_token_accuracy": 0.8723968863487244, + "num_tokens": 567937471.0, + "step": 14629 + }, + { + "epoch": 1.861086375779163, + "grad_norm": 2.627438545227051, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8900166153907776, + "num_tokens": 567979967.0, + "step": 14630 + }, + { + "epoch": 1.8612135860577534, + "grad_norm": 2.8777637481689453, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8729746341705322, + "num_tokens": 568023661.0, + "step": 14631 + }, + { + "epoch": 1.861340796336344, + "grad_norm": 2.5297224521636963, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.9012148380279541, + "num_tokens": 568064763.0, + "step": 14632 + }, + { + "epoch": 1.8614680066149345, + "grad_norm": 2.8149807453155518, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8849493265151978, + "num_tokens": 568101266.0, + "step": 14633 + }, + { + "epoch": 1.861595216893525, + "grad_norm": 2.4693145751953125, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.89296555519104, + "num_tokens": 568145867.0, + "step": 14634 + }, + { + "epoch": 1.8617224271721156, + "grad_norm": 2.7552874088287354, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8857345581054688, + "num_tokens": 568184855.0, + "step": 14635 + }, + { + "epoch": 1.8618496374507059, + "grad_norm": 2.7610034942626953, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8868722319602966, + "num_tokens": 568223959.0, + "step": 14636 + }, + { + "epoch": 1.8619768477292964, + "grad_norm": 2.8148245811462402, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8986705541610718, + "num_tokens": 568254989.0, + "step": 14637 + }, + { + "epoch": 1.862104058007887, + "grad_norm": 2.5318233966827393, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.896857738494873, + "num_tokens": 568293104.0, + "step": 14638 + }, + { + "epoch": 1.8622312682864774, + "grad_norm": 2.3603897094726562, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8862839341163635, + "num_tokens": 568336351.0, + "step": 14639 + }, + { + "epoch": 1.862358478565068, + "grad_norm": 2.7189929485321045, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8783396482467651, + "num_tokens": 568379214.0, + "step": 14640 + }, + { + "epoch": 1.8624856888436585, + "grad_norm": 2.7698311805725098, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8965142965316772, + "num_tokens": 568416438.0, + "step": 14641 + }, + { + "epoch": 1.862612899122249, + "grad_norm": 2.580498456954956, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8822457194328308, + "num_tokens": 568459176.0, + "step": 14642 + }, + { + "epoch": 1.8627401094008396, + "grad_norm": 2.5596814155578613, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8884729146957397, + "num_tokens": 568499883.0, + "step": 14643 + }, + { + "epoch": 1.86286731967943, + "grad_norm": 2.529489517211914, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8848018646240234, + "num_tokens": 568547679.0, + "step": 14644 + }, + { + "epoch": 1.8629945299580206, + "grad_norm": 2.4205946922302246, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8996858596801758, + "num_tokens": 568587210.0, + "step": 14645 + }, + { + "epoch": 1.8631217402366111, + "grad_norm": 2.4744856357574463, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8978200554847717, + "num_tokens": 568627277.0, + "step": 14646 + }, + { + "epoch": 1.8632489505152017, + "grad_norm": 2.4249417781829834, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8915914297103882, + "num_tokens": 568669898.0, + "step": 14647 + }, + { + "epoch": 1.8633761607937922, + "grad_norm": 2.668649196624756, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8863786458969116, + "num_tokens": 568710683.0, + "step": 14648 + }, + { + "epoch": 1.8635033710723827, + "grad_norm": 2.4838645458221436, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8970452547073364, + "num_tokens": 568750769.0, + "step": 14649 + }, + { + "epoch": 1.8636305813509733, + "grad_norm": 2.5418965816497803, + "learning_rate": 1e-06, + "loss": 0.3336, + "mean_token_accuracy": 0.8815569877624512, + "num_tokens": 568792296.0, + "step": 14650 + }, + { + "epoch": 1.8637577916295638, + "grad_norm": 2.6035938262939453, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9115062952041626, + "num_tokens": 568828895.0, + "step": 14651 + }, + { + "epoch": 1.8638850019081543, + "grad_norm": 2.5671401023864746, + "learning_rate": 1e-06, + "loss": 0.3277, + "mean_token_accuracy": 0.8837851285934448, + "num_tokens": 568874320.0, + "step": 14652 + }, + { + "epoch": 1.8640122121867448, + "grad_norm": 2.5654947757720947, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8950411081314087, + "num_tokens": 568911738.0, + "step": 14653 + }, + { + "epoch": 1.8641394224653351, + "grad_norm": 2.7834208011627197, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8892174363136292, + "num_tokens": 568947785.0, + "step": 14654 + }, + { + "epoch": 1.8642666327439257, + "grad_norm": 2.6667652130126953, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8962610960006714, + "num_tokens": 568984053.0, + "step": 14655 + }, + { + "epoch": 1.8643938430225162, + "grad_norm": 2.5282461643218994, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9000436663627625, + "num_tokens": 569020241.0, + "step": 14656 + }, + { + "epoch": 1.8645210533011067, + "grad_norm": 2.7769076824188232, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.878659188747406, + "num_tokens": 569057456.0, + "step": 14657 + }, + { + "epoch": 1.8646482635796973, + "grad_norm": 2.5611658096313477, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8862980604171753, + "num_tokens": 569101844.0, + "step": 14658 + }, + { + "epoch": 1.8647754738582878, + "grad_norm": 2.477768659591675, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8913111090660095, + "num_tokens": 569144815.0, + "step": 14659 + }, + { + "epoch": 1.864902684136878, + "grad_norm": 2.6551787853240967, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8953652381896973, + "num_tokens": 569184123.0, + "step": 14660 + }, + { + "epoch": 1.8650298944154686, + "grad_norm": 2.6173367500305176, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8918795585632324, + "num_tokens": 569222450.0, + "step": 14661 + }, + { + "epoch": 1.8651571046940592, + "grad_norm": 2.4482967853546143, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9033184051513672, + "num_tokens": 569262766.0, + "step": 14662 + }, + { + "epoch": 1.8652843149726497, + "grad_norm": 2.494701623916626, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8874090313911438, + "num_tokens": 569305934.0, + "step": 14663 + }, + { + "epoch": 1.8654115252512402, + "grad_norm": 2.4128506183624268, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.903032124042511, + "num_tokens": 569344962.0, + "step": 14664 + }, + { + "epoch": 1.8655387355298307, + "grad_norm": 2.6547276973724365, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8860209584236145, + "num_tokens": 569383644.0, + "step": 14665 + }, + { + "epoch": 1.8656659458084213, + "grad_norm": 2.4430460929870605, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8986213207244873, + "num_tokens": 569422479.0, + "step": 14666 + }, + { + "epoch": 1.8657931560870118, + "grad_norm": 2.5480494499206543, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8899034857749939, + "num_tokens": 569463952.0, + "step": 14667 + }, + { + "epoch": 1.8659203663656023, + "grad_norm": 2.334846258163452, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8905497193336487, + "num_tokens": 569512304.0, + "step": 14668 + }, + { + "epoch": 1.8660475766441929, + "grad_norm": 2.821969509124756, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8885612487792969, + "num_tokens": 569545566.0, + "step": 14669 + }, + { + "epoch": 1.8661747869227834, + "grad_norm": 2.4570634365081787, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9055330157279968, + "num_tokens": 569583031.0, + "step": 14670 + }, + { + "epoch": 1.866301997201374, + "grad_norm": 2.48093318939209, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8939985036849976, + "num_tokens": 569623740.0, + "step": 14671 + }, + { + "epoch": 1.8664292074799644, + "grad_norm": 2.6277081966400146, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8913804888725281, + "num_tokens": 569658422.0, + "step": 14672 + }, + { + "epoch": 1.866556417758555, + "grad_norm": 2.827145576477051, + "learning_rate": 1e-06, + "loss": 0.3327, + "mean_token_accuracy": 0.8811566829681396, + "num_tokens": 569693096.0, + "step": 14673 + }, + { + "epoch": 1.8666836280371455, + "grad_norm": 2.607377290725708, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8954144716262817, + "num_tokens": 569729149.0, + "step": 14674 + }, + { + "epoch": 1.866810838315736, + "grad_norm": 2.6745259761810303, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8919589519500732, + "num_tokens": 569763461.0, + "step": 14675 + }, + { + "epoch": 1.8669380485943265, + "grad_norm": 2.538029193878174, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9006602764129639, + "num_tokens": 569803201.0, + "step": 14676 + }, + { + "epoch": 1.867065258872917, + "grad_norm": 2.630662441253662, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8815230131149292, + "num_tokens": 569846153.0, + "step": 14677 + }, + { + "epoch": 1.8671924691515076, + "grad_norm": 2.647226572036743, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8894586563110352, + "num_tokens": 569885258.0, + "step": 14678 + }, + { + "epoch": 1.867319679430098, + "grad_norm": 2.4111955165863037, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8918741345405579, + "num_tokens": 569929482.0, + "step": 14679 + }, + { + "epoch": 1.8674468897086884, + "grad_norm": 2.4749257564544678, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8925360441207886, + "num_tokens": 569971428.0, + "step": 14680 + }, + { + "epoch": 1.867574099987279, + "grad_norm": 2.814002513885498, + "learning_rate": 1e-06, + "loss": 0.3627, + "mean_token_accuracy": 0.8699052333831787, + "num_tokens": 570010650.0, + "step": 14681 + }, + { + "epoch": 1.8677013102658695, + "grad_norm": 2.436253547668457, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9077038764953613, + "num_tokens": 570051034.0, + "step": 14682 + }, + { + "epoch": 1.86782852054446, + "grad_norm": 2.854046583175659, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8972798585891724, + "num_tokens": 570082619.0, + "step": 14683 + }, + { + "epoch": 1.8679557308230506, + "grad_norm": 2.700745105743408, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8883153796195984, + "num_tokens": 570119294.0, + "step": 14684 + }, + { + "epoch": 1.8680829411016409, + "grad_norm": 2.66475510597229, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8899496793746948, + "num_tokens": 570158194.0, + "step": 14685 + }, + { + "epoch": 1.8682101513802314, + "grad_norm": 2.884007692337036, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8864140510559082, + "num_tokens": 570199474.0, + "step": 14686 + }, + { + "epoch": 1.868337361658822, + "grad_norm": 2.6480557918548584, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8922890424728394, + "num_tokens": 570236283.0, + "step": 14687 + }, + { + "epoch": 1.8684645719374124, + "grad_norm": 2.7931127548217773, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8974164724349976, + "num_tokens": 570270219.0, + "step": 14688 + }, + { + "epoch": 1.868591782216003, + "grad_norm": 2.7472281455993652, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8943608999252319, + "num_tokens": 570302374.0, + "step": 14689 + }, + { + "epoch": 1.8687189924945935, + "grad_norm": 2.7227442264556885, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8850224018096924, + "num_tokens": 570338917.0, + "step": 14690 + }, + { + "epoch": 1.868846202773184, + "grad_norm": 2.5352365970611572, + "learning_rate": 1e-06, + "loss": 0.3427, + "mean_token_accuracy": 0.875887930393219, + "num_tokens": 570385180.0, + "step": 14691 + }, + { + "epoch": 1.8689734130517746, + "grad_norm": 2.696341037750244, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8902350664138794, + "num_tokens": 570419908.0, + "step": 14692 + }, + { + "epoch": 1.869100623330365, + "grad_norm": 2.449388027191162, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9002941250801086, + "num_tokens": 570462407.0, + "step": 14693 + }, + { + "epoch": 1.8692278336089556, + "grad_norm": 2.173673152923584, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9130071997642517, + "num_tokens": 570510437.0, + "step": 14694 + }, + { + "epoch": 1.8693550438875461, + "grad_norm": 2.42000675201416, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8936362862586975, + "num_tokens": 570550477.0, + "step": 14695 + }, + { + "epoch": 1.8694822541661367, + "grad_norm": 2.5043718814849854, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.8888853788375854, + "num_tokens": 570592841.0, + "step": 14696 + }, + { + "epoch": 1.8696094644447272, + "grad_norm": 2.627441883087158, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8920573592185974, + "num_tokens": 570627548.0, + "step": 14697 + }, + { + "epoch": 1.8697366747233177, + "grad_norm": 2.8625686168670654, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8977522850036621, + "num_tokens": 570656582.0, + "step": 14698 + }, + { + "epoch": 1.8698638850019083, + "grad_norm": 2.4253015518188477, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.892300009727478, + "num_tokens": 570698574.0, + "step": 14699 + }, + { + "epoch": 1.8699910952804988, + "grad_norm": 2.7229790687561035, + "learning_rate": 1e-06, + "loss": 0.3162, + "mean_token_accuracy": 0.8854179382324219, + "num_tokens": 570736063.0, + "step": 14700 + }, + { + "epoch": 1.8701183055590893, + "grad_norm": 2.7217190265655518, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8978201746940613, + "num_tokens": 570770252.0, + "step": 14701 + }, + { + "epoch": 1.8702455158376798, + "grad_norm": 2.5394721031188965, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8956310749053955, + "num_tokens": 570806608.0, + "step": 14702 + }, + { + "epoch": 1.8703727261162701, + "grad_norm": 2.7090134620666504, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.8989052772521973, + "num_tokens": 570838336.0, + "step": 14703 + }, + { + "epoch": 1.8704999363948607, + "grad_norm": 2.334378242492676, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8988499641418457, + "num_tokens": 570880703.0, + "step": 14704 + }, + { + "epoch": 1.8706271466734512, + "grad_norm": 2.6065969467163086, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.8985274434089661, + "num_tokens": 570917354.0, + "step": 14705 + }, + { + "epoch": 1.8707543569520417, + "grad_norm": 2.5752155780792236, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8880545496940613, + "num_tokens": 570959469.0, + "step": 14706 + }, + { + "epoch": 1.8708815672306323, + "grad_norm": 2.686594247817993, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8901975750923157, + "num_tokens": 570997378.0, + "step": 14707 + }, + { + "epoch": 1.8710087775092228, + "grad_norm": 2.7417240142822266, + "learning_rate": 1e-06, + "loss": 0.3542, + "mean_token_accuracy": 0.8718360662460327, + "num_tokens": 571037321.0, + "step": 14708 + }, + { + "epoch": 1.871135987787813, + "grad_norm": 2.5412988662719727, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8977699279785156, + "num_tokens": 571076129.0, + "step": 14709 + }, + { + "epoch": 1.8712631980664036, + "grad_norm": 2.5111169815063477, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8868522047996521, + "num_tokens": 571114469.0, + "step": 14710 + }, + { + "epoch": 1.8713904083449941, + "grad_norm": 2.3138365745544434, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.9004682302474976, + "num_tokens": 571157110.0, + "step": 14711 + }, + { + "epoch": 1.8715176186235847, + "grad_norm": 2.709840774536133, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8843233585357666, + "num_tokens": 571196815.0, + "step": 14712 + }, + { + "epoch": 1.8716448289021752, + "grad_norm": 2.9590542316436768, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8716006875038147, + "num_tokens": 571234446.0, + "step": 14713 + }, + { + "epoch": 1.8717720391807657, + "grad_norm": 2.3798837661743164, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8945299983024597, + "num_tokens": 571280313.0, + "step": 14714 + }, + { + "epoch": 1.8718992494593563, + "grad_norm": 2.6494650840759277, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8889092206954956, + "num_tokens": 571317226.0, + "step": 14715 + }, + { + "epoch": 1.8720264597379468, + "grad_norm": 2.6203505992889404, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8942923545837402, + "num_tokens": 571354422.0, + "step": 14716 + }, + { + "epoch": 1.8721536700165373, + "grad_norm": 2.4181067943573, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8912971615791321, + "num_tokens": 571398547.0, + "step": 14717 + }, + { + "epoch": 1.8722808802951278, + "grad_norm": 2.415909767150879, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9015781879425049, + "num_tokens": 571435279.0, + "step": 14718 + }, + { + "epoch": 1.8724080905737184, + "grad_norm": 2.397667169570923, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8914749622344971, + "num_tokens": 571480031.0, + "step": 14719 + }, + { + "epoch": 1.872535300852309, + "grad_norm": 2.5682287216186523, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.8954147100448608, + "num_tokens": 571517597.0, + "step": 14720 + }, + { + "epoch": 1.8726625111308994, + "grad_norm": 2.678027391433716, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8889312148094177, + "num_tokens": 571554939.0, + "step": 14721 + }, + { + "epoch": 1.87278972140949, + "grad_norm": 2.386141061782837, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8923219442367554, + "num_tokens": 571600118.0, + "step": 14722 + }, + { + "epoch": 1.8729169316880805, + "grad_norm": 2.47271466255188, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9070854187011719, + "num_tokens": 571637938.0, + "step": 14723 + }, + { + "epoch": 1.873044141966671, + "grad_norm": 2.5716946125030518, + "learning_rate": 1e-06, + "loss": 0.3109, + "mean_token_accuracy": 0.8879767656326294, + "num_tokens": 571678831.0, + "step": 14724 + }, + { + "epoch": 1.8731713522452615, + "grad_norm": 2.2174181938171387, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9051494002342224, + "num_tokens": 571722064.0, + "step": 14725 + }, + { + "epoch": 1.873298562523852, + "grad_norm": 2.6368038654327393, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.885683536529541, + "num_tokens": 571762935.0, + "step": 14726 + }, + { + "epoch": 1.8734257728024426, + "grad_norm": 2.6254656314849854, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8824089765548706, + "num_tokens": 571804232.0, + "step": 14727 + }, + { + "epoch": 1.873552983081033, + "grad_norm": 2.5424063205718994, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8941846489906311, + "num_tokens": 571842401.0, + "step": 14728 + }, + { + "epoch": 1.8736801933596234, + "grad_norm": 2.5795631408691406, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8922454714775085, + "num_tokens": 571883841.0, + "step": 14729 + }, + { + "epoch": 1.873807403638214, + "grad_norm": 2.610001564025879, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8957616686820984, + "num_tokens": 571918534.0, + "step": 14730 + }, + { + "epoch": 1.8739346139168045, + "grad_norm": 2.32753849029541, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8895739316940308, + "num_tokens": 571968230.0, + "step": 14731 + }, + { + "epoch": 1.874061824195395, + "grad_norm": 2.5186614990234375, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9059894680976868, + "num_tokens": 572006522.0, + "step": 14732 + }, + { + "epoch": 1.8741890344739855, + "grad_norm": 2.674617052078247, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8848569393157959, + "num_tokens": 572043308.0, + "step": 14733 + }, + { + "epoch": 1.8743162447525759, + "grad_norm": 2.4366607666015625, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.882544755935669, + "num_tokens": 572088521.0, + "step": 14734 + }, + { + "epoch": 1.8744434550311664, + "grad_norm": 2.6361358165740967, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8934735059738159, + "num_tokens": 572124223.0, + "step": 14735 + }, + { + "epoch": 1.874570665309757, + "grad_norm": 2.436433792114258, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9139116406440735, + "num_tokens": 572157766.0, + "step": 14736 + }, + { + "epoch": 1.8746978755883474, + "grad_norm": 2.4601800441741943, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8949102163314819, + "num_tokens": 572200370.0, + "step": 14737 + }, + { + "epoch": 1.874825085866938, + "grad_norm": 2.8936941623687744, + "learning_rate": 1e-06, + "loss": 0.3401, + "mean_token_accuracy": 0.8801435232162476, + "num_tokens": 572237089.0, + "step": 14738 + }, + { + "epoch": 1.8749522961455285, + "grad_norm": 2.49084734916687, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.896277904510498, + "num_tokens": 572277542.0, + "step": 14739 + }, + { + "epoch": 1.875079506424119, + "grad_norm": 2.8118882179260254, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8847314715385437, + "num_tokens": 572317140.0, + "step": 14740 + }, + { + "epoch": 1.8752067167027096, + "grad_norm": 2.7233352661132812, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9006584882736206, + "num_tokens": 572351078.0, + "step": 14741 + }, + { + "epoch": 1.8753339269813, + "grad_norm": 2.5379364490509033, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8956282138824463, + "num_tokens": 572390833.0, + "step": 14742 + }, + { + "epoch": 1.8754611372598906, + "grad_norm": 2.5071029663085938, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8969700932502747, + "num_tokens": 572434599.0, + "step": 14743 + }, + { + "epoch": 1.8755883475384811, + "grad_norm": 2.6379756927490234, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8846755027770996, + "num_tokens": 572476126.0, + "step": 14744 + }, + { + "epoch": 1.8757155578170717, + "grad_norm": 2.43599271774292, + "learning_rate": 1e-06, + "loss": 0.3253, + "mean_token_accuracy": 0.881468653678894, + "num_tokens": 572524559.0, + "step": 14745 + }, + { + "epoch": 1.8758427680956622, + "grad_norm": 2.549830675125122, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.8985851407051086, + "num_tokens": 572564169.0, + "step": 14746 + }, + { + "epoch": 1.8759699783742527, + "grad_norm": 2.6581523418426514, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8952316641807556, + "num_tokens": 572598081.0, + "step": 14747 + }, + { + "epoch": 1.8760971886528433, + "grad_norm": 2.860093593597412, + "learning_rate": 1e-06, + "loss": 0.3594, + "mean_token_accuracy": 0.8759315013885498, + "num_tokens": 572636208.0, + "step": 14748 + }, + { + "epoch": 1.8762243989314338, + "grad_norm": 3.1778147220611572, + "learning_rate": 1e-06, + "loss": 0.3649, + "mean_token_accuracy": 0.8750814199447632, + "num_tokens": 572670138.0, + "step": 14749 + }, + { + "epoch": 1.8763516092100243, + "grad_norm": 2.603278875350952, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8927375078201294, + "num_tokens": 572708996.0, + "step": 14750 + }, + { + "epoch": 1.8764788194886148, + "grad_norm": 2.413795232772827, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8928319215774536, + "num_tokens": 572752640.0, + "step": 14751 + }, + { + "epoch": 1.8766060297672051, + "grad_norm": 2.4731972217559814, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8887730836868286, + "num_tokens": 572797161.0, + "step": 14752 + }, + { + "epoch": 1.8767332400457957, + "grad_norm": 2.309187412261963, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9049643278121948, + "num_tokens": 572838299.0, + "step": 14753 + }, + { + "epoch": 1.8768604503243862, + "grad_norm": 2.7888054847717285, + "learning_rate": 1e-06, + "loss": 0.3265, + "mean_token_accuracy": 0.8824371099472046, + "num_tokens": 572876760.0, + "step": 14754 + }, + { + "epoch": 1.8769876606029767, + "grad_norm": 2.8491384983062744, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8921869993209839, + "num_tokens": 572908411.0, + "step": 14755 + }, + { + "epoch": 1.8771148708815673, + "grad_norm": 2.619680643081665, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8838918209075928, + "num_tokens": 572950731.0, + "step": 14756 + }, + { + "epoch": 1.8772420811601578, + "grad_norm": 2.795642852783203, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8849930167198181, + "num_tokens": 572990793.0, + "step": 14757 + }, + { + "epoch": 1.877369291438748, + "grad_norm": 2.4554688930511475, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9000082612037659, + "num_tokens": 573030983.0, + "step": 14758 + }, + { + "epoch": 1.8774965017173386, + "grad_norm": 2.3783767223358154, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9064092636108398, + "num_tokens": 573069570.0, + "step": 14759 + }, + { + "epoch": 1.8776237119959291, + "grad_norm": 2.739866018295288, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8895533084869385, + "num_tokens": 573105081.0, + "step": 14760 + }, + { + "epoch": 1.8777509222745197, + "grad_norm": 2.4982552528381348, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8950722217559814, + "num_tokens": 573145978.0, + "step": 14761 + }, + { + "epoch": 1.8778781325531102, + "grad_norm": 2.6632213592529297, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9020652174949646, + "num_tokens": 573180343.0, + "step": 14762 + }, + { + "epoch": 1.8780053428317007, + "grad_norm": 2.8832614421844482, + "learning_rate": 1e-06, + "loss": 0.3373, + "mean_token_accuracy": 0.8812307119369507, + "num_tokens": 573215513.0, + "step": 14763 + }, + { + "epoch": 1.8781325531102913, + "grad_norm": 2.6744589805603027, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8922873735427856, + "num_tokens": 573252612.0, + "step": 14764 + }, + { + "epoch": 1.8782597633888818, + "grad_norm": 2.4976818561553955, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.893186092376709, + "num_tokens": 573292729.0, + "step": 14765 + }, + { + "epoch": 1.8783869736674723, + "grad_norm": 2.4900200366973877, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8934009075164795, + "num_tokens": 573335411.0, + "step": 14766 + }, + { + "epoch": 1.8785141839460628, + "grad_norm": 2.6667373180389404, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8859708309173584, + "num_tokens": 573375665.0, + "step": 14767 + }, + { + "epoch": 1.8786413942246534, + "grad_norm": 2.616386651992798, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8913070559501648, + "num_tokens": 573413691.0, + "step": 14768 + }, + { + "epoch": 1.878768604503244, + "grad_norm": 2.4837279319763184, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8872649073600769, + "num_tokens": 573457595.0, + "step": 14769 + }, + { + "epoch": 1.8788958147818344, + "grad_norm": 2.6908366680145264, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9073129892349243, + "num_tokens": 573492865.0, + "step": 14770 + }, + { + "epoch": 1.879023025060425, + "grad_norm": 2.724982261657715, + "learning_rate": 1e-06, + "loss": 0.3258, + "mean_token_accuracy": 0.8804413080215454, + "num_tokens": 573528762.0, + "step": 14771 + }, + { + "epoch": 1.8791502353390155, + "grad_norm": 2.3209586143493652, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.8979474306106567, + "num_tokens": 573573313.0, + "step": 14772 + }, + { + "epoch": 1.879277445617606, + "grad_norm": 2.619325876235962, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9008191227912903, + "num_tokens": 573613058.0, + "step": 14773 + }, + { + "epoch": 1.8794046558961965, + "grad_norm": 2.740177869796753, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8751000165939331, + "num_tokens": 573653840.0, + "step": 14774 + }, + { + "epoch": 1.879531866174787, + "grad_norm": 2.6508796215057373, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8940644264221191, + "num_tokens": 573688998.0, + "step": 14775 + }, + { + "epoch": 1.8796590764533776, + "grad_norm": 2.5018470287323, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8926630020141602, + "num_tokens": 573731932.0, + "step": 14776 + }, + { + "epoch": 1.879786286731968, + "grad_norm": 2.4515721797943115, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8932263851165771, + "num_tokens": 573775607.0, + "step": 14777 + }, + { + "epoch": 1.8799134970105584, + "grad_norm": 2.47617244720459, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9040000438690186, + "num_tokens": 573813377.0, + "step": 14778 + }, + { + "epoch": 1.880040707289149, + "grad_norm": 2.6820225715637207, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8887940049171448, + "num_tokens": 573853566.0, + "step": 14779 + }, + { + "epoch": 1.8801679175677395, + "grad_norm": 2.777146100997925, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8838819265365601, + "num_tokens": 573888514.0, + "step": 14780 + }, + { + "epoch": 1.88029512784633, + "grad_norm": 2.655911445617676, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9007145166397095, + "num_tokens": 573923685.0, + "step": 14781 + }, + { + "epoch": 1.8804223381249205, + "grad_norm": 2.509524345397949, + "learning_rate": 1e-06, + "loss": 0.3421, + "mean_token_accuracy": 0.8758115768432617, + "num_tokens": 573966903.0, + "step": 14782 + }, + { + "epoch": 1.8805495484035109, + "grad_norm": 2.5318124294281006, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8853461742401123, + "num_tokens": 574009353.0, + "step": 14783 + }, + { + "epoch": 1.8806767586821014, + "grad_norm": 2.513437271118164, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8942283391952515, + "num_tokens": 574050704.0, + "step": 14784 + }, + { + "epoch": 1.880803968960692, + "grad_norm": 2.4622058868408203, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8912059664726257, + "num_tokens": 574092754.0, + "step": 14785 + }, + { + "epoch": 1.8809311792392824, + "grad_norm": 2.4908816814422607, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8989443778991699, + "num_tokens": 574132343.0, + "step": 14786 + }, + { + "epoch": 1.881058389517873, + "grad_norm": 2.5618977546691895, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8905625343322754, + "num_tokens": 574170434.0, + "step": 14787 + }, + { + "epoch": 1.8811855997964635, + "grad_norm": 2.316544532775879, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8963500261306763, + "num_tokens": 574215812.0, + "step": 14788 + }, + { + "epoch": 1.881312810075054, + "grad_norm": 2.766794443130493, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8911359310150146, + "num_tokens": 574250040.0, + "step": 14789 + }, + { + "epoch": 1.8814400203536445, + "grad_norm": 2.643585205078125, + "learning_rate": 1e-06, + "loss": 0.3766, + "mean_token_accuracy": 0.8665868043899536, + "num_tokens": 574294433.0, + "step": 14790 + }, + { + "epoch": 1.881567230632235, + "grad_norm": 2.559154748916626, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8889046311378479, + "num_tokens": 574335417.0, + "step": 14791 + }, + { + "epoch": 1.8816944409108256, + "grad_norm": 2.312551736831665, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9083534479141235, + "num_tokens": 574375250.0, + "step": 14792 + }, + { + "epoch": 1.8818216511894161, + "grad_norm": 2.5514073371887207, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.8995683789253235, + "num_tokens": 574415570.0, + "step": 14793 + }, + { + "epoch": 1.8819488614680067, + "grad_norm": 2.6803555488586426, + "learning_rate": 1e-06, + "loss": 0.3178, + "mean_token_accuracy": 0.8828294277191162, + "num_tokens": 574451410.0, + "step": 14794 + }, + { + "epoch": 1.8820760717465972, + "grad_norm": 2.7525248527526855, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8988056778907776, + "num_tokens": 574484781.0, + "step": 14795 + }, + { + "epoch": 1.8822032820251877, + "grad_norm": 2.645164728164673, + "learning_rate": 1e-06, + "loss": 0.323, + "mean_token_accuracy": 0.8824945092201233, + "num_tokens": 574521813.0, + "step": 14796 + }, + { + "epoch": 1.8823304923037782, + "grad_norm": 2.3565189838409424, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.890168309211731, + "num_tokens": 574567502.0, + "step": 14797 + }, + { + "epoch": 1.8824577025823688, + "grad_norm": 2.60599946975708, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8823634386062622, + "num_tokens": 574608063.0, + "step": 14798 + }, + { + "epoch": 1.8825849128609593, + "grad_norm": 2.4752790927886963, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9081742167472839, + "num_tokens": 574647124.0, + "step": 14799 + }, + { + "epoch": 1.8827121231395498, + "grad_norm": 2.7184898853302, + "learning_rate": 1e-06, + "loss": 0.34, + "mean_token_accuracy": 0.8783435225486755, + "num_tokens": 574688951.0, + "step": 14800 + }, + { + "epoch": 1.8828393334181401, + "grad_norm": 2.6506009101867676, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8862090110778809, + "num_tokens": 574727705.0, + "step": 14801 + }, + { + "epoch": 1.8829665436967307, + "grad_norm": 2.629373788833618, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.887496829032898, + "num_tokens": 574766082.0, + "step": 14802 + }, + { + "epoch": 1.8830937539753212, + "grad_norm": 2.644268274307251, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.883027970790863, + "num_tokens": 574806970.0, + "step": 14803 + }, + { + "epoch": 1.8832209642539117, + "grad_norm": 2.425748586654663, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8977472186088562, + "num_tokens": 574850352.0, + "step": 14804 + }, + { + "epoch": 1.8833481745325023, + "grad_norm": 2.416994571685791, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8972323536872864, + "num_tokens": 574893034.0, + "step": 14805 + }, + { + "epoch": 1.8834753848110928, + "grad_norm": 2.622689723968506, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8831691741943359, + "num_tokens": 574932257.0, + "step": 14806 + }, + { + "epoch": 1.883602595089683, + "grad_norm": 2.500751495361328, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.8991569876670837, + "num_tokens": 574971382.0, + "step": 14807 + }, + { + "epoch": 1.8837298053682736, + "grad_norm": 2.4284751415252686, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9087109565734863, + "num_tokens": 575009337.0, + "step": 14808 + }, + { + "epoch": 1.8838570156468641, + "grad_norm": 2.405116081237793, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.894984245300293, + "num_tokens": 575050210.0, + "step": 14809 + }, + { + "epoch": 1.8839842259254547, + "grad_norm": 2.671719789505005, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8993323445320129, + "num_tokens": 575085440.0, + "step": 14810 + }, + { + "epoch": 1.8841114362040452, + "grad_norm": 2.796198844909668, + "learning_rate": 1e-06, + "loss": 0.3176, + "mean_token_accuracy": 0.8856713771820068, + "num_tokens": 575120912.0, + "step": 14811 + }, + { + "epoch": 1.8842386464826357, + "grad_norm": 2.5477099418640137, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9087677597999573, + "num_tokens": 575157479.0, + "step": 14812 + }, + { + "epoch": 1.8843658567612263, + "grad_norm": 2.423408269882202, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9017084836959839, + "num_tokens": 575197038.0, + "step": 14813 + }, + { + "epoch": 1.8844930670398168, + "grad_norm": 2.5954370498657227, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9011824131011963, + "num_tokens": 575232393.0, + "step": 14814 + }, + { + "epoch": 1.8846202773184073, + "grad_norm": 2.783642292022705, + "learning_rate": 1e-06, + "loss": 0.3166, + "mean_token_accuracy": 0.8876600861549377, + "num_tokens": 575268217.0, + "step": 14815 + }, + { + "epoch": 1.8847474875969978, + "grad_norm": 2.2314741611480713, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9011713266372681, + "num_tokens": 575314790.0, + "step": 14816 + }, + { + "epoch": 1.8848746978755884, + "grad_norm": 2.361628293991089, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8927524089813232, + "num_tokens": 575357811.0, + "step": 14817 + }, + { + "epoch": 1.885001908154179, + "grad_norm": 2.7198987007141113, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8820135593414307, + "num_tokens": 575393419.0, + "step": 14818 + }, + { + "epoch": 1.8851291184327694, + "grad_norm": 2.2819228172302246, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8947354555130005, + "num_tokens": 575440876.0, + "step": 14819 + }, + { + "epoch": 1.88525632871136, + "grad_norm": 2.6052613258361816, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8987001180648804, + "num_tokens": 575475132.0, + "step": 14820 + }, + { + "epoch": 1.8853835389899505, + "grad_norm": 2.7634336948394775, + "learning_rate": 1e-06, + "loss": 0.3205, + "mean_token_accuracy": 0.8867576718330383, + "num_tokens": 575510650.0, + "step": 14821 + }, + { + "epoch": 1.885510749268541, + "grad_norm": 2.75189208984375, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.8756852746009827, + "num_tokens": 575554530.0, + "step": 14822 + }, + { + "epoch": 1.8856379595471315, + "grad_norm": 2.4892542362213135, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8979142904281616, + "num_tokens": 575593651.0, + "step": 14823 + }, + { + "epoch": 1.885765169825722, + "grad_norm": 2.706056594848633, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8960850238800049, + "num_tokens": 575629826.0, + "step": 14824 + }, + { + "epoch": 1.8858923801043126, + "grad_norm": 2.8192617893218994, + "learning_rate": 1e-06, + "loss": 0.3616, + "mean_token_accuracy": 0.8723819851875305, + "num_tokens": 575666514.0, + "step": 14825 + }, + { + "epoch": 1.886019590382903, + "grad_norm": 2.7799935340881348, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8928278684616089, + "num_tokens": 575701246.0, + "step": 14826 + }, + { + "epoch": 1.8861468006614934, + "grad_norm": 2.618353843688965, + "learning_rate": 1e-06, + "loss": 0.3348, + "mean_token_accuracy": 0.8822523951530457, + "num_tokens": 575743413.0, + "step": 14827 + }, + { + "epoch": 1.886274010940084, + "grad_norm": 2.650697708129883, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8928385972976685, + "num_tokens": 575781298.0, + "step": 14828 + }, + { + "epoch": 1.8864012212186745, + "grad_norm": 3.0603342056274414, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.90227210521698, + "num_tokens": 575809671.0, + "step": 14829 + }, + { + "epoch": 1.886528431497265, + "grad_norm": 2.667339324951172, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.896079421043396, + "num_tokens": 575843655.0, + "step": 14830 + }, + { + "epoch": 1.8866556417758555, + "grad_norm": 2.5854241847991943, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8859505653381348, + "num_tokens": 575885050.0, + "step": 14831 + }, + { + "epoch": 1.8867828520544458, + "grad_norm": 2.413644790649414, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9017063975334167, + "num_tokens": 575926149.0, + "step": 14832 + }, + { + "epoch": 1.8869100623330364, + "grad_norm": 2.495845079421997, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8903321027755737, + "num_tokens": 575967419.0, + "step": 14833 + }, + { + "epoch": 1.887037272611627, + "grad_norm": 2.4980976581573486, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8982492685317993, + "num_tokens": 576005062.0, + "step": 14834 + }, + { + "epoch": 1.8871644828902174, + "grad_norm": 2.63112211227417, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8885074853897095, + "num_tokens": 576047124.0, + "step": 14835 + }, + { + "epoch": 1.887291693168808, + "grad_norm": 2.6138367652893066, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.882988452911377, + "num_tokens": 576090537.0, + "step": 14836 + }, + { + "epoch": 1.8874189034473985, + "grad_norm": 2.551766872406006, + "learning_rate": 1e-06, + "loss": 0.3438, + "mean_token_accuracy": 0.8776870369911194, + "num_tokens": 576137418.0, + "step": 14837 + }, + { + "epoch": 1.887546113725989, + "grad_norm": 2.5578694343566895, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9061102867126465, + "num_tokens": 576170913.0, + "step": 14838 + }, + { + "epoch": 1.8876733240045795, + "grad_norm": 2.487377166748047, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8928829431533813, + "num_tokens": 576211941.0, + "step": 14839 + }, + { + "epoch": 1.88780053428317, + "grad_norm": 2.5761449337005615, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8918140530586243, + "num_tokens": 576252261.0, + "step": 14840 + }, + { + "epoch": 1.8879277445617606, + "grad_norm": 2.753466844558716, + "learning_rate": 1e-06, + "loss": 0.3638, + "mean_token_accuracy": 0.8713467121124268, + "num_tokens": 576292688.0, + "step": 14841 + }, + { + "epoch": 1.8880549548403511, + "grad_norm": 2.7750420570373535, + "learning_rate": 1e-06, + "loss": 0.3488, + "mean_token_accuracy": 0.8788212537765503, + "num_tokens": 576330764.0, + "step": 14842 + }, + { + "epoch": 1.8881821651189417, + "grad_norm": 2.6247289180755615, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8978860378265381, + "num_tokens": 576366809.0, + "step": 14843 + }, + { + "epoch": 1.8883093753975322, + "grad_norm": 2.602973699569702, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.8905549049377441, + "num_tokens": 576405207.0, + "step": 14844 + }, + { + "epoch": 1.8884365856761227, + "grad_norm": 2.4413769245147705, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8868062496185303, + "num_tokens": 576450271.0, + "step": 14845 + }, + { + "epoch": 1.8885637959547132, + "grad_norm": 2.654768943786621, + "learning_rate": 1e-06, + "loss": 0.3611, + "mean_token_accuracy": 0.8719730973243713, + "num_tokens": 576494921.0, + "step": 14846 + }, + { + "epoch": 1.8886910062333038, + "grad_norm": 2.4185190200805664, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8888120055198669, + "num_tokens": 576535260.0, + "step": 14847 + }, + { + "epoch": 1.8888182165118943, + "grad_norm": 2.5354390144348145, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8935121297836304, + "num_tokens": 576573499.0, + "step": 14848 + }, + { + "epoch": 1.8889454267904848, + "grad_norm": 2.789357900619507, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8819133043289185, + "num_tokens": 576611264.0, + "step": 14849 + }, + { + "epoch": 1.8890726370690751, + "grad_norm": 2.5311553478240967, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8906574845314026, + "num_tokens": 576650073.0, + "step": 14850 + }, + { + "epoch": 1.8891998473476657, + "grad_norm": 2.720465660095215, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8806147575378418, + "num_tokens": 576688275.0, + "step": 14851 + }, + { + "epoch": 1.8893270576262562, + "grad_norm": 2.3673808574676514, + "learning_rate": 1e-06, + "loss": 0.3288, + "mean_token_accuracy": 0.8815414309501648, + "num_tokens": 576736844.0, + "step": 14852 + }, + { + "epoch": 1.8894542679048467, + "grad_norm": 2.6850531101226807, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.893734335899353, + "num_tokens": 576772485.0, + "step": 14853 + }, + { + "epoch": 1.8895814781834372, + "grad_norm": 2.7984414100646973, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8995458483695984, + "num_tokens": 576803447.0, + "step": 14854 + }, + { + "epoch": 1.8897086884620278, + "grad_norm": 2.8244287967681885, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8909422755241394, + "num_tokens": 576836015.0, + "step": 14855 + }, + { + "epoch": 1.889835898740618, + "grad_norm": 2.8869946002960205, + "learning_rate": 1e-06, + "loss": 0.3227, + "mean_token_accuracy": 0.8850303292274475, + "num_tokens": 576868251.0, + "step": 14856 + }, + { + "epoch": 1.8899631090192086, + "grad_norm": 2.641273260116577, + "learning_rate": 1e-06, + "loss": 0.3368, + "mean_token_accuracy": 0.8798903226852417, + "num_tokens": 576909910.0, + "step": 14857 + }, + { + "epoch": 1.8900903192977991, + "grad_norm": 2.4719796180725098, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.8868299722671509, + "num_tokens": 576953354.0, + "step": 14858 + }, + { + "epoch": 1.8902175295763897, + "grad_norm": 2.6254186630249023, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8907502293586731, + "num_tokens": 576989704.0, + "step": 14859 + }, + { + "epoch": 1.8903447398549802, + "grad_norm": 2.4243476390838623, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9029285907745361, + "num_tokens": 577029649.0, + "step": 14860 + }, + { + "epoch": 1.8904719501335707, + "grad_norm": 2.435319185256958, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9103248119354248, + "num_tokens": 577066463.0, + "step": 14861 + }, + { + "epoch": 1.8905991604121613, + "grad_norm": 2.7696115970611572, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8992621898651123, + "num_tokens": 577101561.0, + "step": 14862 + }, + { + "epoch": 1.8907263706907518, + "grad_norm": 2.4696340560913086, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8913295865058899, + "num_tokens": 577142417.0, + "step": 14863 + }, + { + "epoch": 1.8908535809693423, + "grad_norm": 2.7381985187530518, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8889158368110657, + "num_tokens": 577178674.0, + "step": 14864 + }, + { + "epoch": 1.8909807912479328, + "grad_norm": 2.653550148010254, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8902449607849121, + "num_tokens": 577217832.0, + "step": 14865 + }, + { + "epoch": 1.8911080015265234, + "grad_norm": 2.923490285873413, + "learning_rate": 1e-06, + "loss": 0.3631, + "mean_token_accuracy": 0.869064211845398, + "num_tokens": 577251709.0, + "step": 14866 + }, + { + "epoch": 1.891235211805114, + "grad_norm": 2.4494659900665283, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8931347131729126, + "num_tokens": 577292691.0, + "step": 14867 + }, + { + "epoch": 1.8913624220837044, + "grad_norm": 2.4875454902648926, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.894582211971283, + "num_tokens": 577335298.0, + "step": 14868 + }, + { + "epoch": 1.891489632362295, + "grad_norm": 2.4371211528778076, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8990761041641235, + "num_tokens": 577375469.0, + "step": 14869 + }, + { + "epoch": 1.8916168426408855, + "grad_norm": 2.5456302165985107, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9013241529464722, + "num_tokens": 577411107.0, + "step": 14870 + }, + { + "epoch": 1.891744052919476, + "grad_norm": 2.505391836166382, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.895267128944397, + "num_tokens": 577450520.0, + "step": 14871 + }, + { + "epoch": 1.8918712631980665, + "grad_norm": 2.5559470653533936, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8935889601707458, + "num_tokens": 577491526.0, + "step": 14872 + }, + { + "epoch": 1.891998473476657, + "grad_norm": 2.641362190246582, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8889073133468628, + "num_tokens": 577527151.0, + "step": 14873 + }, + { + "epoch": 1.8921256837552476, + "grad_norm": 2.5093252658843994, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.9021583199501038, + "num_tokens": 577562975.0, + "step": 14874 + }, + { + "epoch": 1.892252894033838, + "grad_norm": 2.470339298248291, + "learning_rate": 1e-06, + "loss": 0.3261, + "mean_token_accuracy": 0.8828771114349365, + "num_tokens": 577608569.0, + "step": 14875 + }, + { + "epoch": 1.8923801043124284, + "grad_norm": 2.436392307281494, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9055061340332031, + "num_tokens": 577645558.0, + "step": 14876 + }, + { + "epoch": 1.892507314591019, + "grad_norm": 2.8487095832824707, + "learning_rate": 1e-06, + "loss": 0.3106, + "mean_token_accuracy": 0.888774573802948, + "num_tokens": 577680216.0, + "step": 14877 + }, + { + "epoch": 1.8926345248696095, + "grad_norm": 2.6078972816467285, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8955172896385193, + "num_tokens": 577718689.0, + "step": 14878 + }, + { + "epoch": 1.8927617351482, + "grad_norm": 2.318688154220581, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9076444506645203, + "num_tokens": 577756704.0, + "step": 14879 + }, + { + "epoch": 1.8928889454267905, + "grad_norm": 2.504664897918701, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8993306756019592, + "num_tokens": 577794104.0, + "step": 14880 + }, + { + "epoch": 1.8930161557053808, + "grad_norm": 2.6514015197753906, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9005732536315918, + "num_tokens": 577832329.0, + "step": 14881 + }, + { + "epoch": 1.8931433659839714, + "grad_norm": 2.889127254486084, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8916051983833313, + "num_tokens": 577869331.0, + "step": 14882 + }, + { + "epoch": 1.893270576262562, + "grad_norm": 2.8725736141204834, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8936679363250732, + "num_tokens": 577901831.0, + "step": 14883 + }, + { + "epoch": 1.8933977865411524, + "grad_norm": 2.4851114749908447, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8930099606513977, + "num_tokens": 577944287.0, + "step": 14884 + }, + { + "epoch": 1.893524996819743, + "grad_norm": 2.748467445373535, + "learning_rate": 1e-06, + "loss": 0.3836, + "mean_token_accuracy": 0.866230309009552, + "num_tokens": 577984916.0, + "step": 14885 + }, + { + "epoch": 1.8936522070983335, + "grad_norm": 2.3695743083953857, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.901943027973175, + "num_tokens": 578027477.0, + "step": 14886 + }, + { + "epoch": 1.893779417376924, + "grad_norm": 2.493900775909424, + "learning_rate": 1e-06, + "loss": 0.3193, + "mean_token_accuracy": 0.8859487175941467, + "num_tokens": 578069090.0, + "step": 14887 + }, + { + "epoch": 1.8939066276555145, + "grad_norm": 2.3976216316223145, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.898279070854187, + "num_tokens": 578110222.0, + "step": 14888 + }, + { + "epoch": 1.894033837934105, + "grad_norm": 2.6588785648345947, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8858984708786011, + "num_tokens": 578148585.0, + "step": 14889 + }, + { + "epoch": 1.8941610482126956, + "grad_norm": 2.5190911293029785, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8871236443519592, + "num_tokens": 578191315.0, + "step": 14890 + }, + { + "epoch": 1.8942882584912861, + "grad_norm": 2.7403271198272705, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8920973539352417, + "num_tokens": 578228195.0, + "step": 14891 + }, + { + "epoch": 1.8944154687698767, + "grad_norm": 2.656280040740967, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8797474503517151, + "num_tokens": 578267351.0, + "step": 14892 + }, + { + "epoch": 1.8945426790484672, + "grad_norm": 2.764594078063965, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8850597739219666, + "num_tokens": 578305587.0, + "step": 14893 + }, + { + "epoch": 1.8946698893270577, + "grad_norm": 2.675569772720337, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.903255820274353, + "num_tokens": 578340560.0, + "step": 14894 + }, + { + "epoch": 1.8947970996056482, + "grad_norm": 2.7191410064697266, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8921796083450317, + "num_tokens": 578378156.0, + "step": 14895 + }, + { + "epoch": 1.8949243098842388, + "grad_norm": 2.7688207626342773, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8964486122131348, + "num_tokens": 578410854.0, + "step": 14896 + }, + { + "epoch": 1.8950515201628293, + "grad_norm": 2.680128574371338, + "learning_rate": 1e-06, + "loss": 0.316, + "mean_token_accuracy": 0.88720703125, + "num_tokens": 578449765.0, + "step": 14897 + }, + { + "epoch": 1.8951787304414198, + "grad_norm": 2.7146716117858887, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8915546536445618, + "num_tokens": 578484153.0, + "step": 14898 + }, + { + "epoch": 1.8953059407200101, + "grad_norm": 2.675776243209839, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8912362456321716, + "num_tokens": 578521809.0, + "step": 14899 + }, + { + "epoch": 1.8954331509986007, + "grad_norm": 2.5698320865631104, + "learning_rate": 1e-06, + "loss": 0.3108, + "mean_token_accuracy": 0.890162467956543, + "num_tokens": 578560667.0, + "step": 14900 + }, + { + "epoch": 1.8955603612771912, + "grad_norm": 2.5780065059661865, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.887635350227356, + "num_tokens": 578600341.0, + "step": 14901 + }, + { + "epoch": 1.8956875715557817, + "grad_norm": 2.4576358795166016, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8825730085372925, + "num_tokens": 578645847.0, + "step": 14902 + }, + { + "epoch": 1.8958147818343722, + "grad_norm": 2.501565456390381, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9057271480560303, + "num_tokens": 578681509.0, + "step": 14903 + }, + { + "epoch": 1.8959419921129628, + "grad_norm": 2.6026856899261475, + "learning_rate": 1e-06, + "loss": 0.3058, + "mean_token_accuracy": 0.8898223638534546, + "num_tokens": 578721776.0, + "step": 14904 + }, + { + "epoch": 1.896069202391553, + "grad_norm": 2.3606789112091064, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8949317932128906, + "num_tokens": 578762099.0, + "step": 14905 + }, + { + "epoch": 1.8961964126701436, + "grad_norm": 2.479863166809082, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8950104117393494, + "num_tokens": 578801537.0, + "step": 14906 + }, + { + "epoch": 1.8963236229487341, + "grad_norm": 2.5993330478668213, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8904423713684082, + "num_tokens": 578842140.0, + "step": 14907 + }, + { + "epoch": 1.8964508332273247, + "grad_norm": 2.6220200061798096, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.889409065246582, + "num_tokens": 578885710.0, + "step": 14908 + }, + { + "epoch": 1.8965780435059152, + "grad_norm": 2.5088374614715576, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8985643982887268, + "num_tokens": 578923322.0, + "step": 14909 + }, + { + "epoch": 1.8967052537845057, + "grad_norm": 2.4535534381866455, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8901842832565308, + "num_tokens": 578964929.0, + "step": 14910 + }, + { + "epoch": 1.8968324640630962, + "grad_norm": 2.5527565479278564, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8868232369422913, + "num_tokens": 579003804.0, + "step": 14911 + }, + { + "epoch": 1.8969596743416868, + "grad_norm": 2.7004597187042236, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8917622566223145, + "num_tokens": 579038345.0, + "step": 14912 + }, + { + "epoch": 1.8970868846202773, + "grad_norm": 2.7532243728637695, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8974829912185669, + "num_tokens": 579073885.0, + "step": 14913 + }, + { + "epoch": 1.8972140948988678, + "grad_norm": 2.2828478813171387, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9102946519851685, + "num_tokens": 579111997.0, + "step": 14914 + }, + { + "epoch": 1.8973413051774584, + "grad_norm": 2.7790355682373047, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8853633999824524, + "num_tokens": 579146392.0, + "step": 14915 + }, + { + "epoch": 1.8974685154560489, + "grad_norm": 2.791548728942871, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8847037553787231, + "num_tokens": 579180623.0, + "step": 14916 + }, + { + "epoch": 1.8975957257346394, + "grad_norm": 2.4779202938079834, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9003610610961914, + "num_tokens": 579219234.0, + "step": 14917 + }, + { + "epoch": 1.89772293601323, + "grad_norm": 2.4588613510131836, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8977741003036499, + "num_tokens": 579259065.0, + "step": 14918 + }, + { + "epoch": 1.8978501462918205, + "grad_norm": 2.4570810794830322, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.888504147529602, + "num_tokens": 579301078.0, + "step": 14919 + }, + { + "epoch": 1.897977356570411, + "grad_norm": 2.4466726779937744, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8899269104003906, + "num_tokens": 579342482.0, + "step": 14920 + }, + { + "epoch": 1.8981045668490015, + "grad_norm": 2.5738210678100586, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8904225826263428, + "num_tokens": 579380729.0, + "step": 14921 + }, + { + "epoch": 1.898231777127592, + "grad_norm": 2.690126419067383, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8924362659454346, + "num_tokens": 579414588.0, + "step": 14922 + }, + { + "epoch": 1.8983589874061826, + "grad_norm": 2.609847068786621, + "learning_rate": 1e-06, + "loss": 0.3137, + "mean_token_accuracy": 0.887420654296875, + "num_tokens": 579454258.0, + "step": 14923 + }, + { + "epoch": 1.898486197684773, + "grad_norm": 2.775498628616333, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8815831542015076, + "num_tokens": 579490689.0, + "step": 14924 + }, + { + "epoch": 1.8986134079633634, + "grad_norm": 2.5404741764068604, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8970611095428467, + "num_tokens": 579530307.0, + "step": 14925 + }, + { + "epoch": 1.898740618241954, + "grad_norm": 2.6977808475494385, + "learning_rate": 1e-06, + "loss": 0.3276, + "mean_token_accuracy": 0.8821978569030762, + "num_tokens": 579566955.0, + "step": 14926 + }, + { + "epoch": 1.8988678285205445, + "grad_norm": 2.681670665740967, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9079517722129822, + "num_tokens": 579598978.0, + "step": 14927 + }, + { + "epoch": 1.898995038799135, + "grad_norm": 2.555903911590576, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.899625301361084, + "num_tokens": 579637219.0, + "step": 14928 + }, + { + "epoch": 1.8991222490777255, + "grad_norm": 2.425537109375, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9043200016021729, + "num_tokens": 579679392.0, + "step": 14929 + }, + { + "epoch": 1.8992494593563158, + "grad_norm": 2.526078939437866, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8890541791915894, + "num_tokens": 579720619.0, + "step": 14930 + }, + { + "epoch": 1.8993766696349064, + "grad_norm": 2.55202579498291, + "learning_rate": 1e-06, + "loss": 0.3291, + "mean_token_accuracy": 0.8817477226257324, + "num_tokens": 579760956.0, + "step": 14931 + }, + { + "epoch": 1.899503879913497, + "grad_norm": 2.4935719966888428, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9012740254402161, + "num_tokens": 579803871.0, + "step": 14932 + }, + { + "epoch": 1.8996310901920874, + "grad_norm": 2.552947521209717, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8851886987686157, + "num_tokens": 579847147.0, + "step": 14933 + }, + { + "epoch": 1.899758300470678, + "grad_norm": 2.6042561531066895, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.899358868598938, + "num_tokens": 579885196.0, + "step": 14934 + }, + { + "epoch": 1.8998855107492685, + "grad_norm": 2.839614152908325, + "learning_rate": 1e-06, + "loss": 0.3788, + "mean_token_accuracy": 0.8687171339988708, + "num_tokens": 579923322.0, + "step": 14935 + }, + { + "epoch": 1.900012721027859, + "grad_norm": 2.431032180786133, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9031103253364563, + "num_tokens": 579961114.0, + "step": 14936 + }, + { + "epoch": 1.9001399313064495, + "grad_norm": 2.633772134780884, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8919117450714111, + "num_tokens": 580000596.0, + "step": 14937 + }, + { + "epoch": 1.90026714158504, + "grad_norm": 2.6024062633514404, + "learning_rate": 1e-06, + "loss": 0.3296, + "mean_token_accuracy": 0.8807289004325867, + "num_tokens": 580041644.0, + "step": 14938 + }, + { + "epoch": 1.9003943518636306, + "grad_norm": 2.6902451515197754, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8900333046913147, + "num_tokens": 580077977.0, + "step": 14939 + }, + { + "epoch": 1.9005215621422211, + "grad_norm": 2.7162795066833496, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8935248851776123, + "num_tokens": 580117834.0, + "step": 14940 + }, + { + "epoch": 1.9006487724208116, + "grad_norm": 2.547825574874878, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8971716165542603, + "num_tokens": 580154577.0, + "step": 14941 + }, + { + "epoch": 1.9007759826994022, + "grad_norm": 2.812072277069092, + "learning_rate": 1e-06, + "loss": 0.3564, + "mean_token_accuracy": 0.8766847252845764, + "num_tokens": 580194613.0, + "step": 14942 + }, + { + "epoch": 1.9009031929779927, + "grad_norm": 2.8301000595092773, + "learning_rate": 1e-06, + "loss": 0.339, + "mean_token_accuracy": 0.880729079246521, + "num_tokens": 580232028.0, + "step": 14943 + }, + { + "epoch": 1.9010304032565832, + "grad_norm": 2.52974009513855, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8926941752433777, + "num_tokens": 580274443.0, + "step": 14944 + }, + { + "epoch": 1.9011576135351738, + "grad_norm": 2.3460745811462402, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.887703537940979, + "num_tokens": 580322730.0, + "step": 14945 + }, + { + "epoch": 1.9012848238137643, + "grad_norm": 2.4198639392852783, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8905933499336243, + "num_tokens": 580364749.0, + "step": 14946 + }, + { + "epoch": 1.9014120340923548, + "grad_norm": 2.4579365253448486, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8949362635612488, + "num_tokens": 580406104.0, + "step": 14947 + }, + { + "epoch": 1.9015392443709451, + "grad_norm": 2.6659352779388428, + "learning_rate": 1e-06, + "loss": 0.3153, + "mean_token_accuracy": 0.8855251669883728, + "num_tokens": 580443295.0, + "step": 14948 + }, + { + "epoch": 1.9016664546495357, + "grad_norm": 2.7358343601226807, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8885431289672852, + "num_tokens": 580481701.0, + "step": 14949 + }, + { + "epoch": 1.9017936649281262, + "grad_norm": 3.106945037841797, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.8980836868286133, + "num_tokens": 580519614.0, + "step": 14950 + }, + { + "epoch": 1.9019208752067167, + "grad_norm": 2.6764321327209473, + "learning_rate": 1e-06, + "loss": 0.3173, + "mean_token_accuracy": 0.8848426342010498, + "num_tokens": 580559924.0, + "step": 14951 + }, + { + "epoch": 1.9020480854853072, + "grad_norm": 2.88155460357666, + "learning_rate": 1e-06, + "loss": 0.3451, + "mean_token_accuracy": 0.876041054725647, + "num_tokens": 580593711.0, + "step": 14952 + }, + { + "epoch": 1.9021752957638978, + "grad_norm": 2.618983745574951, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8898404240608215, + "num_tokens": 580635453.0, + "step": 14953 + }, + { + "epoch": 1.902302506042488, + "grad_norm": 2.7757740020751953, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8860394954681396, + "num_tokens": 580671713.0, + "step": 14954 + }, + { + "epoch": 1.9024297163210786, + "grad_norm": 2.599433183670044, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8851633071899414, + "num_tokens": 580713023.0, + "step": 14955 + }, + { + "epoch": 1.9025569265996691, + "grad_norm": 2.6588261127471924, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8933032751083374, + "num_tokens": 580749801.0, + "step": 14956 + }, + { + "epoch": 1.9026841368782597, + "grad_norm": 2.7677927017211914, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8946564793586731, + "num_tokens": 580785901.0, + "step": 14957 + }, + { + "epoch": 1.9028113471568502, + "grad_norm": 3.13511323928833, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8815021514892578, + "num_tokens": 580815170.0, + "step": 14958 + }, + { + "epoch": 1.9029385574354407, + "grad_norm": 2.802497625350952, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8870723247528076, + "num_tokens": 580849739.0, + "step": 14959 + }, + { + "epoch": 1.9030657677140312, + "grad_norm": 2.8332860469818115, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8785884976387024, + "num_tokens": 580885545.0, + "step": 14960 + }, + { + "epoch": 1.9031929779926218, + "grad_norm": 2.5967113971710205, + "learning_rate": 1e-06, + "loss": 0.3424, + "mean_token_accuracy": 0.8764553070068359, + "num_tokens": 580930285.0, + "step": 14961 + }, + { + "epoch": 1.9033201882712123, + "grad_norm": 2.50278639793396, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8982844948768616, + "num_tokens": 580968301.0, + "step": 14962 + }, + { + "epoch": 1.9034473985498028, + "grad_norm": 2.7849977016448975, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8915917873382568, + "num_tokens": 581002688.0, + "step": 14963 + }, + { + "epoch": 1.9035746088283934, + "grad_norm": 2.6643295288085938, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8859153985977173, + "num_tokens": 581042669.0, + "step": 14964 + }, + { + "epoch": 1.9037018191069839, + "grad_norm": 2.6475765705108643, + "learning_rate": 1e-06, + "loss": 0.3496, + "mean_token_accuracy": 0.8741510510444641, + "num_tokens": 581083956.0, + "step": 14965 + }, + { + "epoch": 1.9038290293855744, + "grad_norm": 2.5468506813049316, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.8834959864616394, + "num_tokens": 581124739.0, + "step": 14966 + }, + { + "epoch": 1.903956239664165, + "grad_norm": 2.8316032886505127, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8901190757751465, + "num_tokens": 581155860.0, + "step": 14967 + }, + { + "epoch": 1.9040834499427555, + "grad_norm": 2.5863072872161865, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8942540884017944, + "num_tokens": 581194355.0, + "step": 14968 + }, + { + "epoch": 1.904210660221346, + "grad_norm": 2.524162769317627, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9061282873153687, + "num_tokens": 581230647.0, + "step": 14969 + }, + { + "epoch": 1.9043378704999365, + "grad_norm": 2.5320465564727783, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8901827931404114, + "num_tokens": 581274539.0, + "step": 14970 + }, + { + "epoch": 1.904465080778527, + "grad_norm": 2.410250186920166, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9033284187316895, + "num_tokens": 581311801.0, + "step": 14971 + }, + { + "epoch": 1.9045922910571176, + "grad_norm": 2.8506505489349365, + "learning_rate": 1e-06, + "loss": 0.3218, + "mean_token_accuracy": 0.8837857246398926, + "num_tokens": 581347241.0, + "step": 14972 + }, + { + "epoch": 1.9047195013357079, + "grad_norm": 2.5259382724761963, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8875936269760132, + "num_tokens": 581387665.0, + "step": 14973 + }, + { + "epoch": 1.9048467116142984, + "grad_norm": 2.905888080596924, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8826356530189514, + "num_tokens": 581423901.0, + "step": 14974 + }, + { + "epoch": 1.904973921892889, + "grad_norm": 2.7189950942993164, + "learning_rate": 1e-06, + "loss": 0.3522, + "mean_token_accuracy": 0.8745918273925781, + "num_tokens": 581470554.0, + "step": 14975 + }, + { + "epoch": 1.9051011321714795, + "grad_norm": 2.5960605144500732, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8907460570335388, + "num_tokens": 581507674.0, + "step": 14976 + }, + { + "epoch": 1.90522834245007, + "grad_norm": 2.4137494564056396, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8944997787475586, + "num_tokens": 581551149.0, + "step": 14977 + }, + { + "epoch": 1.9053555527286605, + "grad_norm": 2.6420769691467285, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.891582727432251, + "num_tokens": 581586977.0, + "step": 14978 + }, + { + "epoch": 1.9054827630072508, + "grad_norm": 2.515497922897339, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8892621994018555, + "num_tokens": 581631098.0, + "step": 14979 + }, + { + "epoch": 1.9056099732858414, + "grad_norm": 2.673785448074341, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8881889581680298, + "num_tokens": 581667273.0, + "step": 14980 + }, + { + "epoch": 1.905737183564432, + "grad_norm": 2.7524003982543945, + "learning_rate": 1e-06, + "loss": 0.3363, + "mean_token_accuracy": 0.8773902058601379, + "num_tokens": 581704722.0, + "step": 14981 + }, + { + "epoch": 1.9058643938430224, + "grad_norm": 2.5244123935699463, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8973525166511536, + "num_tokens": 581745939.0, + "step": 14982 + }, + { + "epoch": 1.905991604121613, + "grad_norm": 2.6615757942199707, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8916912078857422, + "num_tokens": 581782109.0, + "step": 14983 + }, + { + "epoch": 1.9061188144002035, + "grad_norm": 2.6334941387176514, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8958683013916016, + "num_tokens": 581817160.0, + "step": 14984 + }, + { + "epoch": 1.906246024678794, + "grad_norm": 2.6903235912323, + "learning_rate": 1e-06, + "loss": 0.344, + "mean_token_accuracy": 0.8736317157745361, + "num_tokens": 581857324.0, + "step": 14985 + }, + { + "epoch": 1.9063732349573845, + "grad_norm": 2.551509380340576, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.8997069597244263, + "num_tokens": 581893320.0, + "step": 14986 + }, + { + "epoch": 1.906500445235975, + "grad_norm": 2.2507917881011963, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8987463712692261, + "num_tokens": 581939247.0, + "step": 14987 + }, + { + "epoch": 1.9066276555145656, + "grad_norm": 2.4794814586639404, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9012565016746521, + "num_tokens": 581975473.0, + "step": 14988 + }, + { + "epoch": 1.9067548657931561, + "grad_norm": 2.7029032707214355, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8951473832130432, + "num_tokens": 582009489.0, + "step": 14989 + }, + { + "epoch": 1.9068820760717466, + "grad_norm": 2.6884546279907227, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.8842395544052124, + "num_tokens": 582047344.0, + "step": 14990 + }, + { + "epoch": 1.9070092863503372, + "grad_norm": 2.713961124420166, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9013773798942566, + "num_tokens": 582080952.0, + "step": 14991 + }, + { + "epoch": 1.9071364966289277, + "grad_norm": 2.376091957092285, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9059910774230957, + "num_tokens": 582122967.0, + "step": 14992 + }, + { + "epoch": 1.9072637069075182, + "grad_norm": 2.560724973678589, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8937911987304688, + "num_tokens": 582163945.0, + "step": 14993 + }, + { + "epoch": 1.9073909171861088, + "grad_norm": 2.5960376262664795, + "learning_rate": 1e-06, + "loss": 0.3505, + "mean_token_accuracy": 0.8735203742980957, + "num_tokens": 582209858.0, + "step": 14994 + }, + { + "epoch": 1.9075181274646993, + "grad_norm": 2.4557080268859863, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9010365009307861, + "num_tokens": 582251675.0, + "step": 14995 + }, + { + "epoch": 1.9076453377432898, + "grad_norm": 2.5733392238616943, + "learning_rate": 1e-06, + "loss": 0.3145, + "mean_token_accuracy": 0.8880383968353271, + "num_tokens": 582291751.0, + "step": 14996 + }, + { + "epoch": 1.9077725480218801, + "grad_norm": 2.417349338531494, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8974043130874634, + "num_tokens": 582335733.0, + "step": 14997 + }, + { + "epoch": 1.9078997583004706, + "grad_norm": 2.593165874481201, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8934853672981262, + "num_tokens": 582376011.0, + "step": 14998 + }, + { + "epoch": 1.9080269685790612, + "grad_norm": 2.8394570350646973, + "learning_rate": 1e-06, + "loss": 0.3282, + "mean_token_accuracy": 0.885871171951294, + "num_tokens": 582414506.0, + "step": 14999 + }, + { + "epoch": 1.9081541788576517, + "grad_norm": 2.7339625358581543, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8851418495178223, + "num_tokens": 582453192.0, + "step": 15000 + }, + { + "epoch": 1.9082813891362422, + "grad_norm": 2.7048499584198, + "learning_rate": 1e-06, + "loss": 0.3377, + "mean_token_accuracy": 0.8800492286682129, + "num_tokens": 582490009.0, + "step": 15001 + }, + { + "epoch": 1.9084085994148328, + "grad_norm": 2.441150665283203, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8950653672218323, + "num_tokens": 582535287.0, + "step": 15002 + }, + { + "epoch": 1.908535809693423, + "grad_norm": 2.5500776767730713, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8910512328147888, + "num_tokens": 582577387.0, + "step": 15003 + }, + { + "epoch": 1.9086630199720136, + "grad_norm": 2.615633249282837, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8912267684936523, + "num_tokens": 582613443.0, + "step": 15004 + }, + { + "epoch": 1.9087902302506041, + "grad_norm": 2.5168838500976562, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8870922327041626, + "num_tokens": 582652075.0, + "step": 15005 + }, + { + "epoch": 1.9089174405291947, + "grad_norm": 2.457562208175659, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9036448001861572, + "num_tokens": 582689162.0, + "step": 15006 + }, + { + "epoch": 1.9090446508077852, + "grad_norm": 2.576021909713745, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8826197385787964, + "num_tokens": 582730201.0, + "step": 15007 + }, + { + "epoch": 1.9091718610863757, + "grad_norm": 2.6172242164611816, + "learning_rate": 1e-06, + "loss": 0.3416, + "mean_token_accuracy": 0.8777949810028076, + "num_tokens": 582771218.0, + "step": 15008 + }, + { + "epoch": 1.9092990713649662, + "grad_norm": 2.5918917655944824, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8874773383140564, + "num_tokens": 582807091.0, + "step": 15009 + }, + { + "epoch": 1.9094262816435568, + "grad_norm": 2.4778125286102295, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9103149175643921, + "num_tokens": 582842401.0, + "step": 15010 + }, + { + "epoch": 1.9095534919221473, + "grad_norm": 2.602722644805908, + "learning_rate": 1e-06, + "loss": 0.2953, + "mean_token_accuracy": 0.8939212560653687, + "num_tokens": 582880822.0, + "step": 15011 + }, + { + "epoch": 1.9096807022007378, + "grad_norm": 2.872359275817871, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8957887887954712, + "num_tokens": 582913056.0, + "step": 15012 + }, + { + "epoch": 1.9098079124793284, + "grad_norm": 2.599287986755371, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8910492658615112, + "num_tokens": 582955113.0, + "step": 15013 + }, + { + "epoch": 1.9099351227579189, + "grad_norm": 2.726860761642456, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8880517482757568, + "num_tokens": 582992999.0, + "step": 15014 + }, + { + "epoch": 1.9100623330365094, + "grad_norm": 2.3649675846099854, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9016992449760437, + "num_tokens": 583034396.0, + "step": 15015 + }, + { + "epoch": 1.9101895433151, + "grad_norm": 2.6493263244628906, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8912551999092102, + "num_tokens": 583073341.0, + "step": 15016 + }, + { + "epoch": 1.9103167535936905, + "grad_norm": 3.075415849685669, + "learning_rate": 1e-06, + "loss": 0.3756, + "mean_token_accuracy": 0.8683115243911743, + "num_tokens": 583108584.0, + "step": 15017 + }, + { + "epoch": 1.910443963872281, + "grad_norm": 2.538792133331299, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8954094648361206, + "num_tokens": 583152221.0, + "step": 15018 + }, + { + "epoch": 1.9105711741508715, + "grad_norm": 3.1761667728424072, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9019181728363037, + "num_tokens": 583188847.0, + "step": 15019 + }, + { + "epoch": 1.910698384429462, + "grad_norm": 2.6116268634796143, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8948734998703003, + "num_tokens": 583226237.0, + "step": 15020 + }, + { + "epoch": 1.9108255947080524, + "grad_norm": 2.649416923522949, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8964369297027588, + "num_tokens": 583263894.0, + "step": 15021 + }, + { + "epoch": 1.9109528049866429, + "grad_norm": 2.8573949337005615, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8915545344352722, + "num_tokens": 583297242.0, + "step": 15022 + }, + { + "epoch": 1.9110800152652334, + "grad_norm": 2.754333972930908, + "learning_rate": 1e-06, + "loss": 0.3915, + "mean_token_accuracy": 0.8597514033317566, + "num_tokens": 583341315.0, + "step": 15023 + }, + { + "epoch": 1.911207225543824, + "grad_norm": 2.7468857765197754, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8927298784255981, + "num_tokens": 583373801.0, + "step": 15024 + }, + { + "epoch": 1.9113344358224145, + "grad_norm": 2.4558167457580566, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8988124132156372, + "num_tokens": 583416191.0, + "step": 15025 + }, + { + "epoch": 1.911461646101005, + "grad_norm": 2.3628487586975098, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9028399586677551, + "num_tokens": 583459292.0, + "step": 15026 + }, + { + "epoch": 1.9115888563795955, + "grad_norm": 2.727905035018921, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8879552483558655, + "num_tokens": 583495126.0, + "step": 15027 + }, + { + "epoch": 1.9117160666581858, + "grad_norm": 2.6347670555114746, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8902218341827393, + "num_tokens": 583534382.0, + "step": 15028 + }, + { + "epoch": 1.9118432769367764, + "grad_norm": 2.4360427856445312, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8826387524604797, + "num_tokens": 583581056.0, + "step": 15029 + }, + { + "epoch": 1.9119704872153669, + "grad_norm": 2.385721445083618, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9020220041275024, + "num_tokens": 583622580.0, + "step": 15030 + }, + { + "epoch": 1.9120976974939574, + "grad_norm": 2.7051620483398438, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8908360004425049, + "num_tokens": 583658018.0, + "step": 15031 + }, + { + "epoch": 1.912224907772548, + "grad_norm": 2.774502754211426, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8921282291412354, + "num_tokens": 583689764.0, + "step": 15032 + }, + { + "epoch": 1.9123521180511385, + "grad_norm": 2.6328814029693604, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8907984495162964, + "num_tokens": 583726646.0, + "step": 15033 + }, + { + "epoch": 1.912479328329729, + "grad_norm": 2.8054745197296143, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8942241072654724, + "num_tokens": 583760468.0, + "step": 15034 + }, + { + "epoch": 1.9126065386083195, + "grad_norm": 2.522810935974121, + "learning_rate": 1e-06, + "loss": 0.3399, + "mean_token_accuracy": 0.8782005310058594, + "num_tokens": 583805152.0, + "step": 15035 + }, + { + "epoch": 1.91273374888691, + "grad_norm": 2.513484239578247, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8984612226486206, + "num_tokens": 583843210.0, + "step": 15036 + }, + { + "epoch": 1.9128609591655006, + "grad_norm": 2.3562541007995605, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8880685567855835, + "num_tokens": 583891633.0, + "step": 15037 + }, + { + "epoch": 1.9129881694440911, + "grad_norm": 2.523749828338623, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9008342027664185, + "num_tokens": 583928240.0, + "step": 15038 + }, + { + "epoch": 1.9131153797226816, + "grad_norm": 2.932695150375366, + "learning_rate": 1e-06, + "loss": 0.3979, + "mean_token_accuracy": 0.8632405996322632, + "num_tokens": 583966939.0, + "step": 15039 + }, + { + "epoch": 1.9132425900012722, + "grad_norm": 2.4361836910247803, + "learning_rate": 1e-06, + "loss": 0.3069, + "mean_token_accuracy": 0.8899415731430054, + "num_tokens": 584008407.0, + "step": 15040 + }, + { + "epoch": 1.9133698002798627, + "grad_norm": 2.602304697036743, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.8966418504714966, + "num_tokens": 584044529.0, + "step": 15041 + }, + { + "epoch": 1.9134970105584532, + "grad_norm": 2.4990367889404297, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.897555947303772, + "num_tokens": 584083095.0, + "step": 15042 + }, + { + "epoch": 1.9136242208370438, + "grad_norm": 2.4925014972686768, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8829174637794495, + "num_tokens": 584124283.0, + "step": 15043 + }, + { + "epoch": 1.9137514311156343, + "grad_norm": 2.7943458557128906, + "learning_rate": 1e-06, + "loss": 0.3566, + "mean_token_accuracy": 0.8730698227882385, + "num_tokens": 584161824.0, + "step": 15044 + }, + { + "epoch": 1.9138786413942248, + "grad_norm": 2.5163168907165527, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8939768671989441, + "num_tokens": 584201635.0, + "step": 15045 + }, + { + "epoch": 1.9140058516728151, + "grad_norm": 2.6245625019073486, + "learning_rate": 1e-06, + "loss": 0.3462, + "mean_token_accuracy": 0.8759028911590576, + "num_tokens": 584245597.0, + "step": 15046 + }, + { + "epoch": 1.9141330619514056, + "grad_norm": 2.8151743412017822, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8896146416664124, + "num_tokens": 584280599.0, + "step": 15047 + }, + { + "epoch": 1.9142602722299962, + "grad_norm": 2.6511785984039307, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8973919749259949, + "num_tokens": 584312794.0, + "step": 15048 + }, + { + "epoch": 1.9143874825085867, + "grad_norm": 2.478659152984619, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8930479288101196, + "num_tokens": 584354812.0, + "step": 15049 + }, + { + "epoch": 1.9145146927871772, + "grad_norm": 2.620156764984131, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8847938179969788, + "num_tokens": 584395684.0, + "step": 15050 + }, + { + "epoch": 1.9146419030657678, + "grad_norm": 2.5393760204315186, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.886693000793457, + "num_tokens": 584441975.0, + "step": 15051 + }, + { + "epoch": 1.914769113344358, + "grad_norm": 2.7009196281433105, + "learning_rate": 1e-06, + "loss": 0.3405, + "mean_token_accuracy": 0.8769259452819824, + "num_tokens": 584484102.0, + "step": 15052 + }, + { + "epoch": 1.9148963236229486, + "grad_norm": 2.556326389312744, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8884215950965881, + "num_tokens": 584525430.0, + "step": 15053 + }, + { + "epoch": 1.9150235339015391, + "grad_norm": 2.57022762298584, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8929080963134766, + "num_tokens": 584564748.0, + "step": 15054 + }, + { + "epoch": 1.9151507441801296, + "grad_norm": 2.638235092163086, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9019051194190979, + "num_tokens": 584596277.0, + "step": 15055 + }, + { + "epoch": 1.9152779544587202, + "grad_norm": 2.3926830291748047, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8951465487480164, + "num_tokens": 584638217.0, + "step": 15056 + }, + { + "epoch": 1.9154051647373107, + "grad_norm": 2.431164264678955, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9015530347824097, + "num_tokens": 584678262.0, + "step": 15057 + }, + { + "epoch": 1.9155323750159012, + "grad_norm": 2.7794430255889893, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.9013290405273438, + "num_tokens": 584714067.0, + "step": 15058 + }, + { + "epoch": 1.9156595852944918, + "grad_norm": 2.402127504348755, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8923953175544739, + "num_tokens": 584759408.0, + "step": 15059 + }, + { + "epoch": 1.9157867955730823, + "grad_norm": 3.271721124649048, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8932269215583801, + "num_tokens": 584786402.0, + "step": 15060 + }, + { + "epoch": 1.9159140058516728, + "grad_norm": 2.831287145614624, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8948908448219299, + "num_tokens": 584818894.0, + "step": 15061 + }, + { + "epoch": 1.9160412161302633, + "grad_norm": 2.583913803100586, + "learning_rate": 1e-06, + "loss": 0.3456, + "mean_token_accuracy": 0.8792098760604858, + "num_tokens": 584860558.0, + "step": 15062 + }, + { + "epoch": 1.9161684264088539, + "grad_norm": 2.9480085372924805, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.8857905268669128, + "num_tokens": 584892733.0, + "step": 15063 + }, + { + "epoch": 1.9162956366874444, + "grad_norm": 2.851668357849121, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.886585533618927, + "num_tokens": 584926532.0, + "step": 15064 + }, + { + "epoch": 1.916422846966035, + "grad_norm": 2.6250667572021484, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8846901655197144, + "num_tokens": 584967277.0, + "step": 15065 + }, + { + "epoch": 1.9165500572446255, + "grad_norm": 2.478517770767212, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.904773473739624, + "num_tokens": 585005372.0, + "step": 15066 + }, + { + "epoch": 1.916677267523216, + "grad_norm": 2.660158157348633, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8886454105377197, + "num_tokens": 585046599.0, + "step": 15067 + }, + { + "epoch": 1.9168044778018065, + "grad_norm": 2.3682727813720703, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8979538679122925, + "num_tokens": 585084537.0, + "step": 15068 + }, + { + "epoch": 1.916931688080397, + "grad_norm": 2.5390148162841797, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8994541764259338, + "num_tokens": 585123244.0, + "step": 15069 + }, + { + "epoch": 1.9170588983589874, + "grad_norm": 2.5710039138793945, + "learning_rate": 1e-06, + "loss": 0.2269, + "mean_token_accuracy": 0.9155541658401489, + "num_tokens": 585154807.0, + "step": 15070 + }, + { + "epoch": 1.9171861086375779, + "grad_norm": 2.5980751514434814, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8890429735183716, + "num_tokens": 585195415.0, + "step": 15071 + }, + { + "epoch": 1.9173133189161684, + "grad_norm": 2.256387948989868, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9057282209396362, + "num_tokens": 585236285.0, + "step": 15072 + }, + { + "epoch": 1.917440529194759, + "grad_norm": 2.4693222045898438, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8972801566123962, + "num_tokens": 585274983.0, + "step": 15073 + }, + { + "epoch": 1.9175677394733495, + "grad_norm": 2.4209964275360107, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9005556106567383, + "num_tokens": 585314255.0, + "step": 15074 + }, + { + "epoch": 1.91769494975194, + "grad_norm": 2.488865375518799, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8879918456077576, + "num_tokens": 585354802.0, + "step": 15075 + }, + { + "epoch": 1.9178221600305305, + "grad_norm": 2.481750965118408, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.8999011516571045, + "num_tokens": 585392720.0, + "step": 15076 + }, + { + "epoch": 1.9179493703091208, + "grad_norm": 2.554980754852295, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8922039270401001, + "num_tokens": 585431933.0, + "step": 15077 + }, + { + "epoch": 1.9180765805877114, + "grad_norm": 2.7587978839874268, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8789331912994385, + "num_tokens": 585470629.0, + "step": 15078 + }, + { + "epoch": 1.9182037908663019, + "grad_norm": 2.841376304626465, + "learning_rate": 1e-06, + "loss": 0.3266, + "mean_token_accuracy": 0.8815180659294128, + "num_tokens": 585507524.0, + "step": 15079 + }, + { + "epoch": 1.9183310011448924, + "grad_norm": 2.4251487255096436, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.8985475301742554, + "num_tokens": 585547412.0, + "step": 15080 + }, + { + "epoch": 1.918458211423483, + "grad_norm": 2.719555377960205, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8838741779327393, + "num_tokens": 585587228.0, + "step": 15081 + }, + { + "epoch": 1.9185854217020735, + "grad_norm": 2.5694093704223633, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8910302519798279, + "num_tokens": 585628667.0, + "step": 15082 + }, + { + "epoch": 1.918712631980664, + "grad_norm": 2.4668304920196533, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8910710215568542, + "num_tokens": 585671776.0, + "step": 15083 + }, + { + "epoch": 1.9188398422592545, + "grad_norm": 2.525881052017212, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8953460454940796, + "num_tokens": 585710534.0, + "step": 15084 + }, + { + "epoch": 1.918967052537845, + "grad_norm": 2.5421254634857178, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.9031131267547607, + "num_tokens": 585751024.0, + "step": 15085 + }, + { + "epoch": 1.9190942628164356, + "grad_norm": 2.490628480911255, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8981491923332214, + "num_tokens": 585790530.0, + "step": 15086 + }, + { + "epoch": 1.919221473095026, + "grad_norm": 2.563976287841797, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8929197192192078, + "num_tokens": 585827681.0, + "step": 15087 + }, + { + "epoch": 1.9193486833736166, + "grad_norm": 2.5544726848602295, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.903734564781189, + "num_tokens": 585864062.0, + "step": 15088 + }, + { + "epoch": 1.9194758936522072, + "grad_norm": 2.6321070194244385, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8817456960678101, + "num_tokens": 585905397.0, + "step": 15089 + }, + { + "epoch": 1.9196031039307977, + "grad_norm": 2.3790791034698486, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.9002939462661743, + "num_tokens": 585948878.0, + "step": 15090 + }, + { + "epoch": 1.9197303142093882, + "grad_norm": 2.506169319152832, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8937398195266724, + "num_tokens": 585988817.0, + "step": 15091 + }, + { + "epoch": 1.9198575244879788, + "grad_norm": 2.6102774143218994, + "learning_rate": 1e-06, + "loss": 0.3431, + "mean_token_accuracy": 0.8773377537727356, + "num_tokens": 586031244.0, + "step": 15092 + }, + { + "epoch": 1.9199847347665693, + "grad_norm": 2.550157070159912, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8940325975418091, + "num_tokens": 586068606.0, + "step": 15093 + }, + { + "epoch": 1.9201119450451598, + "grad_norm": 2.6115951538085938, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.9001041650772095, + "num_tokens": 586105204.0, + "step": 15094 + }, + { + "epoch": 1.9202391553237501, + "grad_norm": 2.4880623817443848, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8952161073684692, + "num_tokens": 586144816.0, + "step": 15095 + }, + { + "epoch": 1.9203663656023406, + "grad_norm": 2.5640053749084473, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.894538164138794, + "num_tokens": 586183472.0, + "step": 15096 + }, + { + "epoch": 1.9204935758809312, + "grad_norm": 2.5657687187194824, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.9001315832138062, + "num_tokens": 586223623.0, + "step": 15097 + }, + { + "epoch": 1.9206207861595217, + "grad_norm": 2.6539840698242188, + "learning_rate": 1e-06, + "loss": 0.3305, + "mean_token_accuracy": 0.8814540505409241, + "num_tokens": 586262639.0, + "step": 15098 + }, + { + "epoch": 1.9207479964381122, + "grad_norm": 2.5985071659088135, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8886107206344604, + "num_tokens": 586304192.0, + "step": 15099 + }, + { + "epoch": 1.9208752067167028, + "grad_norm": 2.3122878074645996, + "learning_rate": 1e-06, + "loss": 0.3186, + "mean_token_accuracy": 0.8872303366661072, + "num_tokens": 586351148.0, + "step": 15100 + }, + { + "epoch": 1.921002416995293, + "grad_norm": 3.015380382537842, + "learning_rate": 1e-06, + "loss": 0.3223, + "mean_token_accuracy": 0.8848506808280945, + "num_tokens": 586383046.0, + "step": 15101 + }, + { + "epoch": 1.9211296272738836, + "grad_norm": 2.694873571395874, + "learning_rate": 1e-06, + "loss": 0.3175, + "mean_token_accuracy": 0.8856796026229858, + "num_tokens": 586420884.0, + "step": 15102 + }, + { + "epoch": 1.9212568375524741, + "grad_norm": 2.523681640625, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8949850797653198, + "num_tokens": 586461688.0, + "step": 15103 + }, + { + "epoch": 1.9213840478310646, + "grad_norm": 2.7019996643066406, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8916357755661011, + "num_tokens": 586497325.0, + "step": 15104 + }, + { + "epoch": 1.9215112581096552, + "grad_norm": 2.729496717453003, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8876067399978638, + "num_tokens": 586535124.0, + "step": 15105 + }, + { + "epoch": 1.9216384683882457, + "grad_norm": 2.5536017417907715, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8998687267303467, + "num_tokens": 586576099.0, + "step": 15106 + }, + { + "epoch": 1.9217656786668362, + "grad_norm": 2.4748027324676514, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8958215713500977, + "num_tokens": 586615792.0, + "step": 15107 + }, + { + "epoch": 1.9218928889454268, + "grad_norm": 2.67541766166687, + "learning_rate": 1e-06, + "loss": 0.3197, + "mean_token_accuracy": 0.8848646283149719, + "num_tokens": 586653329.0, + "step": 15108 + }, + { + "epoch": 1.9220200992240173, + "grad_norm": 2.5451865196228027, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8915376663208008, + "num_tokens": 586694783.0, + "step": 15109 + }, + { + "epoch": 1.9221473095026078, + "grad_norm": 2.6075634956359863, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9078190326690674, + "num_tokens": 586728164.0, + "step": 15110 + }, + { + "epoch": 1.9222745197811983, + "grad_norm": 2.690563678741455, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.8997791409492493, + "num_tokens": 586761050.0, + "step": 15111 + }, + { + "epoch": 1.9224017300597889, + "grad_norm": 2.428096055984497, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9021337628364563, + "num_tokens": 586797653.0, + "step": 15112 + }, + { + "epoch": 1.9225289403383794, + "grad_norm": 2.489816427230835, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8876262903213501, + "num_tokens": 586839678.0, + "step": 15113 + }, + { + "epoch": 1.92265615061697, + "grad_norm": 2.6488754749298096, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.889498233795166, + "num_tokens": 586877290.0, + "step": 15114 + }, + { + "epoch": 1.9227833608955605, + "grad_norm": 2.5509777069091797, + "learning_rate": 1e-06, + "loss": 0.3246, + "mean_token_accuracy": 0.8833034634590149, + "num_tokens": 586920400.0, + "step": 15115 + }, + { + "epoch": 1.922910571174151, + "grad_norm": 3.032942771911621, + "learning_rate": 1e-06, + "loss": 0.3413, + "mean_token_accuracy": 0.878887414932251, + "num_tokens": 586954192.0, + "step": 15116 + }, + { + "epoch": 1.9230377814527415, + "grad_norm": 2.4786553382873535, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8982198238372803, + "num_tokens": 586993780.0, + "step": 15117 + }, + { + "epoch": 1.923164991731332, + "grad_norm": 2.598357915878296, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8973885774612427, + "num_tokens": 587029021.0, + "step": 15118 + }, + { + "epoch": 1.9232922020099223, + "grad_norm": 2.298558235168457, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9031860828399658, + "num_tokens": 587071099.0, + "step": 15119 + }, + { + "epoch": 1.9234194122885129, + "grad_norm": 2.4180378913879395, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8930362462997437, + "num_tokens": 587113650.0, + "step": 15120 + }, + { + "epoch": 1.9235466225671034, + "grad_norm": 2.721096992492676, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8857786655426025, + "num_tokens": 587149748.0, + "step": 15121 + }, + { + "epoch": 1.923673832845694, + "grad_norm": 2.6777608394622803, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8915996551513672, + "num_tokens": 587186972.0, + "step": 15122 + }, + { + "epoch": 1.9238010431242845, + "grad_norm": 2.582484245300293, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8938862681388855, + "num_tokens": 587224009.0, + "step": 15123 + }, + { + "epoch": 1.923928253402875, + "grad_norm": 2.5541791915893555, + "learning_rate": 1e-06, + "loss": 0.322, + "mean_token_accuracy": 0.882007360458374, + "num_tokens": 587264239.0, + "step": 15124 + }, + { + "epoch": 1.9240554636814655, + "grad_norm": 2.9643585681915283, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8899741172790527, + "num_tokens": 587296663.0, + "step": 15125 + }, + { + "epoch": 1.9241826739600558, + "grad_norm": 2.513566732406616, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9063795804977417, + "num_tokens": 587333164.0, + "step": 15126 + }, + { + "epoch": 1.9243098842386464, + "grad_norm": 2.5137434005737305, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8927434682846069, + "num_tokens": 587371917.0, + "step": 15127 + }, + { + "epoch": 1.9244370945172369, + "grad_norm": 2.4317593574523926, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9074813723564148, + "num_tokens": 587411023.0, + "step": 15128 + }, + { + "epoch": 1.9245643047958274, + "grad_norm": 2.6711316108703613, + "learning_rate": 1e-06, + "loss": 0.3161, + "mean_token_accuracy": 0.8854255676269531, + "num_tokens": 587449009.0, + "step": 15129 + }, + { + "epoch": 1.924691515074418, + "grad_norm": 2.731855630874634, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8849784135818481, + "num_tokens": 587484616.0, + "step": 15130 + }, + { + "epoch": 1.9248187253530085, + "grad_norm": 2.85431170463562, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.8818698525428772, + "num_tokens": 587519137.0, + "step": 15131 + }, + { + "epoch": 1.924945935631599, + "grad_norm": 2.934976577758789, + "learning_rate": 1e-06, + "loss": 0.3187, + "mean_token_accuracy": 0.8888195753097534, + "num_tokens": 587551987.0, + "step": 15132 + }, + { + "epoch": 1.9250731459101895, + "grad_norm": 2.561995267868042, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9041648507118225, + "num_tokens": 587589064.0, + "step": 15133 + }, + { + "epoch": 1.92520035618878, + "grad_norm": 2.431988477706909, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8905972242355347, + "num_tokens": 587634453.0, + "step": 15134 + }, + { + "epoch": 1.9253275664673706, + "grad_norm": 2.708066701889038, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8830046653747559, + "num_tokens": 587675034.0, + "step": 15135 + }, + { + "epoch": 1.925454776745961, + "grad_norm": 2.7129993438720703, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8817989826202393, + "num_tokens": 587713910.0, + "step": 15136 + }, + { + "epoch": 1.9255819870245516, + "grad_norm": 2.723696708679199, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8953617811203003, + "num_tokens": 587755682.0, + "step": 15137 + }, + { + "epoch": 1.9257091973031422, + "grad_norm": 2.6483089923858643, + "learning_rate": 1e-06, + "loss": 0.3309, + "mean_token_accuracy": 0.8799771070480347, + "num_tokens": 587798286.0, + "step": 15138 + }, + { + "epoch": 1.9258364075817327, + "grad_norm": 2.819993495941162, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8925215005874634, + "num_tokens": 587836195.0, + "step": 15139 + }, + { + "epoch": 1.9259636178603232, + "grad_norm": 2.485159397125244, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.8991029262542725, + "num_tokens": 587874270.0, + "step": 15140 + }, + { + "epoch": 1.9260908281389137, + "grad_norm": 2.739185094833374, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8885390758514404, + "num_tokens": 587910013.0, + "step": 15141 + }, + { + "epoch": 1.9262180384175043, + "grad_norm": 2.66396164894104, + "learning_rate": 1e-06, + "loss": 0.367, + "mean_token_accuracy": 0.8687730431556702, + "num_tokens": 587950233.0, + "step": 15142 + }, + { + "epoch": 1.9263452486960948, + "grad_norm": 2.705220937728882, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8951665163040161, + "num_tokens": 587983725.0, + "step": 15143 + }, + { + "epoch": 1.926472458974685, + "grad_norm": 2.605672836303711, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8924157619476318, + "num_tokens": 588022993.0, + "step": 15144 + }, + { + "epoch": 1.9265996692532756, + "grad_norm": 2.9330554008483887, + "learning_rate": 1e-06, + "loss": 0.3441, + "mean_token_accuracy": 0.8782172799110413, + "num_tokens": 588058830.0, + "step": 15145 + }, + { + "epoch": 1.9267268795318662, + "grad_norm": 2.575237512588501, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8882936239242554, + "num_tokens": 588099853.0, + "step": 15146 + }, + { + "epoch": 1.9268540898104567, + "grad_norm": 2.5545454025268555, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.899444580078125, + "num_tokens": 588135082.0, + "step": 15147 + }, + { + "epoch": 1.9269813000890472, + "grad_norm": 2.695289134979248, + "learning_rate": 1e-06, + "loss": 0.3417, + "mean_token_accuracy": 0.8775692582130432, + "num_tokens": 588173807.0, + "step": 15148 + }, + { + "epoch": 1.9271085103676378, + "grad_norm": 2.4455316066741943, + "learning_rate": 1e-06, + "loss": 0.307, + "mean_token_accuracy": 0.8912458419799805, + "num_tokens": 588216122.0, + "step": 15149 + }, + { + "epoch": 1.927235720646228, + "grad_norm": 2.93768572807312, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.882125735282898, + "num_tokens": 588250555.0, + "step": 15150 + }, + { + "epoch": 1.9273629309248186, + "grad_norm": 2.6416807174682617, + "learning_rate": 1e-06, + "loss": 0.3361, + "mean_token_accuracy": 0.87660151720047, + "num_tokens": 588292229.0, + "step": 15151 + }, + { + "epoch": 1.9274901412034091, + "grad_norm": 2.429802179336548, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8881146907806396, + "num_tokens": 588335396.0, + "step": 15152 + }, + { + "epoch": 1.9276173514819996, + "grad_norm": 2.55328631401062, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8857858777046204, + "num_tokens": 588377564.0, + "step": 15153 + }, + { + "epoch": 1.9277445617605902, + "grad_norm": 2.428466320037842, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8920935988426208, + "num_tokens": 588420170.0, + "step": 15154 + }, + { + "epoch": 1.9278717720391807, + "grad_norm": 2.704540967941284, + "learning_rate": 1e-06, + "loss": 0.3084, + "mean_token_accuracy": 0.8863574266433716, + "num_tokens": 588457813.0, + "step": 15155 + }, + { + "epoch": 1.9279989823177712, + "grad_norm": 2.5287153720855713, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.8993167877197266, + "num_tokens": 588494166.0, + "step": 15156 + }, + { + "epoch": 1.9281261925963618, + "grad_norm": 2.6129705905914307, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.8866459727287292, + "num_tokens": 588534402.0, + "step": 15157 + }, + { + "epoch": 1.9282534028749523, + "grad_norm": 2.4126968383789062, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.888975977897644, + "num_tokens": 588578516.0, + "step": 15158 + }, + { + "epoch": 1.9283806131535428, + "grad_norm": 2.70365047454834, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.907617449760437, + "num_tokens": 588613770.0, + "step": 15159 + }, + { + "epoch": 1.9285078234321333, + "grad_norm": 2.4654088020324707, + "learning_rate": 1e-06, + "loss": 0.2866, + "mean_token_accuracy": 0.8928982019424438, + "num_tokens": 588656757.0, + "step": 15160 + }, + { + "epoch": 1.9286350337107239, + "grad_norm": 2.8425400257110596, + "learning_rate": 1e-06, + "loss": 0.317, + "mean_token_accuracy": 0.888042151927948, + "num_tokens": 588689668.0, + "step": 15161 + }, + { + "epoch": 1.9287622439893144, + "grad_norm": 2.5524709224700928, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8904229402542114, + "num_tokens": 588731778.0, + "step": 15162 + }, + { + "epoch": 1.928889454267905, + "grad_norm": 2.6405062675476074, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8871381282806396, + "num_tokens": 588767558.0, + "step": 15163 + }, + { + "epoch": 1.9290166645464955, + "grad_norm": 2.5606610774993896, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8928462266921997, + "num_tokens": 588805199.0, + "step": 15164 + }, + { + "epoch": 1.929143874825086, + "grad_norm": 2.4431049823760986, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8933606147766113, + "num_tokens": 588846197.0, + "step": 15165 + }, + { + "epoch": 1.9292710851036765, + "grad_norm": 2.429032325744629, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8950126767158508, + "num_tokens": 588887830.0, + "step": 15166 + }, + { + "epoch": 1.929398295382267, + "grad_norm": 2.4605953693389893, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8953723311424255, + "num_tokens": 588925779.0, + "step": 15167 + }, + { + "epoch": 1.9295255056608573, + "grad_norm": 2.54144287109375, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8956785202026367, + "num_tokens": 588963040.0, + "step": 15168 + }, + { + "epoch": 1.9296527159394479, + "grad_norm": 2.9322829246520996, + "learning_rate": 1e-06, + "loss": 0.3586, + "mean_token_accuracy": 0.873650848865509, + "num_tokens": 589002584.0, + "step": 15169 + }, + { + "epoch": 1.9297799262180384, + "grad_norm": 2.62381911277771, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8939756751060486, + "num_tokens": 589039136.0, + "step": 15170 + }, + { + "epoch": 1.929907136496629, + "grad_norm": 2.957512855529785, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.8919169902801514, + "num_tokens": 589069979.0, + "step": 15171 + }, + { + "epoch": 1.9300343467752195, + "grad_norm": 2.673236608505249, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8901995420455933, + "num_tokens": 589105895.0, + "step": 15172 + }, + { + "epoch": 1.93016155705381, + "grad_norm": 2.525812864303589, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8954793214797974, + "num_tokens": 589143588.0, + "step": 15173 + }, + { + "epoch": 1.9302887673324005, + "grad_norm": 2.775158643722534, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8951225876808167, + "num_tokens": 589178282.0, + "step": 15174 + }, + { + "epoch": 1.9304159776109908, + "grad_norm": 2.4847524166107178, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.907163679599762, + "num_tokens": 589212419.0, + "step": 15175 + }, + { + "epoch": 1.9305431878895813, + "grad_norm": 2.7117116451263428, + "learning_rate": 1e-06, + "loss": 0.319, + "mean_token_accuracy": 0.8850653767585754, + "num_tokens": 589251035.0, + "step": 15176 + }, + { + "epoch": 1.9306703981681719, + "grad_norm": 2.4829189777374268, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9005001783370972, + "num_tokens": 589288941.0, + "step": 15177 + }, + { + "epoch": 1.9307976084467624, + "grad_norm": 2.4278523921966553, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8969727754592896, + "num_tokens": 589327670.0, + "step": 15178 + }, + { + "epoch": 1.930924818725353, + "grad_norm": 2.793421506881714, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8790254592895508, + "num_tokens": 589361876.0, + "step": 15179 + }, + { + "epoch": 1.9310520290039435, + "grad_norm": 2.5613327026367188, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.886176586151123, + "num_tokens": 589400279.0, + "step": 15180 + }, + { + "epoch": 1.931179239282534, + "grad_norm": 2.6718320846557617, + "learning_rate": 1e-06, + "loss": 0.3194, + "mean_token_accuracy": 0.8857209086418152, + "num_tokens": 589440540.0, + "step": 15181 + }, + { + "epoch": 1.9313064495611245, + "grad_norm": 2.6937429904937744, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.8811984658241272, + "num_tokens": 589479637.0, + "step": 15182 + }, + { + "epoch": 1.931433659839715, + "grad_norm": 2.515322208404541, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8917891383171082, + "num_tokens": 589520026.0, + "step": 15183 + }, + { + "epoch": 1.9315608701183056, + "grad_norm": 2.525966167449951, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8899133205413818, + "num_tokens": 589563714.0, + "step": 15184 + }, + { + "epoch": 1.931688080396896, + "grad_norm": 2.7266793251037598, + "learning_rate": 1e-06, + "loss": 0.3061, + "mean_token_accuracy": 0.8876427412033081, + "num_tokens": 589600784.0, + "step": 15185 + }, + { + "epoch": 1.9318152906754866, + "grad_norm": 2.5089428424835205, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8981204628944397, + "num_tokens": 589641159.0, + "step": 15186 + }, + { + "epoch": 1.9319425009540772, + "grad_norm": 2.8354523181915283, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8845696449279785, + "num_tokens": 589678170.0, + "step": 15187 + }, + { + "epoch": 1.9320697112326677, + "grad_norm": 2.930938243865967, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8844631910324097, + "num_tokens": 589711350.0, + "step": 15188 + }, + { + "epoch": 1.9321969215112582, + "grad_norm": 2.7700514793395996, + "learning_rate": 1e-06, + "loss": 0.3095, + "mean_token_accuracy": 0.8859672546386719, + "num_tokens": 589747169.0, + "step": 15189 + }, + { + "epoch": 1.9323241317898487, + "grad_norm": 2.403759241104126, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.898196816444397, + "num_tokens": 589788800.0, + "step": 15190 + }, + { + "epoch": 1.9324513420684393, + "grad_norm": 2.751150131225586, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.8802638649940491, + "num_tokens": 589831712.0, + "step": 15191 + }, + { + "epoch": 1.9325785523470298, + "grad_norm": 2.727367877960205, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8798890709877014, + "num_tokens": 589868121.0, + "step": 15192 + }, + { + "epoch": 1.93270576262562, + "grad_norm": 2.541639566421509, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8941909074783325, + "num_tokens": 589906814.0, + "step": 15193 + }, + { + "epoch": 1.9328329729042106, + "grad_norm": 2.8521924018859863, + "learning_rate": 1e-06, + "loss": 0.3439, + "mean_token_accuracy": 0.8786946535110474, + "num_tokens": 589943873.0, + "step": 15194 + }, + { + "epoch": 1.9329601831828012, + "grad_norm": 2.4058148860931396, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9095616340637207, + "num_tokens": 589981667.0, + "step": 15195 + }, + { + "epoch": 1.9330873934613917, + "grad_norm": 2.5525381565093994, + "learning_rate": 1e-06, + "loss": 0.3537, + "mean_token_accuracy": 0.8730810880661011, + "num_tokens": 590026106.0, + "step": 15196 + }, + { + "epoch": 1.9332146037399822, + "grad_norm": 2.7533750534057617, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8990739583969116, + "num_tokens": 590056646.0, + "step": 15197 + }, + { + "epoch": 1.9333418140185727, + "grad_norm": 2.4952471256256104, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8986793756484985, + "num_tokens": 590095385.0, + "step": 15198 + }, + { + "epoch": 1.933469024297163, + "grad_norm": 2.724910259246826, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.8983727097511292, + "num_tokens": 590126211.0, + "step": 15199 + }, + { + "epoch": 1.9335962345757536, + "grad_norm": 2.7873637676239014, + "learning_rate": 1e-06, + "loss": 0.3869, + "mean_token_accuracy": 0.8649911880493164, + "num_tokens": 590170894.0, + "step": 15200 + }, + { + "epoch": 1.933723444854344, + "grad_norm": 2.4698565006256104, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.891535222530365, + "num_tokens": 590210765.0, + "step": 15201 + }, + { + "epoch": 1.9338506551329346, + "grad_norm": 2.340798854827881, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.902580976486206, + "num_tokens": 590250866.0, + "step": 15202 + }, + { + "epoch": 1.9339778654115252, + "grad_norm": 2.6883487701416016, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.885137140750885, + "num_tokens": 590290764.0, + "step": 15203 + }, + { + "epoch": 1.9341050756901157, + "grad_norm": 3.4053895473480225, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8841692209243774, + "num_tokens": 590335107.0, + "step": 15204 + }, + { + "epoch": 1.9342322859687062, + "grad_norm": 2.5071213245391846, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.893056333065033, + "num_tokens": 590374548.0, + "step": 15205 + }, + { + "epoch": 1.9343594962472968, + "grad_norm": 2.5444588661193848, + "learning_rate": 1e-06, + "loss": 0.3203, + "mean_token_accuracy": 0.8901278972625732, + "num_tokens": 590416225.0, + "step": 15206 + }, + { + "epoch": 1.9344867065258873, + "grad_norm": 2.542372465133667, + "learning_rate": 1e-06, + "loss": 0.3379, + "mean_token_accuracy": 0.8805007338523865, + "num_tokens": 590458782.0, + "step": 15207 + }, + { + "epoch": 1.9346139168044778, + "grad_norm": 2.9176411628723145, + "learning_rate": 1e-06, + "loss": 0.3272, + "mean_token_accuracy": 0.8806096315383911, + "num_tokens": 590492555.0, + "step": 15208 + }, + { + "epoch": 1.9347411270830683, + "grad_norm": 2.646538257598877, + "learning_rate": 1e-06, + "loss": 0.3065, + "mean_token_accuracy": 0.8900321125984192, + "num_tokens": 590529243.0, + "step": 15209 + }, + { + "epoch": 1.9348683373616589, + "grad_norm": 2.207793712615967, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9061710238456726, + "num_tokens": 590574106.0, + "step": 15210 + }, + { + "epoch": 1.9349955476402494, + "grad_norm": 2.8736281394958496, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8922505974769592, + "num_tokens": 590607693.0, + "step": 15211 + }, + { + "epoch": 1.93512275791884, + "grad_norm": 2.3877174854278564, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8999595046043396, + "num_tokens": 590650635.0, + "step": 15212 + }, + { + "epoch": 1.9352499681974304, + "grad_norm": 2.716411828994751, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8985595703125, + "num_tokens": 590682490.0, + "step": 15213 + }, + { + "epoch": 1.935377178476021, + "grad_norm": 2.497358560562134, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8892596960067749, + "num_tokens": 590726638.0, + "step": 15214 + }, + { + "epoch": 1.9355043887546115, + "grad_norm": 2.6060445308685303, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8926122188568115, + "num_tokens": 590767552.0, + "step": 15215 + }, + { + "epoch": 1.935631599033202, + "grad_norm": 2.7253293991088867, + "learning_rate": 1e-06, + "loss": 0.3199, + "mean_token_accuracy": 0.8860710859298706, + "num_tokens": 590802703.0, + "step": 15216 + }, + { + "epoch": 1.9357588093117923, + "grad_norm": 2.484417676925659, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8911398649215698, + "num_tokens": 590846784.0, + "step": 15217 + }, + { + "epoch": 1.9358860195903829, + "grad_norm": 2.5424022674560547, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8960764408111572, + "num_tokens": 590882965.0, + "step": 15218 + }, + { + "epoch": 1.9360132298689734, + "grad_norm": 2.5705196857452393, + "learning_rate": 1e-06, + "loss": 0.3198, + "mean_token_accuracy": 0.8856415748596191, + "num_tokens": 590923469.0, + "step": 15219 + }, + { + "epoch": 1.936140440147564, + "grad_norm": 2.826338291168213, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8798134922981262, + "num_tokens": 590960455.0, + "step": 15220 + }, + { + "epoch": 1.9362676504261545, + "grad_norm": 2.974087715148926, + "learning_rate": 1e-06, + "loss": 0.3297, + "mean_token_accuracy": 0.8831865191459656, + "num_tokens": 591000934.0, + "step": 15221 + }, + { + "epoch": 1.936394860704745, + "grad_norm": 2.77504301071167, + "learning_rate": 1e-06, + "loss": 0.333, + "mean_token_accuracy": 0.8821079134941101, + "num_tokens": 591035692.0, + "step": 15222 + }, + { + "epoch": 1.9365220709833355, + "grad_norm": 2.5086846351623535, + "learning_rate": 1e-06, + "loss": 0.3123, + "mean_token_accuracy": 0.8877739906311035, + "num_tokens": 591080109.0, + "step": 15223 + }, + { + "epoch": 1.9366492812619258, + "grad_norm": 2.4223220348358154, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.902938961982727, + "num_tokens": 591117306.0, + "step": 15224 + }, + { + "epoch": 1.9367764915405163, + "grad_norm": 2.550612688064575, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8889720439910889, + "num_tokens": 591158280.0, + "step": 15225 + }, + { + "epoch": 1.9369037018191069, + "grad_norm": 2.592578411102295, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8958675861358643, + "num_tokens": 591196020.0, + "step": 15226 + }, + { + "epoch": 1.9370309120976974, + "grad_norm": 2.778696060180664, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8998897075653076, + "num_tokens": 591232603.0, + "step": 15227 + }, + { + "epoch": 1.937158122376288, + "grad_norm": 2.4315338134765625, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8934844732284546, + "num_tokens": 591276061.0, + "step": 15228 + }, + { + "epoch": 1.9372853326548785, + "grad_norm": 2.511414051055908, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8983021974563599, + "num_tokens": 591315077.0, + "step": 15229 + }, + { + "epoch": 1.937412542933469, + "grad_norm": 2.589613199234009, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8936372995376587, + "num_tokens": 591353564.0, + "step": 15230 + }, + { + "epoch": 1.9375397532120595, + "grad_norm": 2.6116957664489746, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8885074853897095, + "num_tokens": 591394298.0, + "step": 15231 + }, + { + "epoch": 1.93766696349065, + "grad_norm": 2.8568642139434814, + "learning_rate": 1e-06, + "loss": 0.353, + "mean_token_accuracy": 0.8830486536026001, + "num_tokens": 591434183.0, + "step": 15232 + }, + { + "epoch": 1.9377941737692406, + "grad_norm": 2.702413558959961, + "learning_rate": 1e-06, + "loss": 0.3114, + "mean_token_accuracy": 0.8921579122543335, + "num_tokens": 591472592.0, + "step": 15233 + }, + { + "epoch": 1.937921384047831, + "grad_norm": 2.608556032180786, + "learning_rate": 1e-06, + "loss": 0.3278, + "mean_token_accuracy": 0.8830621838569641, + "num_tokens": 591517469.0, + "step": 15234 + }, + { + "epoch": 1.9380485943264216, + "grad_norm": 2.794893741607666, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8783873319625854, + "num_tokens": 591556508.0, + "step": 15235 + }, + { + "epoch": 1.9381758046050122, + "grad_norm": 2.4469687938690186, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8985499739646912, + "num_tokens": 591598063.0, + "step": 15236 + }, + { + "epoch": 1.9383030148836027, + "grad_norm": 2.7305219173431396, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8991188406944275, + "num_tokens": 591636375.0, + "step": 15237 + }, + { + "epoch": 1.9384302251621932, + "grad_norm": 2.584132194519043, + "learning_rate": 1e-06, + "loss": 0.3268, + "mean_token_accuracy": 0.8822757601737976, + "num_tokens": 591679772.0, + "step": 15238 + }, + { + "epoch": 1.9385574354407837, + "grad_norm": 2.670064926147461, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8858277797698975, + "num_tokens": 591718456.0, + "step": 15239 + }, + { + "epoch": 1.9386846457193743, + "grad_norm": 2.5117568969726562, + "learning_rate": 1e-06, + "loss": 0.309, + "mean_token_accuracy": 0.8882970213890076, + "num_tokens": 591761565.0, + "step": 15240 + }, + { + "epoch": 1.9388118559979648, + "grad_norm": 2.5063562393188477, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8926893472671509, + "num_tokens": 591801011.0, + "step": 15241 + }, + { + "epoch": 1.938939066276555, + "grad_norm": 2.5802454948425293, + "learning_rate": 1e-06, + "loss": 0.3195, + "mean_token_accuracy": 0.886536717414856, + "num_tokens": 591840310.0, + "step": 15242 + }, + { + "epoch": 1.9390662765551456, + "grad_norm": 2.5158851146698, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8943223357200623, + "num_tokens": 591880551.0, + "step": 15243 + }, + { + "epoch": 1.9391934868337362, + "grad_norm": 2.446420907974243, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8874029517173767, + "num_tokens": 591927726.0, + "step": 15244 + }, + { + "epoch": 1.9393206971123267, + "grad_norm": 2.3516082763671875, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9012538194656372, + "num_tokens": 591970235.0, + "step": 15245 + }, + { + "epoch": 1.9394479073909172, + "grad_norm": 2.4914474487304688, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8977540731430054, + "num_tokens": 592008900.0, + "step": 15246 + }, + { + "epoch": 1.9395751176695077, + "grad_norm": 2.543358564376831, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8906579613685608, + "num_tokens": 592049006.0, + "step": 15247 + }, + { + "epoch": 1.939702327948098, + "grad_norm": 2.706115484237671, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8921945095062256, + "num_tokens": 592084812.0, + "step": 15248 + }, + { + "epoch": 1.9398295382266886, + "grad_norm": 2.2151899337768555, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9131817817687988, + "num_tokens": 592125731.0, + "step": 15249 + }, + { + "epoch": 1.939956748505279, + "grad_norm": 2.570470094680786, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9065183401107788, + "num_tokens": 592161259.0, + "step": 15250 + }, + { + "epoch": 1.9400839587838696, + "grad_norm": 2.625720262527466, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8946049809455872, + "num_tokens": 592199066.0, + "step": 15251 + }, + { + "epoch": 1.9402111690624602, + "grad_norm": 2.6282382011413574, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8963017463684082, + "num_tokens": 592237451.0, + "step": 15252 + }, + { + "epoch": 1.9403383793410507, + "grad_norm": 2.6258020401000977, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8868126273155212, + "num_tokens": 592278063.0, + "step": 15253 + }, + { + "epoch": 1.9404655896196412, + "grad_norm": 2.4548263549804688, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8977741003036499, + "num_tokens": 592319400.0, + "step": 15254 + }, + { + "epoch": 1.9405927998982317, + "grad_norm": 2.6640310287475586, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8910636901855469, + "num_tokens": 592354011.0, + "step": 15255 + }, + { + "epoch": 1.9407200101768223, + "grad_norm": 2.3681445121765137, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8854560852050781, + "num_tokens": 592400186.0, + "step": 15256 + }, + { + "epoch": 1.9408472204554128, + "grad_norm": 2.372891426086426, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9055500626564026, + "num_tokens": 592440231.0, + "step": 15257 + }, + { + "epoch": 1.9409744307340033, + "grad_norm": 2.4380297660827637, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9060080051422119, + "num_tokens": 592479824.0, + "step": 15258 + }, + { + "epoch": 1.9411016410125939, + "grad_norm": 2.572099208831787, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8889338970184326, + "num_tokens": 592522071.0, + "step": 15259 + }, + { + "epoch": 1.9412288512911844, + "grad_norm": 2.855613946914673, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8899515867233276, + "num_tokens": 592556745.0, + "step": 15260 + }, + { + "epoch": 1.941356061569775, + "grad_norm": 2.5250799655914307, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9068070650100708, + "num_tokens": 592589206.0, + "step": 15261 + }, + { + "epoch": 1.9414832718483654, + "grad_norm": 2.514267921447754, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8962906002998352, + "num_tokens": 592631534.0, + "step": 15262 + }, + { + "epoch": 1.941610482126956, + "grad_norm": 2.538463592529297, + "learning_rate": 1e-06, + "loss": 0.3026, + "mean_token_accuracy": 0.8908135294914246, + "num_tokens": 592673344.0, + "step": 15263 + }, + { + "epoch": 1.9417376924055465, + "grad_norm": 2.6342830657958984, + "learning_rate": 1e-06, + "loss": 0.3232, + "mean_token_accuracy": 0.8851597309112549, + "num_tokens": 592713280.0, + "step": 15264 + }, + { + "epoch": 1.941864902684137, + "grad_norm": 2.9286983013153076, + "learning_rate": 1e-06, + "loss": 0.3598, + "mean_token_accuracy": 0.8703016042709351, + "num_tokens": 592748891.0, + "step": 15265 + }, + { + "epoch": 1.9419921129627273, + "grad_norm": 2.5978546142578125, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8922678232192993, + "num_tokens": 592789991.0, + "step": 15266 + }, + { + "epoch": 1.9421193232413179, + "grad_norm": 2.6325438022613525, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.8841379284858704, + "num_tokens": 592829137.0, + "step": 15267 + }, + { + "epoch": 1.9422465335199084, + "grad_norm": 2.6302242279052734, + "learning_rate": 1e-06, + "loss": 0.3472, + "mean_token_accuracy": 0.8727400302886963, + "num_tokens": 592869439.0, + "step": 15268 + }, + { + "epoch": 1.942373743798499, + "grad_norm": 2.5600767135620117, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8884379267692566, + "num_tokens": 592910412.0, + "step": 15269 + }, + { + "epoch": 1.9425009540770894, + "grad_norm": 2.7148571014404297, + "learning_rate": 1e-06, + "loss": 0.338, + "mean_token_accuracy": 0.8812956213951111, + "num_tokens": 592947965.0, + "step": 15270 + }, + { + "epoch": 1.94262816435568, + "grad_norm": 2.7421767711639404, + "learning_rate": 1e-06, + "loss": 0.3267, + "mean_token_accuracy": 0.8815668821334839, + "num_tokens": 592985574.0, + "step": 15271 + }, + { + "epoch": 1.9427553746342705, + "grad_norm": 2.63759183883667, + "learning_rate": 1e-06, + "loss": 0.3476, + "mean_token_accuracy": 0.876418948173523, + "num_tokens": 593024326.0, + "step": 15272 + }, + { + "epoch": 1.9428825849128608, + "grad_norm": 2.8176357746124268, + "learning_rate": 1e-06, + "loss": 0.3299, + "mean_token_accuracy": 0.880853533744812, + "num_tokens": 593063555.0, + "step": 15273 + }, + { + "epoch": 1.9430097951914513, + "grad_norm": 2.712810754776001, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8940540552139282, + "num_tokens": 593098089.0, + "step": 15274 + }, + { + "epoch": 1.9431370054700419, + "grad_norm": 2.496410608291626, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.905715823173523, + "num_tokens": 593132239.0, + "step": 15275 + }, + { + "epoch": 1.9432642157486324, + "grad_norm": 2.418426036834717, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8847540616989136, + "num_tokens": 593176793.0, + "step": 15276 + }, + { + "epoch": 1.943391426027223, + "grad_norm": 2.6019771099090576, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8899486064910889, + "num_tokens": 593214370.0, + "step": 15277 + }, + { + "epoch": 1.9435186363058135, + "grad_norm": 2.577955961227417, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8843534588813782, + "num_tokens": 593254633.0, + "step": 15278 + }, + { + "epoch": 1.943645846584404, + "grad_norm": 2.783921718597412, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8930182456970215, + "num_tokens": 593286884.0, + "step": 15279 + }, + { + "epoch": 1.9437730568629945, + "grad_norm": 2.661229372024536, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8859630823135376, + "num_tokens": 593324853.0, + "step": 15280 + }, + { + "epoch": 1.943900267141585, + "grad_norm": 2.7227535247802734, + "learning_rate": 1e-06, + "loss": 0.3141, + "mean_token_accuracy": 0.8881433606147766, + "num_tokens": 593359085.0, + "step": 15281 + }, + { + "epoch": 1.9440274774201756, + "grad_norm": 2.6307668685913086, + "learning_rate": 1e-06, + "loss": 0.304, + "mean_token_accuracy": 0.89467453956604, + "num_tokens": 593402123.0, + "step": 15282 + }, + { + "epoch": 1.944154687698766, + "grad_norm": 2.751981258392334, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8946917057037354, + "num_tokens": 593439631.0, + "step": 15283 + }, + { + "epoch": 1.9442818979773566, + "grad_norm": 2.5712943077087402, + "learning_rate": 1e-06, + "loss": 0.3249, + "mean_token_accuracy": 0.8832399845123291, + "num_tokens": 593479585.0, + "step": 15284 + }, + { + "epoch": 1.9444091082559471, + "grad_norm": 2.613792657852173, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.890762448310852, + "num_tokens": 593517620.0, + "step": 15285 + }, + { + "epoch": 1.9445363185345377, + "grad_norm": 2.6268551349639893, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8884711265563965, + "num_tokens": 593557246.0, + "step": 15286 + }, + { + "epoch": 1.9446635288131282, + "grad_norm": 2.5341103076934814, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.8894868493080139, + "num_tokens": 593601627.0, + "step": 15287 + }, + { + "epoch": 1.9447907390917187, + "grad_norm": 2.6208391189575195, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8856393098831177, + "num_tokens": 593639454.0, + "step": 15288 + }, + { + "epoch": 1.9449179493703093, + "grad_norm": 2.821301221847534, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8886971473693848, + "num_tokens": 593675328.0, + "step": 15289 + }, + { + "epoch": 1.9450451596488998, + "grad_norm": 2.5069761276245117, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8923217058181763, + "num_tokens": 593719658.0, + "step": 15290 + }, + { + "epoch": 1.94517236992749, + "grad_norm": 2.4874112606048584, + "learning_rate": 1e-06, + "loss": 0.3078, + "mean_token_accuracy": 0.8887931108474731, + "num_tokens": 593762056.0, + "step": 15291 + }, + { + "epoch": 1.9452995802060806, + "grad_norm": 2.597388505935669, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8944615125656128, + "num_tokens": 593797500.0, + "step": 15292 + }, + { + "epoch": 1.9454267904846712, + "grad_norm": 2.5584683418273926, + "learning_rate": 1e-06, + "loss": 0.3381, + "mean_token_accuracy": 0.8791751861572266, + "num_tokens": 593841665.0, + "step": 15293 + }, + { + "epoch": 1.9455540007632617, + "grad_norm": 2.785412073135376, + "learning_rate": 1e-06, + "loss": 0.33, + "mean_token_accuracy": 0.8827831745147705, + "num_tokens": 593875390.0, + "step": 15294 + }, + { + "epoch": 1.9456812110418522, + "grad_norm": 2.6225361824035645, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.89637291431427, + "num_tokens": 593910841.0, + "step": 15295 + }, + { + "epoch": 1.9458084213204427, + "grad_norm": 2.4081640243530273, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8930206298828125, + "num_tokens": 593953221.0, + "step": 15296 + }, + { + "epoch": 1.945935631599033, + "grad_norm": 2.532076120376587, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.883480966091156, + "num_tokens": 593995016.0, + "step": 15297 + }, + { + "epoch": 1.9460628418776236, + "grad_norm": 2.76253080368042, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8809703588485718, + "num_tokens": 594033016.0, + "step": 15298 + }, + { + "epoch": 1.946190052156214, + "grad_norm": 2.7715702056884766, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9022146463394165, + "num_tokens": 594065122.0, + "step": 15299 + }, + { + "epoch": 1.9463172624348046, + "grad_norm": 2.612563133239746, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8890142440795898, + "num_tokens": 594104478.0, + "step": 15300 + }, + { + "epoch": 1.9464444727133952, + "grad_norm": 2.4938173294067383, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9085238575935364, + "num_tokens": 594144572.0, + "step": 15301 + }, + { + "epoch": 1.9465716829919857, + "grad_norm": 2.628504514694214, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8985475301742554, + "num_tokens": 594181651.0, + "step": 15302 + }, + { + "epoch": 1.9466988932705762, + "grad_norm": 2.8903465270996094, + "learning_rate": 1e-06, + "loss": 0.3553, + "mean_token_accuracy": 0.873126745223999, + "num_tokens": 594219036.0, + "step": 15303 + }, + { + "epoch": 1.9468261035491667, + "grad_norm": 2.6675095558166504, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8951465487480164, + "num_tokens": 594255022.0, + "step": 15304 + }, + { + "epoch": 1.9469533138277573, + "grad_norm": 2.634019374847412, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8873103857040405, + "num_tokens": 594294426.0, + "step": 15305 + }, + { + "epoch": 1.9470805241063478, + "grad_norm": 2.7869701385498047, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8840206861495972, + "num_tokens": 594332047.0, + "step": 15306 + }, + { + "epoch": 1.9472077343849383, + "grad_norm": 2.7746212482452393, + "learning_rate": 1e-06, + "loss": 0.3133, + "mean_token_accuracy": 0.8851550817489624, + "num_tokens": 594368431.0, + "step": 15307 + }, + { + "epoch": 1.9473349446635289, + "grad_norm": 2.667609691619873, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9011160731315613, + "num_tokens": 594401015.0, + "step": 15308 + }, + { + "epoch": 1.9474621549421194, + "grad_norm": 2.470564603805542, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.8998169898986816, + "num_tokens": 594440619.0, + "step": 15309 + }, + { + "epoch": 1.94758936522071, + "grad_norm": 2.913215160369873, + "learning_rate": 1e-06, + "loss": 0.3512, + "mean_token_accuracy": 0.8756418228149414, + "num_tokens": 594480912.0, + "step": 15310 + }, + { + "epoch": 1.9477165754993004, + "grad_norm": 2.550931453704834, + "learning_rate": 1e-06, + "loss": 0.3292, + "mean_token_accuracy": 0.8804953098297119, + "num_tokens": 594523504.0, + "step": 15311 + }, + { + "epoch": 1.947843785777891, + "grad_norm": 2.512195587158203, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.9004563689231873, + "num_tokens": 594562286.0, + "step": 15312 + }, + { + "epoch": 1.9479709960564815, + "grad_norm": 2.684159994125366, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8912143111228943, + "num_tokens": 594598888.0, + "step": 15313 + }, + { + "epoch": 1.948098206335072, + "grad_norm": 2.6252944469451904, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9076341390609741, + "num_tokens": 594631328.0, + "step": 15314 + }, + { + "epoch": 1.9482254166136623, + "grad_norm": 2.662701368331909, + "learning_rate": 1e-06, + "loss": 0.3056, + "mean_token_accuracy": 0.8898795247077942, + "num_tokens": 594670828.0, + "step": 15315 + }, + { + "epoch": 1.9483526268922529, + "grad_norm": 2.833611249923706, + "learning_rate": 1e-06, + "loss": 0.3275, + "mean_token_accuracy": 0.8829382061958313, + "num_tokens": 594705404.0, + "step": 15316 + }, + { + "epoch": 1.9484798371708434, + "grad_norm": 2.8331117630004883, + "learning_rate": 1e-06, + "loss": 0.3323, + "mean_token_accuracy": 0.8817498087882996, + "num_tokens": 594739926.0, + "step": 15317 + }, + { + "epoch": 1.948607047449434, + "grad_norm": 2.7271735668182373, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8933850526809692, + "num_tokens": 594773075.0, + "step": 15318 + }, + { + "epoch": 1.9487342577280244, + "grad_norm": 2.5991289615631104, + "learning_rate": 1e-06, + "loss": 0.3225, + "mean_token_accuracy": 0.8850719928741455, + "num_tokens": 594816265.0, + "step": 15319 + }, + { + "epoch": 1.948861468006615, + "grad_norm": 2.4250400066375732, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9048605561256409, + "num_tokens": 594855798.0, + "step": 15320 + }, + { + "epoch": 1.9489886782852053, + "grad_norm": 2.735218048095703, + "learning_rate": 1e-06, + "loss": 0.3302, + "mean_token_accuracy": 0.8796529173851013, + "num_tokens": 594894409.0, + "step": 15321 + }, + { + "epoch": 1.9491158885637958, + "grad_norm": 2.6749494075775146, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8849260807037354, + "num_tokens": 594933186.0, + "step": 15322 + }, + { + "epoch": 1.9492430988423863, + "grad_norm": 2.6732680797576904, + "learning_rate": 1e-06, + "loss": 0.3206, + "mean_token_accuracy": 0.8863483667373657, + "num_tokens": 594972973.0, + "step": 15323 + }, + { + "epoch": 1.9493703091209769, + "grad_norm": 2.5867772102355957, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9033328890800476, + "num_tokens": 595007174.0, + "step": 15324 + }, + { + "epoch": 1.9494975193995674, + "grad_norm": 2.621572732925415, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8900262117385864, + "num_tokens": 595047796.0, + "step": 15325 + }, + { + "epoch": 1.949624729678158, + "grad_norm": 2.7136547565460205, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.8877663612365723, + "num_tokens": 595088704.0, + "step": 15326 + }, + { + "epoch": 1.9497519399567484, + "grad_norm": 2.501831531524658, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8936107158660889, + "num_tokens": 595129233.0, + "step": 15327 + }, + { + "epoch": 1.949879150235339, + "grad_norm": 2.7820663452148438, + "learning_rate": 1e-06, + "loss": 0.3372, + "mean_token_accuracy": 0.8795784711837769, + "num_tokens": 595166083.0, + "step": 15328 + }, + { + "epoch": 1.9500063605139295, + "grad_norm": 2.5761783123016357, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8938543796539307, + "num_tokens": 595205973.0, + "step": 15329 + }, + { + "epoch": 1.95013357079252, + "grad_norm": 2.616295337677002, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8911178112030029, + "num_tokens": 595245848.0, + "step": 15330 + }, + { + "epoch": 1.9502607810711106, + "grad_norm": 2.9146292209625244, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9020925760269165, + "num_tokens": 595274482.0, + "step": 15331 + }, + { + "epoch": 1.950387991349701, + "grad_norm": 2.398829698562622, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8919790983200073, + "num_tokens": 595315786.0, + "step": 15332 + }, + { + "epoch": 1.9505152016282916, + "grad_norm": 2.418053388595581, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8968614935874939, + "num_tokens": 595356633.0, + "step": 15333 + }, + { + "epoch": 1.9506424119068821, + "grad_norm": 2.64658784866333, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9062653183937073, + "num_tokens": 595393196.0, + "step": 15334 + }, + { + "epoch": 1.9507696221854727, + "grad_norm": 2.7565548419952393, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.89649897813797, + "num_tokens": 595424180.0, + "step": 15335 + }, + { + "epoch": 1.9508968324640632, + "grad_norm": 2.566884756088257, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.894446074962616, + "num_tokens": 595466815.0, + "step": 15336 + }, + { + "epoch": 1.9510240427426537, + "grad_norm": 2.5254504680633545, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9010453224182129, + "num_tokens": 595500476.0, + "step": 15337 + }, + { + "epoch": 1.9511512530212443, + "grad_norm": 3.2277450561523438, + "learning_rate": 1e-06, + "loss": 0.3318, + "mean_token_accuracy": 0.8834552764892578, + "num_tokens": 595532950.0, + "step": 15338 + }, + { + "epoch": 1.9512784632998348, + "grad_norm": 2.3581762313842773, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.905954122543335, + "num_tokens": 595572592.0, + "step": 15339 + }, + { + "epoch": 1.951405673578425, + "grad_norm": 2.746325969696045, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8823955059051514, + "num_tokens": 595609701.0, + "step": 15340 + }, + { + "epoch": 1.9515328838570156, + "grad_norm": 2.5812253952026367, + "learning_rate": 1e-06, + "loss": 0.3011, + "mean_token_accuracy": 0.8906217813491821, + "num_tokens": 595648363.0, + "step": 15341 + }, + { + "epoch": 1.9516600941356061, + "grad_norm": 2.9926490783691406, + "learning_rate": 1e-06, + "loss": 0.3478, + "mean_token_accuracy": 0.8763238191604614, + "num_tokens": 595683881.0, + "step": 15342 + }, + { + "epoch": 1.9517873044141967, + "grad_norm": 2.780421733856201, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8901052474975586, + "num_tokens": 595718020.0, + "step": 15343 + }, + { + "epoch": 1.9519145146927872, + "grad_norm": 2.318974256515503, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8894855976104736, + "num_tokens": 595766339.0, + "step": 15344 + }, + { + "epoch": 1.9520417249713777, + "grad_norm": 2.6725339889526367, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8958560228347778, + "num_tokens": 595802571.0, + "step": 15345 + }, + { + "epoch": 1.952168935249968, + "grad_norm": 2.6648340225219727, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8912397623062134, + "num_tokens": 595840740.0, + "step": 15346 + }, + { + "epoch": 1.9522961455285586, + "grad_norm": 2.7759878635406494, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8910088539123535, + "num_tokens": 595874103.0, + "step": 15347 + }, + { + "epoch": 1.952423355807149, + "grad_norm": 2.637915849685669, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8890454769134521, + "num_tokens": 595911260.0, + "step": 15348 + }, + { + "epoch": 1.9525505660857396, + "grad_norm": 2.5415589809417725, + "learning_rate": 1e-06, + "loss": 0.2952, + "mean_token_accuracy": 0.8923177719116211, + "num_tokens": 595950273.0, + "step": 15349 + }, + { + "epoch": 1.9526777763643302, + "grad_norm": 2.5481884479522705, + "learning_rate": 1e-06, + "loss": 0.3099, + "mean_token_accuracy": 0.8883036971092224, + "num_tokens": 595989183.0, + "step": 15350 + }, + { + "epoch": 1.9528049866429207, + "grad_norm": 2.763404607772827, + "learning_rate": 1e-06, + "loss": 0.3366, + "mean_token_accuracy": 0.8794363737106323, + "num_tokens": 596025975.0, + "step": 15351 + }, + { + "epoch": 1.9529321969215112, + "grad_norm": 2.693068504333496, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8963327407836914, + "num_tokens": 596060242.0, + "step": 15352 + }, + { + "epoch": 1.9530594072001017, + "grad_norm": 2.3554458618164062, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.906224250793457, + "num_tokens": 596101720.0, + "step": 15353 + }, + { + "epoch": 1.9531866174786923, + "grad_norm": 2.658290147781372, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8976027369499207, + "num_tokens": 596138416.0, + "step": 15354 + }, + { + "epoch": 1.9533138277572828, + "grad_norm": 2.7041122913360596, + "learning_rate": 1e-06, + "loss": 0.3242, + "mean_token_accuracy": 0.8880213499069214, + "num_tokens": 596176754.0, + "step": 15355 + }, + { + "epoch": 1.9534410380358733, + "grad_norm": 2.7901341915130615, + "learning_rate": 1e-06, + "loss": 0.3287, + "mean_token_accuracy": 0.8799605965614319, + "num_tokens": 596215828.0, + "step": 15356 + }, + { + "epoch": 1.9535682483144639, + "grad_norm": 2.7758028507232666, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8897981643676758, + "num_tokens": 596250743.0, + "step": 15357 + }, + { + "epoch": 1.9536954585930544, + "grad_norm": 2.5075676441192627, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8976873159408569, + "num_tokens": 596288452.0, + "step": 15358 + }, + { + "epoch": 1.953822668871645, + "grad_norm": 2.81894850730896, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8868589997291565, + "num_tokens": 596321874.0, + "step": 15359 + }, + { + "epoch": 1.9539498791502354, + "grad_norm": 2.6342244148254395, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8909354209899902, + "num_tokens": 596357922.0, + "step": 15360 + }, + { + "epoch": 1.954077089428826, + "grad_norm": 2.880627393722534, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8867746591567993, + "num_tokens": 596391059.0, + "step": 15361 + }, + { + "epoch": 1.9542042997074165, + "grad_norm": 2.3321590423583984, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8944581747055054, + "num_tokens": 596437466.0, + "step": 15362 + }, + { + "epoch": 1.954331509986007, + "grad_norm": 2.891772508621216, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.8905760049819946, + "num_tokens": 596471066.0, + "step": 15363 + }, + { + "epoch": 1.9544587202645973, + "grad_norm": 2.707899808883667, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.893177330493927, + "num_tokens": 596507670.0, + "step": 15364 + }, + { + "epoch": 1.9545859305431879, + "grad_norm": 2.856543779373169, + "learning_rate": 1e-06, + "loss": 0.3244, + "mean_token_accuracy": 0.8866767883300781, + "num_tokens": 596548367.0, + "step": 15365 + }, + { + "epoch": 1.9547131408217784, + "grad_norm": 2.782285213470459, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8901287317276001, + "num_tokens": 596584093.0, + "step": 15366 + }, + { + "epoch": 1.954840351100369, + "grad_norm": 2.6171693801879883, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8936594724655151, + "num_tokens": 596618054.0, + "step": 15367 + }, + { + "epoch": 1.9549675613789594, + "grad_norm": 2.544779062271118, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9008911848068237, + "num_tokens": 596655057.0, + "step": 15368 + }, + { + "epoch": 1.95509477165755, + "grad_norm": 2.6498472690582275, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.8984556198120117, + "num_tokens": 596690857.0, + "step": 15369 + }, + { + "epoch": 1.9552219819361403, + "grad_norm": 2.878833293914795, + "learning_rate": 1e-06, + "loss": 0.3307, + "mean_token_accuracy": 0.8808079957962036, + "num_tokens": 596727568.0, + "step": 15370 + }, + { + "epoch": 1.9553491922147308, + "grad_norm": 2.7513275146484375, + "learning_rate": 1e-06, + "loss": 0.3524, + "mean_token_accuracy": 0.8712887167930603, + "num_tokens": 596767980.0, + "step": 15371 + }, + { + "epoch": 1.9554764024933213, + "grad_norm": 2.896373987197876, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8855268955230713, + "num_tokens": 596802576.0, + "step": 15372 + }, + { + "epoch": 1.9556036127719119, + "grad_norm": 2.7514302730560303, + "learning_rate": 1e-06, + "loss": 0.3262, + "mean_token_accuracy": 0.8831003308296204, + "num_tokens": 596837861.0, + "step": 15373 + }, + { + "epoch": 1.9557308230505024, + "grad_norm": 2.496976613998413, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.892850935459137, + "num_tokens": 596880981.0, + "step": 15374 + }, + { + "epoch": 1.955858033329093, + "grad_norm": 2.5802338123321533, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8922601938247681, + "num_tokens": 596918293.0, + "step": 15375 + }, + { + "epoch": 1.9559852436076834, + "grad_norm": 2.789910078048706, + "learning_rate": 1e-06, + "loss": 0.3459, + "mean_token_accuracy": 0.8751451969146729, + "num_tokens": 596958668.0, + "step": 15376 + }, + { + "epoch": 1.956112453886274, + "grad_norm": 2.529491901397705, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8973366618156433, + "num_tokens": 597001868.0, + "step": 15377 + }, + { + "epoch": 1.9562396641648645, + "grad_norm": 2.6091842651367188, + "learning_rate": 1e-06, + "loss": 0.3136, + "mean_token_accuracy": 0.8861117362976074, + "num_tokens": 597041228.0, + "step": 15378 + }, + { + "epoch": 1.956366874443455, + "grad_norm": 2.5602948665618896, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8887091875076294, + "num_tokens": 597084494.0, + "step": 15379 + }, + { + "epoch": 1.9564940847220456, + "grad_norm": 2.5125021934509277, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8911011219024658, + "num_tokens": 597126589.0, + "step": 15380 + }, + { + "epoch": 1.956621295000636, + "grad_norm": 2.6710004806518555, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9003663063049316, + "num_tokens": 597161713.0, + "step": 15381 + }, + { + "epoch": 1.9567485052792266, + "grad_norm": 2.7062811851501465, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8951750993728638, + "num_tokens": 597201334.0, + "step": 15382 + }, + { + "epoch": 1.9568757155578171, + "grad_norm": 2.774972438812256, + "learning_rate": 1e-06, + "loss": 0.3558, + "mean_token_accuracy": 0.8735437393188477, + "num_tokens": 597244152.0, + "step": 15383 + }, + { + "epoch": 1.9570029258364077, + "grad_norm": 2.895220994949341, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8849400877952576, + "num_tokens": 597281867.0, + "step": 15384 + }, + { + "epoch": 1.9571301361149982, + "grad_norm": 2.4243524074554443, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.897826611995697, + "num_tokens": 597322339.0, + "step": 15385 + }, + { + "epoch": 1.9572573463935887, + "grad_norm": 2.5768322944641113, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.8990992903709412, + "num_tokens": 597356161.0, + "step": 15386 + }, + { + "epoch": 1.9573845566721793, + "grad_norm": 2.8617379665374756, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8932406902313232, + "num_tokens": 597387935.0, + "step": 15387 + }, + { + "epoch": 1.9575117669507698, + "grad_norm": 2.5823612213134766, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9031969904899597, + "num_tokens": 597424227.0, + "step": 15388 + }, + { + "epoch": 1.95763897722936, + "grad_norm": 2.851731777191162, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.8988445997238159, + "num_tokens": 597458113.0, + "step": 15389 + }, + { + "epoch": 1.9577661875079506, + "grad_norm": 2.7315316200256348, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8901235461235046, + "num_tokens": 597492753.0, + "step": 15390 + }, + { + "epoch": 1.9578933977865411, + "grad_norm": 2.7524447441101074, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9004915356636047, + "num_tokens": 597525040.0, + "step": 15391 + }, + { + "epoch": 1.9580206080651317, + "grad_norm": 2.4426679611206055, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.9014999866485596, + "num_tokens": 597564863.0, + "step": 15392 + }, + { + "epoch": 1.9581478183437222, + "grad_norm": 2.465406656265259, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8937931060791016, + "num_tokens": 597607740.0, + "step": 15393 + }, + { + "epoch": 1.9582750286223127, + "grad_norm": 2.8104822635650635, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8885847926139832, + "num_tokens": 597641929.0, + "step": 15394 + }, + { + "epoch": 1.958402238900903, + "grad_norm": 2.633229970932007, + "learning_rate": 1e-06, + "loss": 0.3202, + "mean_token_accuracy": 0.8825556039810181, + "num_tokens": 597683777.0, + "step": 15395 + }, + { + "epoch": 1.9585294491794936, + "grad_norm": 2.619999408721924, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8960527181625366, + "num_tokens": 597719786.0, + "step": 15396 + }, + { + "epoch": 1.958656659458084, + "grad_norm": 2.6228525638580322, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8947027921676636, + "num_tokens": 597755726.0, + "step": 15397 + }, + { + "epoch": 1.9587838697366746, + "grad_norm": 2.3903684616088867, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9036152362823486, + "num_tokens": 597799448.0, + "step": 15398 + }, + { + "epoch": 1.9589110800152651, + "grad_norm": 2.8238611221313477, + "learning_rate": 1e-06, + "loss": 0.3271, + "mean_token_accuracy": 0.8830360770225525, + "num_tokens": 597836948.0, + "step": 15399 + }, + { + "epoch": 1.9590382902938557, + "grad_norm": 2.4671568870544434, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8940778374671936, + "num_tokens": 597874866.0, + "step": 15400 + }, + { + "epoch": 1.9591655005724462, + "grad_norm": 2.5023112297058105, + "learning_rate": 1e-06, + "loss": 0.3073, + "mean_token_accuracy": 0.8899884819984436, + "num_tokens": 597917379.0, + "step": 15401 + }, + { + "epoch": 1.9592927108510367, + "grad_norm": 2.5094988346099854, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8927420377731323, + "num_tokens": 597959758.0, + "step": 15402 + }, + { + "epoch": 1.9594199211296273, + "grad_norm": 2.6520872116088867, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8943115472793579, + "num_tokens": 597996216.0, + "step": 15403 + }, + { + "epoch": 1.9595471314082178, + "grad_norm": 2.610621690750122, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8948701620101929, + "num_tokens": 598035914.0, + "step": 15404 + }, + { + "epoch": 1.9596743416868083, + "grad_norm": 2.5262813568115234, + "learning_rate": 1e-06, + "loss": 0.3159, + "mean_token_accuracy": 0.886463463306427, + "num_tokens": 598080622.0, + "step": 15405 + }, + { + "epoch": 1.9598015519653988, + "grad_norm": 2.5157570838928223, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8988666534423828, + "num_tokens": 598121089.0, + "step": 15406 + }, + { + "epoch": 1.9599287622439894, + "grad_norm": 2.700491428375244, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.887347400188446, + "num_tokens": 598155278.0, + "step": 15407 + }, + { + "epoch": 1.96005597252258, + "grad_norm": 2.4007678031921387, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9097369909286499, + "num_tokens": 598194140.0, + "step": 15408 + }, + { + "epoch": 1.9601831828011704, + "grad_norm": 2.748154878616333, + "learning_rate": 1e-06, + "loss": 0.3344, + "mean_token_accuracy": 0.8810480237007141, + "num_tokens": 598233421.0, + "step": 15409 + }, + { + "epoch": 1.960310393079761, + "grad_norm": 2.4734671115875244, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8970192670822144, + "num_tokens": 598278159.0, + "step": 15410 + }, + { + "epoch": 1.9604376033583515, + "grad_norm": 2.8705456256866455, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.897388756275177, + "num_tokens": 598311780.0, + "step": 15411 + }, + { + "epoch": 1.960564813636942, + "grad_norm": 2.647404909133911, + "learning_rate": 1e-06, + "loss": 0.3499, + "mean_token_accuracy": 0.8746168613433838, + "num_tokens": 598352751.0, + "step": 15412 + }, + { + "epoch": 1.9606920239155323, + "grad_norm": 2.5068933963775635, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9057580232620239, + "num_tokens": 598389027.0, + "step": 15413 + }, + { + "epoch": 1.9608192341941229, + "grad_norm": 2.4771735668182373, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.8840824365615845, + "num_tokens": 598430645.0, + "step": 15414 + }, + { + "epoch": 1.9609464444727134, + "grad_norm": 2.547409772872925, + "learning_rate": 1e-06, + "loss": 0.3293, + "mean_token_accuracy": 0.8796869516372681, + "num_tokens": 598471175.0, + "step": 15415 + }, + { + "epoch": 1.961073654751304, + "grad_norm": 2.702082395553589, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8964636325836182, + "num_tokens": 598513445.0, + "step": 15416 + }, + { + "epoch": 1.9612008650298944, + "grad_norm": 2.395979166030884, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9023375511169434, + "num_tokens": 598552845.0, + "step": 15417 + }, + { + "epoch": 1.961328075308485, + "grad_norm": 2.2938547134399414, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.904350757598877, + "num_tokens": 598598631.0, + "step": 15418 + }, + { + "epoch": 1.9614552855870753, + "grad_norm": 2.5609138011932373, + "learning_rate": 1e-06, + "loss": 0.3254, + "mean_token_accuracy": 0.8826524615287781, + "num_tokens": 598642597.0, + "step": 15419 + }, + { + "epoch": 1.9615824958656658, + "grad_norm": 2.4903924465179443, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8911892175674438, + "num_tokens": 598684151.0, + "step": 15420 + }, + { + "epoch": 1.9617097061442563, + "grad_norm": 2.906853199005127, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8928530216217041, + "num_tokens": 598715697.0, + "step": 15421 + }, + { + "epoch": 1.9618369164228469, + "grad_norm": 2.506295680999756, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8944610357284546, + "num_tokens": 598755462.0, + "step": 15422 + }, + { + "epoch": 1.9619641267014374, + "grad_norm": 2.69791316986084, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.9001405239105225, + "num_tokens": 598789695.0, + "step": 15423 + }, + { + "epoch": 1.962091336980028, + "grad_norm": 2.4891626834869385, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9024554491043091, + "num_tokens": 598827743.0, + "step": 15424 + }, + { + "epoch": 1.9622185472586184, + "grad_norm": 2.355107307434082, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9052023887634277, + "num_tokens": 598870131.0, + "step": 15425 + }, + { + "epoch": 1.962345757537209, + "grad_norm": 2.8369076251983643, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8810533285140991, + "num_tokens": 598909326.0, + "step": 15426 + }, + { + "epoch": 1.9624729678157995, + "grad_norm": 2.4249162673950195, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8995102047920227, + "num_tokens": 598949405.0, + "step": 15427 + }, + { + "epoch": 1.96260017809439, + "grad_norm": 2.643496513366699, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8993585705757141, + "num_tokens": 598984682.0, + "step": 15428 + }, + { + "epoch": 1.9627273883729806, + "grad_norm": 2.601343870162964, + "learning_rate": 1e-06, + "loss": 0.3425, + "mean_token_accuracy": 0.8771328330039978, + "num_tokens": 599031492.0, + "step": 15429 + }, + { + "epoch": 1.962854598651571, + "grad_norm": 2.4972267150878906, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8977969884872437, + "num_tokens": 599070985.0, + "step": 15430 + }, + { + "epoch": 1.9629818089301616, + "grad_norm": 2.7307679653167725, + "learning_rate": 1e-06, + "loss": 0.3342, + "mean_token_accuracy": 0.8808611631393433, + "num_tokens": 599110699.0, + "step": 15431 + }, + { + "epoch": 1.9631090192087521, + "grad_norm": 2.6064107418060303, + "learning_rate": 1e-06, + "loss": 0.3046, + "mean_token_accuracy": 0.8902252912521362, + "num_tokens": 599150065.0, + "step": 15432 + }, + { + "epoch": 1.9632362294873427, + "grad_norm": 2.3515305519104004, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.892327070236206, + "num_tokens": 599191332.0, + "step": 15433 + }, + { + "epoch": 1.9633634397659332, + "grad_norm": 2.765625, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8997210264205933, + "num_tokens": 599221626.0, + "step": 15434 + }, + { + "epoch": 1.9634906500445237, + "grad_norm": 2.771413803100586, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9027318358421326, + "num_tokens": 599251604.0, + "step": 15435 + }, + { + "epoch": 1.9636178603231143, + "grad_norm": 2.4183731079101562, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8948587775230408, + "num_tokens": 599294170.0, + "step": 15436 + }, + { + "epoch": 1.9637450706017048, + "grad_norm": 2.7412450313568115, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.8870005011558533, + "num_tokens": 599335151.0, + "step": 15437 + }, + { + "epoch": 1.963872280880295, + "grad_norm": 2.5334455966949463, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8944920301437378, + "num_tokens": 599374925.0, + "step": 15438 + }, + { + "epoch": 1.9639994911588856, + "grad_norm": 2.580596923828125, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8938729166984558, + "num_tokens": 599413737.0, + "step": 15439 + }, + { + "epoch": 1.9641267014374761, + "grad_norm": 2.749562978744507, + "learning_rate": 1e-06, + "loss": 0.3504, + "mean_token_accuracy": 0.8746761083602905, + "num_tokens": 599450204.0, + "step": 15440 + }, + { + "epoch": 1.9642539117160667, + "grad_norm": 2.718865156173706, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8931859731674194, + "num_tokens": 599487089.0, + "step": 15441 + }, + { + "epoch": 1.9643811219946572, + "grad_norm": 2.817474365234375, + "learning_rate": 1e-06, + "loss": 0.326, + "mean_token_accuracy": 0.8825241923332214, + "num_tokens": 599526103.0, + "step": 15442 + }, + { + "epoch": 1.9645083322732477, + "grad_norm": 2.8278462886810303, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.89105623960495, + "num_tokens": 599560233.0, + "step": 15443 + }, + { + "epoch": 1.964635542551838, + "grad_norm": 2.602280616760254, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9062871932983398, + "num_tokens": 599593620.0, + "step": 15444 + }, + { + "epoch": 1.9647627528304286, + "grad_norm": 2.4867773056030273, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.896910548210144, + "num_tokens": 599633684.0, + "step": 15445 + }, + { + "epoch": 1.964889963109019, + "grad_norm": 2.7170329093933105, + "learning_rate": 1e-06, + "loss": 0.3189, + "mean_token_accuracy": 0.8865869641304016, + "num_tokens": 599671321.0, + "step": 15446 + }, + { + "epoch": 1.9650171733876096, + "grad_norm": 2.694295883178711, + "learning_rate": 1e-06, + "loss": 0.362, + "mean_token_accuracy": 0.8749250173568726, + "num_tokens": 599715305.0, + "step": 15447 + }, + { + "epoch": 1.9651443836662001, + "grad_norm": 2.6098923683166504, + "learning_rate": 1e-06, + "loss": 0.321, + "mean_token_accuracy": 0.8835210800170898, + "num_tokens": 599753697.0, + "step": 15448 + }, + { + "epoch": 1.9652715939447907, + "grad_norm": 2.608004331588745, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8959304094314575, + "num_tokens": 599789330.0, + "step": 15449 + }, + { + "epoch": 1.9653988042233812, + "grad_norm": 2.4231507778167725, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9026236534118652, + "num_tokens": 599830468.0, + "step": 15450 + }, + { + "epoch": 1.9655260145019717, + "grad_norm": 2.4691617488861084, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8941344022750854, + "num_tokens": 599871676.0, + "step": 15451 + }, + { + "epoch": 1.9656532247805623, + "grad_norm": 2.4802753925323486, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9031463861465454, + "num_tokens": 599910139.0, + "step": 15452 + }, + { + "epoch": 1.9657804350591528, + "grad_norm": 2.542003631591797, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.895612895488739, + "num_tokens": 599950702.0, + "step": 15453 + }, + { + "epoch": 1.9659076453377433, + "grad_norm": 2.576457977294922, + "learning_rate": 1e-06, + "loss": 0.3247, + "mean_token_accuracy": 0.8833838105201721, + "num_tokens": 599992376.0, + "step": 15454 + }, + { + "epoch": 1.9660348556163338, + "grad_norm": 2.7688989639282227, + "learning_rate": 1e-06, + "loss": 0.3311, + "mean_token_accuracy": 0.8818093538284302, + "num_tokens": 600027141.0, + "step": 15455 + }, + { + "epoch": 1.9661620658949244, + "grad_norm": 2.449230909347534, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.899994969367981, + "num_tokens": 600071152.0, + "step": 15456 + }, + { + "epoch": 1.966289276173515, + "grad_norm": 2.61535906791687, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8869544863700867, + "num_tokens": 600111374.0, + "step": 15457 + }, + { + "epoch": 1.9664164864521054, + "grad_norm": 2.596498966217041, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8944664001464844, + "num_tokens": 600149905.0, + "step": 15458 + }, + { + "epoch": 1.966543696730696, + "grad_norm": 2.654860019683838, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8892675042152405, + "num_tokens": 600189579.0, + "step": 15459 + }, + { + "epoch": 1.9666709070092865, + "grad_norm": 2.64093017578125, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8910030126571655, + "num_tokens": 600225414.0, + "step": 15460 + }, + { + "epoch": 1.966798117287877, + "grad_norm": 2.6019954681396484, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9056331515312195, + "num_tokens": 600261535.0, + "step": 15461 + }, + { + "epoch": 1.9669253275664673, + "grad_norm": 2.4123904705047607, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.8972759246826172, + "num_tokens": 600299370.0, + "step": 15462 + }, + { + "epoch": 1.9670525378450578, + "grad_norm": 2.4181947708129883, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8993786573410034, + "num_tokens": 600339007.0, + "step": 15463 + }, + { + "epoch": 1.9671797481236484, + "grad_norm": 2.6449263095855713, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.899907112121582, + "num_tokens": 600373769.0, + "step": 15464 + }, + { + "epoch": 1.967306958402239, + "grad_norm": 2.8517961502075195, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.894298255443573, + "num_tokens": 600406896.0, + "step": 15465 + }, + { + "epoch": 1.9674341686808294, + "grad_norm": 2.45272159576416, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.894315242767334, + "num_tokens": 600449351.0, + "step": 15466 + }, + { + "epoch": 1.96756137895942, + "grad_norm": 2.600055694580078, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8900668621063232, + "num_tokens": 600485896.0, + "step": 15467 + }, + { + "epoch": 1.9676885892380103, + "grad_norm": 2.614008903503418, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9049668312072754, + "num_tokens": 600518666.0, + "step": 15468 + }, + { + "epoch": 1.9678157995166008, + "grad_norm": 2.6312990188598633, + "learning_rate": 1e-06, + "loss": 0.3704, + "mean_token_accuracy": 0.867517352104187, + "num_tokens": 600565492.0, + "step": 15469 + }, + { + "epoch": 1.9679430097951913, + "grad_norm": 2.8523173332214355, + "learning_rate": 1e-06, + "loss": 0.3182, + "mean_token_accuracy": 0.8870232105255127, + "num_tokens": 600604552.0, + "step": 15470 + }, + { + "epoch": 1.9680702200737819, + "grad_norm": 2.4518303871154785, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8890869617462158, + "num_tokens": 600645010.0, + "step": 15471 + }, + { + "epoch": 1.9681974303523724, + "grad_norm": 2.7493374347686768, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8833712935447693, + "num_tokens": 600681693.0, + "step": 15472 + }, + { + "epoch": 1.968324640630963, + "grad_norm": 2.709197998046875, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8900075554847717, + "num_tokens": 600721272.0, + "step": 15473 + }, + { + "epoch": 1.9684518509095534, + "grad_norm": 2.499444007873535, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.8961573839187622, + "num_tokens": 600759076.0, + "step": 15474 + }, + { + "epoch": 1.968579061188144, + "grad_norm": 2.478771209716797, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8972902894020081, + "num_tokens": 600797513.0, + "step": 15475 + }, + { + "epoch": 1.9687062714667345, + "grad_norm": 2.5758755207061768, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8923513889312744, + "num_tokens": 600837199.0, + "step": 15476 + }, + { + "epoch": 1.968833481745325, + "grad_norm": 2.693171977996826, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8926161527633667, + "num_tokens": 600874075.0, + "step": 15477 + }, + { + "epoch": 1.9689606920239155, + "grad_norm": 2.7387163639068604, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.896869957447052, + "num_tokens": 600905513.0, + "step": 15478 + }, + { + "epoch": 1.969087902302506, + "grad_norm": 2.568946361541748, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8898169994354248, + "num_tokens": 600944787.0, + "step": 15479 + }, + { + "epoch": 1.9692151125810966, + "grad_norm": 2.7976534366607666, + "learning_rate": 1e-06, + "loss": 0.3319, + "mean_token_accuracy": 0.8807106614112854, + "num_tokens": 600981834.0, + "step": 15480 + }, + { + "epoch": 1.9693423228596871, + "grad_norm": 2.6707816123962402, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8906946778297424, + "num_tokens": 601019971.0, + "step": 15481 + }, + { + "epoch": 1.9694695331382777, + "grad_norm": 2.560945510864258, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8996628522872925, + "num_tokens": 601059702.0, + "step": 15482 + }, + { + "epoch": 1.9695967434168682, + "grad_norm": 2.435598134994507, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9004262089729309, + "num_tokens": 601101482.0, + "step": 15483 + }, + { + "epoch": 1.9697239536954587, + "grad_norm": 2.527399778366089, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.900511622428894, + "num_tokens": 601142599.0, + "step": 15484 + }, + { + "epoch": 1.9698511639740492, + "grad_norm": 2.3957953453063965, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9078816175460815, + "num_tokens": 601182127.0, + "step": 15485 + }, + { + "epoch": 1.9699783742526398, + "grad_norm": 2.6789262294769287, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8878849744796753, + "num_tokens": 601219452.0, + "step": 15486 + }, + { + "epoch": 1.97010558453123, + "grad_norm": 2.7730493545532227, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8890654444694519, + "num_tokens": 601253235.0, + "step": 15487 + }, + { + "epoch": 1.9702327948098206, + "grad_norm": 2.5567219257354736, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8969238996505737, + "num_tokens": 601289561.0, + "step": 15488 + }, + { + "epoch": 1.9703600050884111, + "grad_norm": 2.6390767097473145, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.901354193687439, + "num_tokens": 601328693.0, + "step": 15489 + }, + { + "epoch": 1.9704872153670017, + "grad_norm": 2.7237918376922607, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8895710110664368, + "num_tokens": 601365726.0, + "step": 15490 + }, + { + "epoch": 1.9706144256455922, + "grad_norm": 2.472752094268799, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9077119827270508, + "num_tokens": 601403384.0, + "step": 15491 + }, + { + "epoch": 1.9707416359241827, + "grad_norm": 2.604188919067383, + "learning_rate": 1e-06, + "loss": 0.3543, + "mean_token_accuracy": 0.8734917640686035, + "num_tokens": 601448541.0, + "step": 15492 + }, + { + "epoch": 1.970868846202773, + "grad_norm": 2.536513566970825, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8845471739768982, + "num_tokens": 601489300.0, + "step": 15493 + }, + { + "epoch": 1.9709960564813636, + "grad_norm": 2.7243645191192627, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8938058614730835, + "num_tokens": 601523134.0, + "step": 15494 + }, + { + "epoch": 1.971123266759954, + "grad_norm": 2.5356171131134033, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8958525061607361, + "num_tokens": 601563714.0, + "step": 15495 + }, + { + "epoch": 1.9712504770385446, + "grad_norm": 2.838315010070801, + "learning_rate": 1e-06, + "loss": 0.3444, + "mean_token_accuracy": 0.8763636350631714, + "num_tokens": 601599390.0, + "step": 15496 + }, + { + "epoch": 1.9713776873171351, + "grad_norm": 2.4395530223846436, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8952076435089111, + "num_tokens": 601639054.0, + "step": 15497 + }, + { + "epoch": 1.9715048975957257, + "grad_norm": 2.475374221801758, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9068165421485901, + "num_tokens": 601678061.0, + "step": 15498 + }, + { + "epoch": 1.9716321078743162, + "grad_norm": 2.5001723766326904, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9015093445777893, + "num_tokens": 601716657.0, + "step": 15499 + }, + { + "epoch": 1.9717593181529067, + "grad_norm": 2.707371950149536, + "learning_rate": 1e-06, + "loss": 0.3298, + "mean_token_accuracy": 0.8812078237533569, + "num_tokens": 601755421.0, + "step": 15500 + }, + { + "epoch": 1.9718865284314973, + "grad_norm": 2.6379001140594482, + "learning_rate": 1e-06, + "loss": 0.2981, + "mean_token_accuracy": 0.8943756818771362, + "num_tokens": 601795020.0, + "step": 15501 + }, + { + "epoch": 1.9720137387100878, + "grad_norm": 2.5781259536743164, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8980159759521484, + "num_tokens": 601833252.0, + "step": 15502 + }, + { + "epoch": 1.9721409489886783, + "grad_norm": 2.609449625015259, + "learning_rate": 1e-06, + "loss": 0.3313, + "mean_token_accuracy": 0.8805398941040039, + "num_tokens": 601874054.0, + "step": 15503 + }, + { + "epoch": 1.9722681592672688, + "grad_norm": 2.4610726833343506, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8971161246299744, + "num_tokens": 601911268.0, + "step": 15504 + }, + { + "epoch": 1.9723953695458594, + "grad_norm": 2.6368494033813477, + "learning_rate": 1e-06, + "loss": 0.3466, + "mean_token_accuracy": 0.8784298300743103, + "num_tokens": 601950802.0, + "step": 15505 + }, + { + "epoch": 1.97252257982445, + "grad_norm": 2.771681070327759, + "learning_rate": 1e-06, + "loss": 0.3545, + "mean_token_accuracy": 0.8728918433189392, + "num_tokens": 601988440.0, + "step": 15506 + }, + { + "epoch": 1.9726497901030404, + "grad_norm": 2.6903069019317627, + "learning_rate": 1e-06, + "loss": 0.365, + "mean_token_accuracy": 0.8699931502342224, + "num_tokens": 602029198.0, + "step": 15507 + }, + { + "epoch": 1.972777000381631, + "grad_norm": 2.559213399887085, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8862311244010925, + "num_tokens": 602070828.0, + "step": 15508 + }, + { + "epoch": 1.9729042106602215, + "grad_norm": 2.708526611328125, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8864941000938416, + "num_tokens": 602110155.0, + "step": 15509 + }, + { + "epoch": 1.973031420938812, + "grad_norm": 2.8337130546569824, + "learning_rate": 1e-06, + "loss": 0.3283, + "mean_token_accuracy": 0.8833143711090088, + "num_tokens": 602145910.0, + "step": 15510 + }, + { + "epoch": 1.9731586312174023, + "grad_norm": 2.7014927864074707, + "learning_rate": 1e-06, + "loss": 0.3331, + "mean_token_accuracy": 0.8836727738380432, + "num_tokens": 602185383.0, + "step": 15511 + }, + { + "epoch": 1.9732858414959928, + "grad_norm": 2.489103078842163, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9008144736289978, + "num_tokens": 602224520.0, + "step": 15512 + }, + { + "epoch": 1.9734130517745834, + "grad_norm": 2.6235594749450684, + "learning_rate": 1e-06, + "loss": 0.3142, + "mean_token_accuracy": 0.8860388994216919, + "num_tokens": 602262766.0, + "step": 15513 + }, + { + "epoch": 1.973540262053174, + "grad_norm": 2.7305240631103516, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8886147737503052, + "num_tokens": 602298944.0, + "step": 15514 + }, + { + "epoch": 1.9736674723317644, + "grad_norm": 2.6008355617523193, + "learning_rate": 1e-06, + "loss": 0.3135, + "mean_token_accuracy": 0.8893474340438843, + "num_tokens": 602339897.0, + "step": 15515 + }, + { + "epoch": 1.973794682610355, + "grad_norm": 2.5085692405700684, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8890128135681152, + "num_tokens": 602383537.0, + "step": 15516 + }, + { + "epoch": 1.9739218928889453, + "grad_norm": 2.347907781600952, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9044363498687744, + "num_tokens": 602424480.0, + "step": 15517 + }, + { + "epoch": 1.9740491031675358, + "grad_norm": 2.3304810523986816, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8908470869064331, + "num_tokens": 602467607.0, + "step": 15518 + }, + { + "epoch": 1.9741763134461263, + "grad_norm": 3.0356550216674805, + "learning_rate": 1e-06, + "loss": 0.3215, + "mean_token_accuracy": 0.885219931602478, + "num_tokens": 602497563.0, + "step": 15519 + }, + { + "epoch": 1.9743035237247168, + "grad_norm": 2.727756977081299, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8919821977615356, + "num_tokens": 602531988.0, + "step": 15520 + }, + { + "epoch": 1.9744307340033074, + "grad_norm": 2.6406919956207275, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8939946293830872, + "num_tokens": 602568990.0, + "step": 15521 + }, + { + "epoch": 1.974557944281898, + "grad_norm": 2.608832597732544, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8881490230560303, + "num_tokens": 602606384.0, + "step": 15522 + }, + { + "epoch": 1.9746851545604884, + "grad_norm": 2.4989709854125977, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9022306203842163, + "num_tokens": 602643042.0, + "step": 15523 + }, + { + "epoch": 1.974812364839079, + "grad_norm": 2.7771992683410645, + "learning_rate": 1e-06, + "loss": 0.3358, + "mean_token_accuracy": 0.8822802901268005, + "num_tokens": 602679203.0, + "step": 15524 + }, + { + "epoch": 1.9749395751176695, + "grad_norm": 2.801119327545166, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8850837349891663, + "num_tokens": 602712160.0, + "step": 15525 + }, + { + "epoch": 1.97506678539626, + "grad_norm": 2.678853750228882, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8845652341842651, + "num_tokens": 602749665.0, + "step": 15526 + }, + { + "epoch": 1.9751939956748505, + "grad_norm": 2.8721954822540283, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8878027200698853, + "num_tokens": 602779787.0, + "step": 15527 + }, + { + "epoch": 1.975321205953441, + "grad_norm": 3.0727224349975586, + "learning_rate": 1e-06, + "loss": 0.3322, + "mean_token_accuracy": 0.8792809247970581, + "num_tokens": 602811312.0, + "step": 15528 + }, + { + "epoch": 1.9754484162320316, + "grad_norm": 2.6462020874023438, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8953920006752014, + "num_tokens": 602849769.0, + "step": 15529 + }, + { + "epoch": 1.9755756265106221, + "grad_norm": 2.8466527462005615, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8851678371429443, + "num_tokens": 602887841.0, + "step": 15530 + }, + { + "epoch": 1.9757028367892127, + "grad_norm": 2.5210721492767334, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8869491815567017, + "num_tokens": 602930738.0, + "step": 15531 + }, + { + "epoch": 1.9758300470678032, + "grad_norm": 2.54537296295166, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.8997520208358765, + "num_tokens": 602967153.0, + "step": 15532 + }, + { + "epoch": 1.9759572573463937, + "grad_norm": 2.6535239219665527, + "learning_rate": 1e-06, + "loss": 0.3005, + "mean_token_accuracy": 0.893203616142273, + "num_tokens": 603005017.0, + "step": 15533 + }, + { + "epoch": 1.9760844676249842, + "grad_norm": 2.456857919692993, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8993794322013855, + "num_tokens": 603045348.0, + "step": 15534 + }, + { + "epoch": 1.9762116779035748, + "grad_norm": 2.591986656188965, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.895560622215271, + "num_tokens": 603083318.0, + "step": 15535 + }, + { + "epoch": 1.976338888182165, + "grad_norm": 2.9092013835906982, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8924504518508911, + "num_tokens": 603116418.0, + "step": 15536 + }, + { + "epoch": 1.9764660984607556, + "grad_norm": 2.344935894012451, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.899297297000885, + "num_tokens": 603162156.0, + "step": 15537 + }, + { + "epoch": 1.9765933087393461, + "grad_norm": 2.321185827255249, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9009779691696167, + "num_tokens": 603208318.0, + "step": 15538 + }, + { + "epoch": 1.9767205190179367, + "grad_norm": 2.526047945022583, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8969281911849976, + "num_tokens": 603243819.0, + "step": 15539 + }, + { + "epoch": 1.9768477292965272, + "grad_norm": 2.3918967247009277, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9009683132171631, + "num_tokens": 603285031.0, + "step": 15540 + }, + { + "epoch": 1.9769749395751177, + "grad_norm": 2.6365575790405273, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8900794982910156, + "num_tokens": 603320913.0, + "step": 15541 + }, + { + "epoch": 1.977102149853708, + "grad_norm": 2.571007013320923, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.9000480771064758, + "num_tokens": 603361624.0, + "step": 15542 + }, + { + "epoch": 1.9772293601322986, + "grad_norm": 2.708311080932617, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9041187763214111, + "num_tokens": 603391614.0, + "step": 15543 + }, + { + "epoch": 1.977356570410889, + "grad_norm": 2.6983320713043213, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8993374109268188, + "num_tokens": 603427356.0, + "step": 15544 + }, + { + "epoch": 1.9774837806894796, + "grad_norm": 2.654472589492798, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9044198989868164, + "num_tokens": 603459316.0, + "step": 15545 + }, + { + "epoch": 1.9776109909680701, + "grad_norm": 2.558727264404297, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8900018334388733, + "num_tokens": 603501191.0, + "step": 15546 + }, + { + "epoch": 1.9777382012466607, + "grad_norm": 2.783702850341797, + "learning_rate": 1e-06, + "loss": 0.3866, + "mean_token_accuracy": 0.8622308969497681, + "num_tokens": 603543401.0, + "step": 15547 + }, + { + "epoch": 1.9778654115252512, + "grad_norm": 2.4732887744903564, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.89774090051651, + "num_tokens": 603583271.0, + "step": 15548 + }, + { + "epoch": 1.9779926218038417, + "grad_norm": 2.8408610820770264, + "learning_rate": 1e-06, + "loss": 0.3076, + "mean_token_accuracy": 0.8871912956237793, + "num_tokens": 603619118.0, + "step": 15549 + }, + { + "epoch": 1.9781198320824323, + "grad_norm": 2.6304521560668945, + "learning_rate": 1e-06, + "loss": 0.359, + "mean_token_accuracy": 0.8710915446281433, + "num_tokens": 603662911.0, + "step": 15550 + }, + { + "epoch": 1.9782470423610228, + "grad_norm": 2.7694239616394043, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8935865759849548, + "num_tokens": 603699451.0, + "step": 15551 + }, + { + "epoch": 1.9783742526396133, + "grad_norm": 2.5569000244140625, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8980377912521362, + "num_tokens": 603736533.0, + "step": 15552 + }, + { + "epoch": 1.9785014629182038, + "grad_norm": 2.926604747772217, + "learning_rate": 1e-06, + "loss": 0.3572, + "mean_token_accuracy": 0.8744099736213684, + "num_tokens": 603772540.0, + "step": 15553 + }, + { + "epoch": 1.9786286731967944, + "grad_norm": 2.589601993560791, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.89815354347229, + "num_tokens": 603807432.0, + "step": 15554 + }, + { + "epoch": 1.978755883475385, + "grad_norm": 2.5439701080322266, + "learning_rate": 1e-06, + "loss": 0.3248, + "mean_token_accuracy": 0.881603479385376, + "num_tokens": 603848657.0, + "step": 15555 + }, + { + "epoch": 1.9788830937539754, + "grad_norm": 2.547206163406372, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8840879797935486, + "num_tokens": 603891767.0, + "step": 15556 + }, + { + "epoch": 1.979010304032566, + "grad_norm": 2.7797927856445312, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8869463205337524, + "num_tokens": 603930912.0, + "step": 15557 + }, + { + "epoch": 1.9791375143111565, + "grad_norm": 2.722341775894165, + "learning_rate": 1e-06, + "loss": 0.3083, + "mean_token_accuracy": 0.8877221941947937, + "num_tokens": 603969645.0, + "step": 15558 + }, + { + "epoch": 1.979264724589747, + "grad_norm": 2.531881093978882, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.902327835559845, + "num_tokens": 604008553.0, + "step": 15559 + }, + { + "epoch": 1.9793919348683373, + "grad_norm": 2.5678534507751465, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8875118494033813, + "num_tokens": 604055838.0, + "step": 15560 + }, + { + "epoch": 1.9795191451469278, + "grad_norm": 2.692631483078003, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8955969214439392, + "num_tokens": 604091158.0, + "step": 15561 + }, + { + "epoch": 1.9796463554255184, + "grad_norm": 2.498662233352661, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8947242498397827, + "num_tokens": 604130690.0, + "step": 15562 + }, + { + "epoch": 1.979773565704109, + "grad_norm": 2.410264492034912, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.9019379615783691, + "num_tokens": 604172488.0, + "step": 15563 + }, + { + "epoch": 1.9799007759826994, + "grad_norm": 2.5816617012023926, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8876502513885498, + "num_tokens": 604211796.0, + "step": 15564 + }, + { + "epoch": 1.98002798626129, + "grad_norm": 2.836700916290283, + "learning_rate": 1e-06, + "loss": 0.3376, + "mean_token_accuracy": 0.8778939247131348, + "num_tokens": 604248136.0, + "step": 15565 + }, + { + "epoch": 1.9801551965398803, + "grad_norm": 2.6215240955352783, + "learning_rate": 1e-06, + "loss": 0.3104, + "mean_token_accuracy": 0.8839412927627563, + "num_tokens": 604282177.0, + "step": 15566 + }, + { + "epoch": 1.9802824068184708, + "grad_norm": 2.75626277923584, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8908871412277222, + "num_tokens": 604319457.0, + "step": 15567 + }, + { + "epoch": 1.9804096170970613, + "grad_norm": 2.6181640625, + "learning_rate": 1e-06, + "loss": 0.3102, + "mean_token_accuracy": 0.8895204067230225, + "num_tokens": 604357295.0, + "step": 15568 + }, + { + "epoch": 1.9805368273756518, + "grad_norm": 2.8737783432006836, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8968281745910645, + "num_tokens": 604387472.0, + "step": 15569 + }, + { + "epoch": 1.9806640376542424, + "grad_norm": 2.8913931846618652, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.889440655708313, + "num_tokens": 604421203.0, + "step": 15570 + }, + { + "epoch": 1.980791247932833, + "grad_norm": 2.4276862144470215, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9146198034286499, + "num_tokens": 604455638.0, + "step": 15571 + }, + { + "epoch": 1.9809184582114234, + "grad_norm": 2.8011860847473145, + "learning_rate": 1e-06, + "loss": 0.3128, + "mean_token_accuracy": 0.8882358074188232, + "num_tokens": 604490268.0, + "step": 15572 + }, + { + "epoch": 1.981045668490014, + "grad_norm": 2.787928581237793, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8923911452293396, + "num_tokens": 604526243.0, + "step": 15573 + }, + { + "epoch": 1.9811728787686045, + "grad_norm": 2.4655568599700928, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9091531038284302, + "num_tokens": 604558543.0, + "step": 15574 + }, + { + "epoch": 1.981300089047195, + "grad_norm": 2.8732097148895264, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8939100503921509, + "num_tokens": 604593449.0, + "step": 15575 + }, + { + "epoch": 1.9814272993257855, + "grad_norm": 2.2997395992279053, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9044535160064697, + "num_tokens": 604635364.0, + "step": 15576 + }, + { + "epoch": 1.981554509604376, + "grad_norm": 2.728187084197998, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8933972716331482, + "num_tokens": 604671381.0, + "step": 15577 + }, + { + "epoch": 1.9816817198829666, + "grad_norm": 2.978269338607788, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.890522301197052, + "num_tokens": 604701761.0, + "step": 15578 + }, + { + "epoch": 1.9818089301615571, + "grad_norm": 2.4560048580169678, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9033692479133606, + "num_tokens": 604741346.0, + "step": 15579 + }, + { + "epoch": 1.9819361404401477, + "grad_norm": 2.7019386291503906, + "learning_rate": 1e-06, + "loss": 0.3131, + "mean_token_accuracy": 0.8885436058044434, + "num_tokens": 604775866.0, + "step": 15580 + }, + { + "epoch": 1.9820633507187382, + "grad_norm": 2.7899222373962402, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9030323028564453, + "num_tokens": 604806910.0, + "step": 15581 + }, + { + "epoch": 1.9821905609973287, + "grad_norm": 2.921159029006958, + "learning_rate": 1e-06, + "loss": 0.351, + "mean_token_accuracy": 0.8731756210327148, + "num_tokens": 604839456.0, + "step": 15582 + }, + { + "epoch": 1.9823177712759192, + "grad_norm": 2.382997512817383, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9014585018157959, + "num_tokens": 604878450.0, + "step": 15583 + }, + { + "epoch": 1.9824449815545098, + "grad_norm": 2.715327739715576, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8921163082122803, + "num_tokens": 604912997.0, + "step": 15584 + }, + { + "epoch": 1.9825721918331, + "grad_norm": 2.6225016117095947, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.892304003238678, + "num_tokens": 604948289.0, + "step": 15585 + }, + { + "epoch": 1.9826994021116906, + "grad_norm": 2.4225196838378906, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8932672739028931, + "num_tokens": 604990154.0, + "step": 15586 + }, + { + "epoch": 1.9828266123902811, + "grad_norm": 2.5656704902648926, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8885493874549866, + "num_tokens": 605029230.0, + "step": 15587 + }, + { + "epoch": 1.9829538226688717, + "grad_norm": 2.375056028366089, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.900935709476471, + "num_tokens": 605071432.0, + "step": 15588 + }, + { + "epoch": 1.9830810329474622, + "grad_norm": 2.3692517280578613, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8852173686027527, + "num_tokens": 605115781.0, + "step": 15589 + }, + { + "epoch": 1.9832082432260527, + "grad_norm": 10.47694206237793, + "learning_rate": 1e-06, + "loss": 0.3214, + "mean_token_accuracy": 0.882827639579773, + "num_tokens": 605156153.0, + "step": 15590 + }, + { + "epoch": 1.983335453504643, + "grad_norm": 2.420116662979126, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8929349184036255, + "num_tokens": 605198673.0, + "step": 15591 + }, + { + "epoch": 1.9834626637832335, + "grad_norm": 2.496281385421753, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8973097801208496, + "num_tokens": 605236161.0, + "step": 15592 + }, + { + "epoch": 1.983589874061824, + "grad_norm": 2.4649341106414795, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.890504002571106, + "num_tokens": 605276978.0, + "step": 15593 + }, + { + "epoch": 1.9837170843404146, + "grad_norm": 2.846470355987549, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8923934698104858, + "num_tokens": 605311267.0, + "step": 15594 + }, + { + "epoch": 1.9838442946190051, + "grad_norm": 2.505903959274292, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9011025428771973, + "num_tokens": 605348681.0, + "step": 15595 + }, + { + "epoch": 1.9839715048975957, + "grad_norm": 2.6729888916015625, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9056519269943237, + "num_tokens": 605381181.0, + "step": 15596 + }, + { + "epoch": 1.9840987151761862, + "grad_norm": 2.592820167541504, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9038174748420715, + "num_tokens": 605416167.0, + "step": 15597 + }, + { + "epoch": 1.9842259254547767, + "grad_norm": 2.6603450775146484, + "learning_rate": 1e-06, + "loss": 0.3188, + "mean_token_accuracy": 0.8851189017295837, + "num_tokens": 605457390.0, + "step": 15598 + }, + { + "epoch": 1.9843531357333672, + "grad_norm": 2.6115188598632812, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8843136429786682, + "num_tokens": 605499435.0, + "step": 15599 + }, + { + "epoch": 1.9844803460119578, + "grad_norm": 2.7015302181243896, + "learning_rate": 1e-06, + "loss": 0.328, + "mean_token_accuracy": 0.8847752213478088, + "num_tokens": 605538354.0, + "step": 15600 + }, + { + "epoch": 1.9846075562905483, + "grad_norm": 2.6098358631134033, + "learning_rate": 1e-06, + "loss": 0.3403, + "mean_token_accuracy": 0.8778033256530762, + "num_tokens": 605582455.0, + "step": 15601 + }, + { + "epoch": 1.9847347665691388, + "grad_norm": 2.645995616912842, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9078490734100342, + "num_tokens": 605612082.0, + "step": 15602 + }, + { + "epoch": 1.9848619768477294, + "grad_norm": 2.479524612426758, + "learning_rate": 1e-06, + "loss": 0.314, + "mean_token_accuracy": 0.8873496651649475, + "num_tokens": 605655494.0, + "step": 15603 + }, + { + "epoch": 1.9849891871263199, + "grad_norm": 2.544476270675659, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8940496444702148, + "num_tokens": 605695707.0, + "step": 15604 + }, + { + "epoch": 1.9851163974049104, + "grad_norm": 2.5847389698028564, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8859361410140991, + "num_tokens": 605737201.0, + "step": 15605 + }, + { + "epoch": 1.985243607683501, + "grad_norm": 2.6416635513305664, + "learning_rate": 1e-06, + "loss": 0.3115, + "mean_token_accuracy": 0.8864156007766724, + "num_tokens": 605775981.0, + "step": 15606 + }, + { + "epoch": 1.9853708179620915, + "grad_norm": 2.6315441131591797, + "learning_rate": 1e-06, + "loss": 0.3183, + "mean_token_accuracy": 0.8849920630455017, + "num_tokens": 605815539.0, + "step": 15607 + }, + { + "epoch": 1.985498028240682, + "grad_norm": 2.5964202880859375, + "learning_rate": 1e-06, + "loss": 0.3089, + "mean_token_accuracy": 0.8895019292831421, + "num_tokens": 605855680.0, + "step": 15608 + }, + { + "epoch": 1.9856252385192723, + "grad_norm": 2.4040119647979736, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.896886944770813, + "num_tokens": 605899332.0, + "step": 15609 + }, + { + "epoch": 1.9857524487978628, + "grad_norm": 2.583411693572998, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8994983434677124, + "num_tokens": 605936052.0, + "step": 15610 + }, + { + "epoch": 1.9858796590764534, + "grad_norm": 2.6053519248962402, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8913244009017944, + "num_tokens": 605973934.0, + "step": 15611 + }, + { + "epoch": 1.986006869355044, + "grad_norm": 2.5369949340820312, + "learning_rate": 1e-06, + "loss": 0.3121, + "mean_token_accuracy": 0.888084352016449, + "num_tokens": 606015238.0, + "step": 15612 + }, + { + "epoch": 1.9861340796336344, + "grad_norm": 2.499220609664917, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9002076387405396, + "num_tokens": 606052356.0, + "step": 15613 + }, + { + "epoch": 1.986261289912225, + "grad_norm": 2.675902843475342, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8912472128868103, + "num_tokens": 606088279.0, + "step": 15614 + }, + { + "epoch": 1.9863885001908153, + "grad_norm": 2.6107017993927, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8956232070922852, + "num_tokens": 606124225.0, + "step": 15615 + }, + { + "epoch": 1.9865157104694058, + "grad_norm": 2.4414494037628174, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9097332954406738, + "num_tokens": 606160824.0, + "step": 15616 + }, + { + "epoch": 1.9866429207479963, + "grad_norm": 2.5933806896209717, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8869609832763672, + "num_tokens": 606197699.0, + "step": 15617 + }, + { + "epoch": 1.9867701310265868, + "grad_norm": 2.712728500366211, + "learning_rate": 1e-06, + "loss": 0.3378, + "mean_token_accuracy": 0.8798079490661621, + "num_tokens": 606241355.0, + "step": 15618 + }, + { + "epoch": 1.9868973413051774, + "grad_norm": 2.554373264312744, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.8970813751220703, + "num_tokens": 606281199.0, + "step": 15619 + }, + { + "epoch": 1.987024551583768, + "grad_norm": 2.4101204872131348, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8966245055198669, + "num_tokens": 606322141.0, + "step": 15620 + }, + { + "epoch": 1.9871517618623584, + "grad_norm": 2.634282112121582, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8930895328521729, + "num_tokens": 606358656.0, + "step": 15621 + }, + { + "epoch": 1.987278972140949, + "grad_norm": 2.760225534439087, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.892220139503479, + "num_tokens": 606389714.0, + "step": 15622 + }, + { + "epoch": 1.9874061824195395, + "grad_norm": 2.4121603965759277, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8918145895004272, + "num_tokens": 606434993.0, + "step": 15623 + }, + { + "epoch": 1.98753339269813, + "grad_norm": 2.6839263439178467, + "learning_rate": 1e-06, + "loss": 0.3367, + "mean_token_accuracy": 0.8809361457824707, + "num_tokens": 606476221.0, + "step": 15624 + }, + { + "epoch": 1.9876606029767205, + "grad_norm": 2.559554100036621, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8892435431480408, + "num_tokens": 606516497.0, + "step": 15625 + }, + { + "epoch": 1.987787813255311, + "grad_norm": 2.609100103378296, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.902191162109375, + "num_tokens": 606548661.0, + "step": 15626 + }, + { + "epoch": 1.9879150235339016, + "grad_norm": 2.516000270843506, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8924524784088135, + "num_tokens": 606592416.0, + "step": 15627 + }, + { + "epoch": 1.9880422338124921, + "grad_norm": 2.3993940353393555, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9070582389831543, + "num_tokens": 606631543.0, + "step": 15628 + }, + { + "epoch": 1.9881694440910826, + "grad_norm": 2.4684503078460693, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8911963701248169, + "num_tokens": 606674789.0, + "step": 15629 + }, + { + "epoch": 1.9882966543696732, + "grad_norm": 2.7217326164245605, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.88847815990448, + "num_tokens": 606710118.0, + "step": 15630 + }, + { + "epoch": 1.9884238646482637, + "grad_norm": 2.400684118270874, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.889971911907196, + "num_tokens": 606752597.0, + "step": 15631 + }, + { + "epoch": 1.9885510749268542, + "grad_norm": 2.4534380435943604, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8944724798202515, + "num_tokens": 606795537.0, + "step": 15632 + }, + { + "epoch": 1.9886782852054448, + "grad_norm": 2.7272908687591553, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8928327560424805, + "num_tokens": 606832131.0, + "step": 15633 + }, + { + "epoch": 1.988805495484035, + "grad_norm": 2.667252540588379, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8958157300949097, + "num_tokens": 606867412.0, + "step": 15634 + }, + { + "epoch": 1.9889327057626256, + "grad_norm": 2.52577805519104, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8916206359863281, + "num_tokens": 606907739.0, + "step": 15635 + }, + { + "epoch": 1.9890599160412161, + "grad_norm": 2.6115591526031494, + "learning_rate": 1e-06, + "loss": 0.3457, + "mean_token_accuracy": 0.8748073577880859, + "num_tokens": 606948661.0, + "step": 15636 + }, + { + "epoch": 1.9891871263198067, + "grad_norm": 2.5285985469818115, + "learning_rate": 1e-06, + "loss": 0.3429, + "mean_token_accuracy": 0.880865216255188, + "num_tokens": 606995515.0, + "step": 15637 + }, + { + "epoch": 1.9893143365983972, + "grad_norm": 2.517671823501587, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8962551355361938, + "num_tokens": 607032158.0, + "step": 15638 + }, + { + "epoch": 1.9894415468769877, + "grad_norm": 2.7003228664398193, + "learning_rate": 1e-06, + "loss": 0.3221, + "mean_token_accuracy": 0.8874921798706055, + "num_tokens": 607070381.0, + "step": 15639 + }, + { + "epoch": 1.989568757155578, + "grad_norm": 2.479600667953491, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8940808176994324, + "num_tokens": 607108450.0, + "step": 15640 + }, + { + "epoch": 1.9896959674341685, + "grad_norm": 2.748586893081665, + "learning_rate": 1e-06, + "loss": 0.3191, + "mean_token_accuracy": 0.8850208520889282, + "num_tokens": 607144807.0, + "step": 15641 + }, + { + "epoch": 1.989823177712759, + "grad_norm": 2.474139451980591, + "learning_rate": 1e-06, + "loss": 0.3049, + "mean_token_accuracy": 0.8909986019134521, + "num_tokens": 607183560.0, + "step": 15642 + }, + { + "epoch": 1.9899503879913496, + "grad_norm": 2.7006235122680664, + "learning_rate": 1e-06, + "loss": 0.3208, + "mean_token_accuracy": 0.8856251835823059, + "num_tokens": 607220092.0, + "step": 15643 + }, + { + "epoch": 1.9900775982699401, + "grad_norm": 2.373255491256714, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8950614929199219, + "num_tokens": 607263497.0, + "step": 15644 + }, + { + "epoch": 1.9902048085485307, + "grad_norm": 2.3769679069519043, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8961988687515259, + "num_tokens": 607306054.0, + "step": 15645 + }, + { + "epoch": 1.9903320188271212, + "grad_norm": 2.5941896438598633, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8895745277404785, + "num_tokens": 607345681.0, + "step": 15646 + }, + { + "epoch": 1.9904592291057117, + "grad_norm": 2.467947244644165, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.8876352906227112, + "num_tokens": 607387334.0, + "step": 15647 + }, + { + "epoch": 1.9905864393843022, + "grad_norm": 2.4574520587921143, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8888489007949829, + "num_tokens": 607428761.0, + "step": 15648 + }, + { + "epoch": 1.9907136496628928, + "grad_norm": 2.4542901515960693, + "learning_rate": 1e-06, + "loss": 0.3192, + "mean_token_accuracy": 0.8830553293228149, + "num_tokens": 607472119.0, + "step": 15649 + }, + { + "epoch": 1.9908408599414833, + "grad_norm": 2.641749382019043, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8919367790222168, + "num_tokens": 607511063.0, + "step": 15650 + }, + { + "epoch": 1.9909680702200738, + "grad_norm": 2.629671096801758, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8983185291290283, + "num_tokens": 607548965.0, + "step": 15651 + }, + { + "epoch": 1.9910952804986644, + "grad_norm": 2.724034309387207, + "learning_rate": 1e-06, + "loss": 0.3152, + "mean_token_accuracy": 0.8857153058052063, + "num_tokens": 607585877.0, + "step": 15652 + }, + { + "epoch": 1.9912224907772549, + "grad_norm": 2.593249797821045, + "learning_rate": 1e-06, + "loss": 0.3241, + "mean_token_accuracy": 0.8853261470794678, + "num_tokens": 607624941.0, + "step": 15653 + }, + { + "epoch": 1.9913497010558454, + "grad_norm": 2.3678693771362305, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8908778429031372, + "num_tokens": 607669640.0, + "step": 15654 + }, + { + "epoch": 1.991476911334436, + "grad_norm": 2.62766695022583, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8979088068008423, + "num_tokens": 607706325.0, + "step": 15655 + }, + { + "epoch": 1.9916041216130265, + "grad_norm": 2.4119694232940674, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8984659910202026, + "num_tokens": 607745388.0, + "step": 15656 + }, + { + "epoch": 1.991731331891617, + "grad_norm": 2.7593836784362793, + "learning_rate": 1e-06, + "loss": 0.3234, + "mean_token_accuracy": 0.885240912437439, + "num_tokens": 607784220.0, + "step": 15657 + }, + { + "epoch": 1.9918585421702073, + "grad_norm": 2.533189296722412, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9097532033920288, + "num_tokens": 607819020.0, + "step": 15658 + }, + { + "epoch": 1.9919857524487978, + "grad_norm": 2.873081684112549, + "learning_rate": 1e-06, + "loss": 0.3229, + "mean_token_accuracy": 0.885431706905365, + "num_tokens": 607849401.0, + "step": 15659 + }, + { + "epoch": 1.9921129627273884, + "grad_norm": 2.5774128437042236, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8954693675041199, + "num_tokens": 607885269.0, + "step": 15660 + }, + { + "epoch": 1.9922401730059789, + "grad_norm": 2.9079272747039795, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8880137205123901, + "num_tokens": 607915519.0, + "step": 15661 + }, + { + "epoch": 1.9923673832845694, + "grad_norm": 2.364645004272461, + "learning_rate": 1e-06, + "loss": 0.3285, + "mean_token_accuracy": 0.884451687335968, + "num_tokens": 607965604.0, + "step": 15662 + }, + { + "epoch": 1.99249459356316, + "grad_norm": 2.5256638526916504, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.8990317583084106, + "num_tokens": 608004520.0, + "step": 15663 + }, + { + "epoch": 1.9926218038417502, + "grad_norm": 2.7532002925872803, + "learning_rate": 1e-06, + "loss": 0.3449, + "mean_token_accuracy": 0.8720501065254211, + "num_tokens": 608043252.0, + "step": 15664 + }, + { + "epoch": 1.9927490141203408, + "grad_norm": 2.931239604949951, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8920398950576782, + "num_tokens": 608073101.0, + "step": 15665 + }, + { + "epoch": 1.9928762243989313, + "grad_norm": 2.627779960632324, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8939775228500366, + "num_tokens": 608112355.0, + "step": 15666 + }, + { + "epoch": 1.9930034346775218, + "grad_norm": 2.425039291381836, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9047070145606995, + "num_tokens": 608151418.0, + "step": 15667 + }, + { + "epoch": 1.9931306449561124, + "grad_norm": 2.687736749649048, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8974907398223877, + "num_tokens": 608187755.0, + "step": 15668 + }, + { + "epoch": 1.993257855234703, + "grad_norm": 2.7030279636383057, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8858880400657654, + "num_tokens": 608227748.0, + "step": 15669 + }, + { + "epoch": 1.9933850655132934, + "grad_norm": 3.1647040843963623, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.906922459602356, + "num_tokens": 608264425.0, + "step": 15670 + }, + { + "epoch": 1.993512275791884, + "grad_norm": 2.6375274658203125, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8928526043891907, + "num_tokens": 608300579.0, + "step": 15671 + }, + { + "epoch": 1.9936394860704745, + "grad_norm": 2.567668914794922, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8978658318519592, + "num_tokens": 608339008.0, + "step": 15672 + }, + { + "epoch": 1.993766696349065, + "grad_norm": 2.666240930557251, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8930898904800415, + "num_tokens": 608374178.0, + "step": 15673 + }, + { + "epoch": 1.9938939066276555, + "grad_norm": 2.4858102798461914, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9068690538406372, + "num_tokens": 608412782.0, + "step": 15674 + }, + { + "epoch": 1.994021116906246, + "grad_norm": 2.4773106575012207, + "learning_rate": 1e-06, + "loss": 0.3126, + "mean_token_accuracy": 0.886314868927002, + "num_tokens": 608452487.0, + "step": 15675 + }, + { + "epoch": 1.9941483271848366, + "grad_norm": 2.4691579341888428, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8986443281173706, + "num_tokens": 608492159.0, + "step": 15676 + }, + { + "epoch": 1.9942755374634271, + "grad_norm": 2.640033006668091, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8886810541152954, + "num_tokens": 608531370.0, + "step": 15677 + }, + { + "epoch": 1.9944027477420176, + "grad_norm": 2.6739320755004883, + "learning_rate": 1e-06, + "loss": 0.3243, + "mean_token_accuracy": 0.884795606136322, + "num_tokens": 608573357.0, + "step": 15678 + }, + { + "epoch": 1.9945299580206082, + "grad_norm": 2.407578468322754, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8894414901733398, + "num_tokens": 608618116.0, + "step": 15679 + }, + { + "epoch": 1.9946571682991987, + "grad_norm": 2.6886062622070312, + "learning_rate": 1e-06, + "loss": 0.3257, + "mean_token_accuracy": 0.8841365575790405, + "num_tokens": 608656749.0, + "step": 15680 + }, + { + "epoch": 1.9947843785777892, + "grad_norm": 2.3558106422424316, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8871508240699768, + "num_tokens": 608706634.0, + "step": 15681 + }, + { + "epoch": 1.9949115888563798, + "grad_norm": 2.468393325805664, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8966233730316162, + "num_tokens": 608744388.0, + "step": 15682 + }, + { + "epoch": 1.99503879913497, + "grad_norm": 2.483720541000366, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8898241519927979, + "num_tokens": 608789628.0, + "step": 15683 + }, + { + "epoch": 1.9951660094135606, + "grad_norm": 2.742281198501587, + "learning_rate": 1e-06, + "loss": 0.3181, + "mean_token_accuracy": 0.8873329162597656, + "num_tokens": 608827268.0, + "step": 15684 + }, + { + "epoch": 1.9952932196921511, + "grad_norm": 2.4978835582733154, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8932099342346191, + "num_tokens": 608869728.0, + "step": 15685 + }, + { + "epoch": 1.9954204299707416, + "grad_norm": 2.5329203605651855, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8877140283584595, + "num_tokens": 608911790.0, + "step": 15686 + }, + { + "epoch": 1.9955476402493322, + "grad_norm": 2.66064190864563, + "learning_rate": 1e-06, + "loss": 0.3357, + "mean_token_accuracy": 0.8798490762710571, + "num_tokens": 608952352.0, + "step": 15687 + }, + { + "epoch": 1.9956748505279227, + "grad_norm": 2.265773057937622, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.9002277851104736, + "num_tokens": 608994228.0, + "step": 15688 + }, + { + "epoch": 1.995802060806513, + "grad_norm": 2.6809499263763428, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8949865102767944, + "num_tokens": 609029379.0, + "step": 15689 + }, + { + "epoch": 1.9959292710851035, + "grad_norm": 2.588223457336426, + "learning_rate": 1e-06, + "loss": 0.341, + "mean_token_accuracy": 0.8762044310569763, + "num_tokens": 609071456.0, + "step": 15690 + }, + { + "epoch": 1.996056481363694, + "grad_norm": 2.3327255249023438, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9077052474021912, + "num_tokens": 609113636.0, + "step": 15691 + }, + { + "epoch": 1.9961836916422846, + "grad_norm": 2.644134759902954, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8916559219360352, + "num_tokens": 609153250.0, + "step": 15692 + }, + { + "epoch": 1.9963109019208751, + "grad_norm": 2.333604335784912, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9079990386962891, + "num_tokens": 609194440.0, + "step": 15693 + }, + { + "epoch": 1.9964381121994657, + "grad_norm": 2.422945261001587, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8869297504425049, + "num_tokens": 609237269.0, + "step": 15694 + }, + { + "epoch": 1.9965653224780562, + "grad_norm": 2.583674430847168, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8916898965835571, + "num_tokens": 609276337.0, + "step": 15695 + }, + { + "epoch": 1.9966925327566467, + "grad_norm": 2.3692445755004883, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9015300869941711, + "num_tokens": 609315681.0, + "step": 15696 + }, + { + "epoch": 1.9968197430352372, + "grad_norm": 2.627908706665039, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8973817229270935, + "num_tokens": 609352012.0, + "step": 15697 + }, + { + "epoch": 1.9969469533138278, + "grad_norm": 2.505300998687744, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8979479074478149, + "num_tokens": 609389915.0, + "step": 15698 + }, + { + "epoch": 1.9970741635924183, + "grad_norm": 2.5439138412475586, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8822533488273621, + "num_tokens": 609430796.0, + "step": 15699 + }, + { + "epoch": 1.9972013738710088, + "grad_norm": 2.4800939559936523, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8883309960365295, + "num_tokens": 609470854.0, + "step": 15700 + }, + { + "epoch": 1.9973285841495994, + "grad_norm": 2.825792074203491, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8850977420806885, + "num_tokens": 609506275.0, + "step": 15701 + }, + { + "epoch": 1.9974557944281899, + "grad_norm": 2.849576711654663, + "learning_rate": 1e-06, + "loss": 0.3432, + "mean_token_accuracy": 0.880162239074707, + "num_tokens": 609543814.0, + "step": 15702 + }, + { + "epoch": 1.9975830047067804, + "grad_norm": 2.5938491821289062, + "learning_rate": 1e-06, + "loss": 0.3071, + "mean_token_accuracy": 0.8882696032524109, + "num_tokens": 609582536.0, + "step": 15703 + }, + { + "epoch": 1.997710214985371, + "grad_norm": 2.5349347591400146, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.882479190826416, + "num_tokens": 609623218.0, + "step": 15704 + }, + { + "epoch": 1.9978374252639615, + "grad_norm": 2.5801968574523926, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.88902348279953, + "num_tokens": 609660534.0, + "step": 15705 + }, + { + "epoch": 1.997964635542552, + "grad_norm": 2.643592596054077, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9016062021255493, + "num_tokens": 609696005.0, + "step": 15706 + }, + { + "epoch": 1.9980918458211423, + "grad_norm": 2.612459421157837, + "learning_rate": 1e-06, + "loss": 0.3163, + "mean_token_accuracy": 0.8865727186203003, + "num_tokens": 609736465.0, + "step": 15707 + }, + { + "epoch": 1.9982190560997328, + "grad_norm": 2.5220680236816406, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8962117433547974, + "num_tokens": 609775072.0, + "step": 15708 + }, + { + "epoch": 1.9983462663783234, + "grad_norm": 2.5537962913513184, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8981684446334839, + "num_tokens": 609816663.0, + "step": 15709 + }, + { + "epoch": 1.9984734766569139, + "grad_norm": 2.5409021377563477, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8983324766159058, + "num_tokens": 609855687.0, + "step": 15710 + }, + { + "epoch": 1.9986006869355044, + "grad_norm": 2.868241786956787, + "learning_rate": 1e-06, + "loss": 0.3574, + "mean_token_accuracy": 0.8783907890319824, + "num_tokens": 609896035.0, + "step": 15711 + }, + { + "epoch": 1.998727897214095, + "grad_norm": 2.4552133083343506, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8969826698303223, + "num_tokens": 609936287.0, + "step": 15712 + }, + { + "epoch": 1.9988551074926852, + "grad_norm": 2.490633487701416, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8905719518661499, + "num_tokens": 609978579.0, + "step": 15713 + }, + { + "epoch": 1.9989823177712758, + "grad_norm": 2.343545913696289, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.893387496471405, + "num_tokens": 610023410.0, + "step": 15714 + }, + { + "epoch": 1.9991095280498663, + "grad_norm": 2.6363656520843506, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8876185417175293, + "num_tokens": 610062294.0, + "step": 15715 + }, + { + "epoch": 1.9992367383284568, + "grad_norm": 2.7454628944396973, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9025964736938477, + "num_tokens": 610095337.0, + "step": 15716 + }, + { + "epoch": 1.9993639486070474, + "grad_norm": 2.48818302154541, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.888672947883606, + "num_tokens": 610135979.0, + "step": 15717 + }, + { + "epoch": 1.9994911588856379, + "grad_norm": 2.522045373916626, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8937447667121887, + "num_tokens": 610177353.0, + "step": 15718 + }, + { + "epoch": 1.9996183691642284, + "grad_norm": 2.9053895473480225, + "learning_rate": 1e-06, + "loss": 0.3149, + "mean_token_accuracy": 0.8845196962356567, + "num_tokens": 610212545.0, + "step": 15719 + }, + { + "epoch": 1.999745579442819, + "grad_norm": 2.5453341007232666, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8983293771743774, + "num_tokens": 610247662.0, + "step": 15720 + }, + { + "epoch": 1.9998727897214095, + "grad_norm": 2.5805468559265137, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.894717812538147, + "num_tokens": 610286882.0, + "step": 15721 + }, + { + "epoch": 2.0, + "grad_norm": 2.508150339126587, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8963633179664612, + "num_tokens": 610324690.0, + "step": 15722 + }, + { + "epoch": 2.0001272102785905, + "grad_norm": 2.245736837387085, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9107334017753601, + "num_tokens": 610365752.0, + "step": 15723 + }, + { + "epoch": 2.000254420557181, + "grad_norm": 2.3783364295959473, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.8998162746429443, + "num_tokens": 610407024.0, + "step": 15724 + }, + { + "epoch": 2.0003816308357716, + "grad_norm": 2.401427984237671, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9035367369651794, + "num_tokens": 610450164.0, + "step": 15725 + }, + { + "epoch": 2.000508841114362, + "grad_norm": 2.2843165397644043, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9017988443374634, + "num_tokens": 610496348.0, + "step": 15726 + }, + { + "epoch": 2.0006360513929526, + "grad_norm": 2.422869920730591, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.908913254737854, + "num_tokens": 610536940.0, + "step": 15727 + }, + { + "epoch": 2.000763261671543, + "grad_norm": 2.4246392250061035, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9106992483139038, + "num_tokens": 610577578.0, + "step": 15728 + }, + { + "epoch": 2.0008904719501337, + "grad_norm": 2.389746904373169, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.904960036277771, + "num_tokens": 610619844.0, + "step": 15729 + }, + { + "epoch": 2.0010176822287242, + "grad_norm": 2.7776846885681152, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.907809853553772, + "num_tokens": 610658212.0, + "step": 15730 + }, + { + "epoch": 2.0011448925073148, + "grad_norm": 2.6271908283233643, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9095916748046875, + "num_tokens": 610693290.0, + "step": 15731 + }, + { + "epoch": 2.0012721027859053, + "grad_norm": 2.509039878845215, + "learning_rate": 1e-06, + "loss": 0.2329, + "mean_token_accuracy": 0.911108672618866, + "num_tokens": 610732569.0, + "step": 15732 + }, + { + "epoch": 2.001399313064496, + "grad_norm": 2.692497730255127, + "learning_rate": 1e-06, + "loss": 0.2435, + "mean_token_accuracy": 0.9099605083465576, + "num_tokens": 610770106.0, + "step": 15733 + }, + { + "epoch": 2.0015265233430863, + "grad_norm": 2.8144290447235107, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.8957793712615967, + "num_tokens": 610809813.0, + "step": 15734 + }, + { + "epoch": 2.0016537336216764, + "grad_norm": 2.870572805404663, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8972679376602173, + "num_tokens": 610849716.0, + "step": 15735 + }, + { + "epoch": 2.001780943900267, + "grad_norm": 2.689570665359497, + "learning_rate": 1e-06, + "loss": 0.2445, + "mean_token_accuracy": 0.9112637042999268, + "num_tokens": 610888881.0, + "step": 15736 + }, + { + "epoch": 2.0019081541788575, + "grad_norm": 2.6473820209503174, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9118767380714417, + "num_tokens": 610929279.0, + "step": 15737 + }, + { + "epoch": 2.002035364457448, + "grad_norm": 2.9809041023254395, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8972445130348206, + "num_tokens": 610968819.0, + "step": 15738 + }, + { + "epoch": 2.0021625747360385, + "grad_norm": 2.772918701171875, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8963736891746521, + "num_tokens": 611014145.0, + "step": 15739 + }, + { + "epoch": 2.002289785014629, + "grad_norm": 2.586527109146118, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9109381437301636, + "num_tokens": 611056196.0, + "step": 15740 + }, + { + "epoch": 2.0024169952932196, + "grad_norm": 2.7594332695007324, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.905813992023468, + "num_tokens": 611092727.0, + "step": 15741 + }, + { + "epoch": 2.00254420557181, + "grad_norm": 3.148937940597534, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8926117420196533, + "num_tokens": 611125660.0, + "step": 15742 + }, + { + "epoch": 2.0026714158504006, + "grad_norm": 2.763397693634033, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9017550945281982, + "num_tokens": 611169318.0, + "step": 15743 + }, + { + "epoch": 2.002798626128991, + "grad_norm": 2.6101651191711426, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9133425354957581, + "num_tokens": 611208164.0, + "step": 15744 + }, + { + "epoch": 2.0029258364075817, + "grad_norm": 2.6139187812805176, + "learning_rate": 1e-06, + "loss": 0.2177, + "mean_token_accuracy": 0.9180321097373962, + "num_tokens": 611243758.0, + "step": 15745 + }, + { + "epoch": 2.0030530466861722, + "grad_norm": 2.522909641265869, + "learning_rate": 1e-06, + "loss": 0.2098, + "mean_token_accuracy": 0.9248228669166565, + "num_tokens": 611284056.0, + "step": 15746 + }, + { + "epoch": 2.0031802569647628, + "grad_norm": 2.6460893154144287, + "learning_rate": 1e-06, + "loss": 0.2235, + "mean_token_accuracy": 0.9162559509277344, + "num_tokens": 611322637.0, + "step": 15747 + }, + { + "epoch": 2.0033074672433533, + "grad_norm": 2.653231143951416, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9065970778465271, + "num_tokens": 611360881.0, + "step": 15748 + }, + { + "epoch": 2.003434677521944, + "grad_norm": 2.56288743019104, + "learning_rate": 1e-06, + "loss": 0.2278, + "mean_token_accuracy": 0.9136323928833008, + "num_tokens": 611401305.0, + "step": 15749 + }, + { + "epoch": 2.0035618878005343, + "grad_norm": 2.839296817779541, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8961082100868225, + "num_tokens": 611439347.0, + "step": 15750 + }, + { + "epoch": 2.003689098079125, + "grad_norm": 2.6408793926239014, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9009090065956116, + "num_tokens": 611481955.0, + "step": 15751 + }, + { + "epoch": 2.0038163083577154, + "grad_norm": 2.7000370025634766, + "learning_rate": 1e-06, + "loss": 0.233, + "mean_token_accuracy": 0.912666380405426, + "num_tokens": 611519652.0, + "step": 15752 + }, + { + "epoch": 2.003943518636306, + "grad_norm": 2.6528701782226562, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9076387882232666, + "num_tokens": 611561019.0, + "step": 15753 + }, + { + "epoch": 2.0040707289148965, + "grad_norm": 2.832303285598755, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9069992303848267, + "num_tokens": 611600464.0, + "step": 15754 + }, + { + "epoch": 2.004197939193487, + "grad_norm": 3.0486621856689453, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.89167320728302, + "num_tokens": 611638314.0, + "step": 15755 + }, + { + "epoch": 2.0043251494720775, + "grad_norm": 2.675652027130127, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9084397554397583, + "num_tokens": 611676229.0, + "step": 15756 + }, + { + "epoch": 2.004452359750668, + "grad_norm": 2.890603542327881, + "learning_rate": 1e-06, + "loss": 0.2234, + "mean_token_accuracy": 0.9174313545227051, + "num_tokens": 611708611.0, + "step": 15757 + }, + { + "epoch": 2.0045795700292586, + "grad_norm": 2.7485604286193848, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9045326709747314, + "num_tokens": 611743702.0, + "step": 15758 + }, + { + "epoch": 2.0047067803078487, + "grad_norm": 2.8471384048461914, + "learning_rate": 1e-06, + "loss": 0.2153, + "mean_token_accuracy": 0.9201306700706482, + "num_tokens": 611775844.0, + "step": 15759 + }, + { + "epoch": 2.004833990586439, + "grad_norm": 2.7856698036193848, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9066421985626221, + "num_tokens": 611813494.0, + "step": 15760 + }, + { + "epoch": 2.0049612008650297, + "grad_norm": 2.734196424484253, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.904017448425293, + "num_tokens": 611854284.0, + "step": 15761 + }, + { + "epoch": 2.0050884111436202, + "grad_norm": 3.0647997856140137, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9039531946182251, + "num_tokens": 611888794.0, + "step": 15762 + }, + { + "epoch": 2.0052156214222108, + "grad_norm": 2.849885940551758, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9120062589645386, + "num_tokens": 611924882.0, + "step": 15763 + }, + { + "epoch": 2.0053428317008013, + "grad_norm": 2.832365036010742, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8908101320266724, + "num_tokens": 611968458.0, + "step": 15764 + }, + { + "epoch": 2.005470041979392, + "grad_norm": 2.9802918434143066, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9004775881767273, + "num_tokens": 612004808.0, + "step": 15765 + }, + { + "epoch": 2.0055972522579824, + "grad_norm": 2.9968862533569336, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.914829432964325, + "num_tokens": 612035517.0, + "step": 15766 + }, + { + "epoch": 2.005724462536573, + "grad_norm": 2.7162282466888428, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9079844951629639, + "num_tokens": 612073176.0, + "step": 15767 + }, + { + "epoch": 2.0058516728151634, + "grad_norm": 2.88655161857605, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9013258218765259, + "num_tokens": 612109599.0, + "step": 15768 + }, + { + "epoch": 2.005978883093754, + "grad_norm": 3.1011664867401123, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9003875255584717, + "num_tokens": 612145400.0, + "step": 15769 + }, + { + "epoch": 2.0061060933723445, + "grad_norm": 2.8632147312164307, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9013161063194275, + "num_tokens": 612184518.0, + "step": 15770 + }, + { + "epoch": 2.006233303650935, + "grad_norm": 2.5863373279571533, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9078871011734009, + "num_tokens": 612227882.0, + "step": 15771 + }, + { + "epoch": 2.0063605139295255, + "grad_norm": 2.839543104171753, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9022692441940308, + "num_tokens": 612266825.0, + "step": 15772 + }, + { + "epoch": 2.006487724208116, + "grad_norm": 2.8491554260253906, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8960965275764465, + "num_tokens": 612307172.0, + "step": 15773 + }, + { + "epoch": 2.0066149344867066, + "grad_norm": 2.424973726272583, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.9108266830444336, + "num_tokens": 612356667.0, + "step": 15774 + }, + { + "epoch": 2.006742144765297, + "grad_norm": 2.4430549144744873, + "learning_rate": 1e-06, + "loss": 0.2019, + "mean_token_accuracy": 0.9235567450523376, + "num_tokens": 612395243.0, + "step": 15775 + }, + { + "epoch": 2.0068693550438876, + "grad_norm": 2.7960212230682373, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8986303806304932, + "num_tokens": 612438675.0, + "step": 15776 + }, + { + "epoch": 2.006996565322478, + "grad_norm": 2.5298993587493896, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9009119272232056, + "num_tokens": 612485411.0, + "step": 15777 + }, + { + "epoch": 2.0071237756010687, + "grad_norm": 2.4480624198913574, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9141585826873779, + "num_tokens": 612530877.0, + "step": 15778 + }, + { + "epoch": 2.007250985879659, + "grad_norm": 2.9234392642974854, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8984900116920471, + "num_tokens": 612569662.0, + "step": 15779 + }, + { + "epoch": 2.0073781961582498, + "grad_norm": 2.887517213821411, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9083874225616455, + "num_tokens": 612603027.0, + "step": 15780 + }, + { + "epoch": 2.0075054064368403, + "grad_norm": 2.698838710784912, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.9117490649223328, + "num_tokens": 612641015.0, + "step": 15781 + }, + { + "epoch": 2.007632616715431, + "grad_norm": 2.6478042602539062, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9079138040542603, + "num_tokens": 612685616.0, + "step": 15782 + }, + { + "epoch": 2.0077598269940213, + "grad_norm": 2.6928882598876953, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9022970199584961, + "num_tokens": 612727900.0, + "step": 15783 + }, + { + "epoch": 2.0078870372726114, + "grad_norm": 2.80772066116333, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9013108015060425, + "num_tokens": 612767254.0, + "step": 15784 + }, + { + "epoch": 2.008014247551202, + "grad_norm": 2.9381237030029297, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9076890349388123, + "num_tokens": 612802748.0, + "step": 15785 + }, + { + "epoch": 2.0081414578297925, + "grad_norm": 2.8309407234191895, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9010725617408752, + "num_tokens": 612841548.0, + "step": 15786 + }, + { + "epoch": 2.008268668108383, + "grad_norm": 2.9371449947357178, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9071313142776489, + "num_tokens": 612873599.0, + "step": 15787 + }, + { + "epoch": 2.0083958783869735, + "grad_norm": 2.7448010444641113, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9043049812316895, + "num_tokens": 612915280.0, + "step": 15788 + }, + { + "epoch": 2.008523088665564, + "grad_norm": 2.8459997177124023, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8964415788650513, + "num_tokens": 612952998.0, + "step": 15789 + }, + { + "epoch": 2.0086502989441546, + "grad_norm": 2.7170238494873047, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9065505862236023, + "num_tokens": 612990505.0, + "step": 15790 + }, + { + "epoch": 2.008777509222745, + "grad_norm": 3.0048630237579346, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.908388078212738, + "num_tokens": 613023342.0, + "step": 15791 + }, + { + "epoch": 2.0089047195013356, + "grad_norm": 2.9727401733398438, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9008381366729736, + "num_tokens": 613058357.0, + "step": 15792 + }, + { + "epoch": 2.009031929779926, + "grad_norm": 2.7326486110687256, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9095287919044495, + "num_tokens": 613097000.0, + "step": 15793 + }, + { + "epoch": 2.0091591400585167, + "grad_norm": 2.787648916244507, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8992475867271423, + "num_tokens": 613138062.0, + "step": 15794 + }, + { + "epoch": 2.0092863503371072, + "grad_norm": 2.4993953704833984, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9075000882148743, + "num_tokens": 613180577.0, + "step": 15795 + }, + { + "epoch": 2.0094135606156978, + "grad_norm": 2.8015265464782715, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9048577547073364, + "num_tokens": 613218508.0, + "step": 15796 + }, + { + "epoch": 2.0095407708942883, + "grad_norm": 2.8620810508728027, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9015784859657288, + "num_tokens": 613260525.0, + "step": 15797 + }, + { + "epoch": 2.009667981172879, + "grad_norm": 2.8307371139526367, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9088715314865112, + "num_tokens": 613300819.0, + "step": 15798 + }, + { + "epoch": 2.0097951914514693, + "grad_norm": 2.6131372451782227, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9080950021743774, + "num_tokens": 613341870.0, + "step": 15799 + }, + { + "epoch": 2.00992240173006, + "grad_norm": 2.8082540035247803, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9072856307029724, + "num_tokens": 613380914.0, + "step": 15800 + }, + { + "epoch": 2.0100496120086504, + "grad_norm": 2.7537713050842285, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9074761867523193, + "num_tokens": 613421770.0, + "step": 15801 + }, + { + "epoch": 2.010176822287241, + "grad_norm": 2.7873454093933105, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9060782194137573, + "num_tokens": 613459835.0, + "step": 15802 + }, + { + "epoch": 2.0103040325658315, + "grad_norm": 2.852142095565796, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9038642644882202, + "num_tokens": 613499302.0, + "step": 15803 + }, + { + "epoch": 2.010431242844422, + "grad_norm": 2.7884533405303955, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9117839336395264, + "num_tokens": 613536976.0, + "step": 15804 + }, + { + "epoch": 2.0105584531230125, + "grad_norm": 2.633106231689453, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9062032699584961, + "num_tokens": 613579317.0, + "step": 15805 + }, + { + "epoch": 2.010685663401603, + "grad_norm": 2.7587027549743652, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9116450548171997, + "num_tokens": 613617194.0, + "step": 15806 + }, + { + "epoch": 2.0108128736801936, + "grad_norm": 2.876636028289795, + "learning_rate": 1e-06, + "loss": 0.2303, + "mean_token_accuracy": 0.9117960929870605, + "num_tokens": 613651264.0, + "step": 15807 + }, + { + "epoch": 2.0109400839587837, + "grad_norm": 3.159870147705078, + "learning_rate": 1e-06, + "loss": 0.3103, + "mean_token_accuracy": 0.8854221701622009, + "num_tokens": 613692394.0, + "step": 15808 + }, + { + "epoch": 2.011067294237374, + "grad_norm": 2.649507761001587, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9057445526123047, + "num_tokens": 613735238.0, + "step": 15809 + }, + { + "epoch": 2.0111945045159647, + "grad_norm": 2.875782012939453, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9092486500740051, + "num_tokens": 613768016.0, + "step": 15810 + }, + { + "epoch": 2.0113217147945552, + "grad_norm": 2.520658254623413, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9099401235580444, + "num_tokens": 613811222.0, + "step": 15811 + }, + { + "epoch": 2.0114489250731458, + "grad_norm": 2.9914755821228027, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9060190320014954, + "num_tokens": 613844507.0, + "step": 15812 + }, + { + "epoch": 2.0115761353517363, + "grad_norm": 2.5381274223327637, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.911343514919281, + "num_tokens": 613888943.0, + "step": 15813 + }, + { + "epoch": 2.011703345630327, + "grad_norm": 2.8590807914733887, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.9095603823661804, + "num_tokens": 613923922.0, + "step": 15814 + }, + { + "epoch": 2.0118305559089174, + "grad_norm": 3.180483341217041, + "learning_rate": 1e-06, + "loss": 0.2205, + "mean_token_accuracy": 0.918357253074646, + "num_tokens": 613952947.0, + "step": 15815 + }, + { + "epoch": 2.011957766187508, + "grad_norm": 2.7258636951446533, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.906247079372406, + "num_tokens": 613992313.0, + "step": 15816 + }, + { + "epoch": 2.0120849764660984, + "grad_norm": 2.482112407684326, + "learning_rate": 1e-06, + "loss": 0.232, + "mean_token_accuracy": 0.9146960973739624, + "num_tokens": 614039302.0, + "step": 15817 + }, + { + "epoch": 2.012212186744689, + "grad_norm": 3.0915637016296387, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8886206150054932, + "num_tokens": 614078648.0, + "step": 15818 + }, + { + "epoch": 2.0123393970232795, + "grad_norm": 2.6887593269348145, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9006799459457397, + "num_tokens": 614122643.0, + "step": 15819 + }, + { + "epoch": 2.01246660730187, + "grad_norm": 3.2779054641723633, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8960731029510498, + "num_tokens": 614156473.0, + "step": 15820 + }, + { + "epoch": 2.0125938175804605, + "grad_norm": 2.752206802368164, + "learning_rate": 1e-06, + "loss": 0.2416, + "mean_token_accuracy": 0.9097535610198975, + "num_tokens": 614195067.0, + "step": 15821 + }, + { + "epoch": 2.012721027859051, + "grad_norm": 2.846358060836792, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9055197834968567, + "num_tokens": 614235820.0, + "step": 15822 + }, + { + "epoch": 2.0128482381376416, + "grad_norm": 2.8396947383880615, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.8987469673156738, + "num_tokens": 614274147.0, + "step": 15823 + }, + { + "epoch": 2.012975448416232, + "grad_norm": 2.682914972305298, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9092696905136108, + "num_tokens": 614315897.0, + "step": 15824 + }, + { + "epoch": 2.0131026586948226, + "grad_norm": 2.788346290588379, + "learning_rate": 1e-06, + "loss": 0.2315, + "mean_token_accuracy": 0.9150735139846802, + "num_tokens": 614352961.0, + "step": 15825 + }, + { + "epoch": 2.013229868973413, + "grad_norm": 2.722217321395874, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9122995734214783, + "num_tokens": 614393266.0, + "step": 15826 + }, + { + "epoch": 2.0133570792520037, + "grad_norm": 2.5100297927856445, + "learning_rate": 1e-06, + "loss": 0.2129, + "mean_token_accuracy": 0.9206944108009338, + "num_tokens": 614434940.0, + "step": 15827 + }, + { + "epoch": 2.013484289530594, + "grad_norm": 3.006182909011841, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8950833082199097, + "num_tokens": 614473714.0, + "step": 15828 + }, + { + "epoch": 2.0136114998091847, + "grad_norm": 2.9001238346099854, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9047381281852722, + "num_tokens": 614509534.0, + "step": 15829 + }, + { + "epoch": 2.0137387100877753, + "grad_norm": 2.5659799575805664, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9051052331924438, + "num_tokens": 614559093.0, + "step": 15830 + }, + { + "epoch": 2.013865920366366, + "grad_norm": 2.9126975536346436, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9103962182998657, + "num_tokens": 614594775.0, + "step": 15831 + }, + { + "epoch": 2.0139931306449563, + "grad_norm": 2.6068551540374756, + "learning_rate": 1e-06, + "loss": 0.2313, + "mean_token_accuracy": 0.915407657623291, + "num_tokens": 614635751.0, + "step": 15832 + }, + { + "epoch": 2.0141203409235464, + "grad_norm": 2.8504841327667236, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9112088680267334, + "num_tokens": 614669657.0, + "step": 15833 + }, + { + "epoch": 2.014247551202137, + "grad_norm": 3.0302932262420654, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9047181010246277, + "num_tokens": 614702298.0, + "step": 15834 + }, + { + "epoch": 2.0143747614807275, + "grad_norm": 2.600353479385376, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9067542552947998, + "num_tokens": 614746334.0, + "step": 15835 + }, + { + "epoch": 2.014501971759318, + "grad_norm": 2.5546212196350098, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9066792726516724, + "num_tokens": 614791052.0, + "step": 15836 + }, + { + "epoch": 2.0146291820379085, + "grad_norm": 2.7587342262268066, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9072834849357605, + "num_tokens": 614832703.0, + "step": 15837 + }, + { + "epoch": 2.014756392316499, + "grad_norm": 2.826007843017578, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9104790091514587, + "num_tokens": 614868466.0, + "step": 15838 + }, + { + "epoch": 2.0148836025950896, + "grad_norm": 2.8736140727996826, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9077874422073364, + "num_tokens": 614906576.0, + "step": 15839 + }, + { + "epoch": 2.01501081287368, + "grad_norm": 2.6168582439422607, + "learning_rate": 1e-06, + "loss": 0.2267, + "mean_token_accuracy": 0.9155904650688171, + "num_tokens": 614945681.0, + "step": 15840 + }, + { + "epoch": 2.0151380231522706, + "grad_norm": 2.824572801589966, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9042654037475586, + "num_tokens": 614983278.0, + "step": 15841 + }, + { + "epoch": 2.015265233430861, + "grad_norm": 2.690540075302124, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9081559181213379, + "num_tokens": 615025392.0, + "step": 15842 + }, + { + "epoch": 2.0153924437094517, + "grad_norm": 3.076373338699341, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9104517102241516, + "num_tokens": 615058366.0, + "step": 15843 + }, + { + "epoch": 2.0155196539880422, + "grad_norm": 2.708868980407715, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9088295698165894, + "num_tokens": 615100527.0, + "step": 15844 + }, + { + "epoch": 2.0156468642666328, + "grad_norm": 2.6281557083129883, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9085378050804138, + "num_tokens": 615142156.0, + "step": 15845 + }, + { + "epoch": 2.0157740745452233, + "grad_norm": 3.0682337284088135, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9129724502563477, + "num_tokens": 615176237.0, + "step": 15846 + }, + { + "epoch": 2.015901284823814, + "grad_norm": 3.0087475776672363, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8912824392318726, + "num_tokens": 615220248.0, + "step": 15847 + }, + { + "epoch": 2.0160284951024043, + "grad_norm": 2.787912130355835, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9076480269432068, + "num_tokens": 615262392.0, + "step": 15848 + }, + { + "epoch": 2.016155705380995, + "grad_norm": 2.7822628021240234, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9107409119606018, + "num_tokens": 615297492.0, + "step": 15849 + }, + { + "epoch": 2.0162829156595854, + "grad_norm": 2.8371682167053223, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9071234464645386, + "num_tokens": 615334169.0, + "step": 15850 + }, + { + "epoch": 2.016410125938176, + "grad_norm": 2.780078411102295, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9093274474143982, + "num_tokens": 615369572.0, + "step": 15851 + }, + { + "epoch": 2.0165373362167665, + "grad_norm": 2.5627031326293945, + "learning_rate": 1e-06, + "loss": 0.2296, + "mean_token_accuracy": 0.9159278273582458, + "num_tokens": 615410052.0, + "step": 15852 + }, + { + "epoch": 2.016664546495357, + "grad_norm": 2.973562240600586, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9105821847915649, + "num_tokens": 615447020.0, + "step": 15853 + }, + { + "epoch": 2.0167917567739475, + "grad_norm": 2.766292095184326, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9040219187736511, + "num_tokens": 615484545.0, + "step": 15854 + }, + { + "epoch": 2.016918967052538, + "grad_norm": 2.6240105628967285, + "learning_rate": 1e-06, + "loss": 0.2314, + "mean_token_accuracy": 0.9143036007881165, + "num_tokens": 615527101.0, + "step": 15855 + }, + { + "epoch": 2.0170461773311286, + "grad_norm": 2.8355658054351807, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9039308428764343, + "num_tokens": 615563478.0, + "step": 15856 + }, + { + "epoch": 2.0171733876097186, + "grad_norm": 2.71683669090271, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9095725417137146, + "num_tokens": 615605517.0, + "step": 15857 + }, + { + "epoch": 2.017300597888309, + "grad_norm": 3.1555070877075195, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.905290424823761, + "num_tokens": 615637649.0, + "step": 15858 + }, + { + "epoch": 2.0174278081668997, + "grad_norm": 2.7212438583374023, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9027478694915771, + "num_tokens": 615680113.0, + "step": 15859 + }, + { + "epoch": 2.0175550184454902, + "grad_norm": 2.5942206382751465, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9090583324432373, + "num_tokens": 615722381.0, + "step": 15860 + }, + { + "epoch": 2.0176822287240808, + "grad_norm": 2.874054193496704, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9015931487083435, + "num_tokens": 615759612.0, + "step": 15861 + }, + { + "epoch": 2.0178094390026713, + "grad_norm": 2.6114656925201416, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9035173654556274, + "num_tokens": 615801601.0, + "step": 15862 + }, + { + "epoch": 2.017936649281262, + "grad_norm": 3.5103814601898193, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.8992618918418884, + "num_tokens": 615843955.0, + "step": 15863 + }, + { + "epoch": 2.0180638595598523, + "grad_norm": 2.8222756385803223, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8958495855331421, + "num_tokens": 615882608.0, + "step": 15864 + }, + { + "epoch": 2.018191069838443, + "grad_norm": 2.84963321685791, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9022396802902222, + "num_tokens": 615920295.0, + "step": 15865 + }, + { + "epoch": 2.0183182801170334, + "grad_norm": 2.70654559135437, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9034786224365234, + "num_tokens": 615962949.0, + "step": 15866 + }, + { + "epoch": 2.018445490395624, + "grad_norm": 2.903334856033325, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9033643007278442, + "num_tokens": 615999688.0, + "step": 15867 + }, + { + "epoch": 2.0185727006742145, + "grad_norm": 2.713728427886963, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9004296064376831, + "num_tokens": 616043470.0, + "step": 15868 + }, + { + "epoch": 2.018699910952805, + "grad_norm": 2.9239487648010254, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9116435647010803, + "num_tokens": 616079231.0, + "step": 15869 + }, + { + "epoch": 2.0188271212313955, + "grad_norm": 2.7544803619384766, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9108630418777466, + "num_tokens": 616121208.0, + "step": 15870 + }, + { + "epoch": 2.018954331509986, + "grad_norm": 3.0948567390441895, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9014368057250977, + "num_tokens": 616154423.0, + "step": 15871 + }, + { + "epoch": 2.0190815417885766, + "grad_norm": 2.9847981929779053, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.8977624177932739, + "num_tokens": 616188580.0, + "step": 15872 + }, + { + "epoch": 2.019208752067167, + "grad_norm": 2.656604528427124, + "learning_rate": 1e-06, + "loss": 0.2195, + "mean_token_accuracy": 0.9164861440658569, + "num_tokens": 616224892.0, + "step": 15873 + }, + { + "epoch": 2.0193359623457576, + "grad_norm": 2.9091453552246094, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9057360887527466, + "num_tokens": 616261374.0, + "step": 15874 + }, + { + "epoch": 2.019463172624348, + "grad_norm": 2.6735098361968994, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9034053683280945, + "num_tokens": 616301959.0, + "step": 15875 + }, + { + "epoch": 2.0195903829029387, + "grad_norm": 3.0724668502807617, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9029327034950256, + "num_tokens": 616341077.0, + "step": 15876 + }, + { + "epoch": 2.019717593181529, + "grad_norm": 2.8221709728240967, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9011838436126709, + "num_tokens": 616377587.0, + "step": 15877 + }, + { + "epoch": 2.0198448034601197, + "grad_norm": 2.7986645698547363, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9096379280090332, + "num_tokens": 616413576.0, + "step": 15878 + }, + { + "epoch": 2.0199720137387103, + "grad_norm": 2.7808642387390137, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9119570255279541, + "num_tokens": 616451255.0, + "step": 15879 + }, + { + "epoch": 2.020099224017301, + "grad_norm": 2.715163230895996, + "learning_rate": 1e-06, + "loss": 0.2277, + "mean_token_accuracy": 0.918219268321991, + "num_tokens": 616487677.0, + "step": 15880 + }, + { + "epoch": 2.0202264342958913, + "grad_norm": 2.858466386795044, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9052634239196777, + "num_tokens": 616523954.0, + "step": 15881 + }, + { + "epoch": 2.0203536445744814, + "grad_norm": 2.7613656520843506, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9062191843986511, + "num_tokens": 616564147.0, + "step": 15882 + }, + { + "epoch": 2.020480854853072, + "grad_norm": 2.9958577156066895, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9110293388366699, + "num_tokens": 616598550.0, + "step": 15883 + }, + { + "epoch": 2.0206080651316625, + "grad_norm": 2.869436502456665, + "learning_rate": 1e-06, + "loss": 0.2257, + "mean_token_accuracy": 0.9149727821350098, + "num_tokens": 616630781.0, + "step": 15884 + }, + { + "epoch": 2.020735275410253, + "grad_norm": 2.6359708309173584, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9026771783828735, + "num_tokens": 616674931.0, + "step": 15885 + }, + { + "epoch": 2.0208624856888435, + "grad_norm": 2.7919094562530518, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.910408616065979, + "num_tokens": 616710011.0, + "step": 15886 + }, + { + "epoch": 2.020989695967434, + "grad_norm": 2.6541004180908203, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8973420858383179, + "num_tokens": 616756206.0, + "step": 15887 + }, + { + "epoch": 2.0211169062460246, + "grad_norm": 2.9968109130859375, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9058851003646851, + "num_tokens": 616797604.0, + "step": 15888 + }, + { + "epoch": 2.021244116524615, + "grad_norm": 2.9093031883239746, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9086230993270874, + "num_tokens": 616833774.0, + "step": 15889 + }, + { + "epoch": 2.0213713268032056, + "grad_norm": 2.9279041290283203, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9068284630775452, + "num_tokens": 616869447.0, + "step": 15890 + }, + { + "epoch": 2.021498537081796, + "grad_norm": 3.2084596157073975, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.897060751914978, + "num_tokens": 616905414.0, + "step": 15891 + }, + { + "epoch": 2.0216257473603867, + "grad_norm": 3.0841574668884277, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9044407606124878, + "num_tokens": 616937098.0, + "step": 15892 + }, + { + "epoch": 2.021752957638977, + "grad_norm": 2.6305320262908936, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8977339267730713, + "num_tokens": 616985538.0, + "step": 15893 + }, + { + "epoch": 2.0218801679175677, + "grad_norm": 2.7897753715515137, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9053429961204529, + "num_tokens": 617026753.0, + "step": 15894 + }, + { + "epoch": 2.0220073781961583, + "grad_norm": 2.7825636863708496, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9121172428131104, + "num_tokens": 617065282.0, + "step": 15895 + }, + { + "epoch": 2.022134588474749, + "grad_norm": 2.7632217407226562, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9101108312606812, + "num_tokens": 617104580.0, + "step": 15896 + }, + { + "epoch": 2.0222617987533393, + "grad_norm": 2.8766679763793945, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9045922756195068, + "num_tokens": 617140545.0, + "step": 15897 + }, + { + "epoch": 2.02238900903193, + "grad_norm": 3.2218334674835205, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8990347385406494, + "num_tokens": 617175575.0, + "step": 15898 + }, + { + "epoch": 2.0225162193105204, + "grad_norm": 2.877028226852417, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9080641269683838, + "num_tokens": 617215687.0, + "step": 15899 + }, + { + "epoch": 2.022643429589111, + "grad_norm": 2.813697338104248, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9106708765029907, + "num_tokens": 617254694.0, + "step": 15900 + }, + { + "epoch": 2.0227706398677014, + "grad_norm": 2.751345634460449, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9092050790786743, + "num_tokens": 617293717.0, + "step": 15901 + }, + { + "epoch": 2.022897850146292, + "grad_norm": 2.70495343208313, + "learning_rate": 1e-06, + "loss": 0.2128, + "mean_token_accuracy": 0.9195443987846375, + "num_tokens": 617334001.0, + "step": 15902 + }, + { + "epoch": 2.0230250604248825, + "grad_norm": 2.9177300930023193, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9085378050804138, + "num_tokens": 617370130.0, + "step": 15903 + }, + { + "epoch": 2.023152270703473, + "grad_norm": 2.5206007957458496, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9000980854034424, + "num_tokens": 617415424.0, + "step": 15904 + }, + { + "epoch": 2.0232794809820636, + "grad_norm": 3.040534734725952, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.8999247550964355, + "num_tokens": 617449482.0, + "step": 15905 + }, + { + "epoch": 2.0234066912606536, + "grad_norm": 3.3631362915039062, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.904644787311554, + "num_tokens": 617479362.0, + "step": 15906 + }, + { + "epoch": 2.023533901539244, + "grad_norm": 3.023900270462036, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8963047862052917, + "num_tokens": 617515675.0, + "step": 15907 + }, + { + "epoch": 2.0236611118178347, + "grad_norm": 2.8990182876586914, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9072450399398804, + "num_tokens": 617554843.0, + "step": 15908 + }, + { + "epoch": 2.0237883220964252, + "grad_norm": 2.668009042739868, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9088767766952515, + "num_tokens": 617594889.0, + "step": 15909 + }, + { + "epoch": 2.0239155323750158, + "grad_norm": 2.6883420944213867, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9028902649879456, + "num_tokens": 617637038.0, + "step": 15910 + }, + { + "epoch": 2.0240427426536063, + "grad_norm": 2.990217447280884, + "learning_rate": 1e-06, + "loss": 0.2311, + "mean_token_accuracy": 0.9144477844238281, + "num_tokens": 617672120.0, + "step": 15911 + }, + { + "epoch": 2.024169952932197, + "grad_norm": 2.9068872928619385, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9131486415863037, + "num_tokens": 617706572.0, + "step": 15912 + }, + { + "epoch": 2.0242971632107873, + "grad_norm": 2.904886245727539, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9045132398605347, + "num_tokens": 617746679.0, + "step": 15913 + }, + { + "epoch": 2.024424373489378, + "grad_norm": 2.9851863384246826, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9056060314178467, + "num_tokens": 617781013.0, + "step": 15914 + }, + { + "epoch": 2.0245515837679684, + "grad_norm": 3.044605255126953, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9120282530784607, + "num_tokens": 617813832.0, + "step": 15915 + }, + { + "epoch": 2.024678794046559, + "grad_norm": 3.149892568588257, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8970866799354553, + "num_tokens": 617850996.0, + "step": 15916 + }, + { + "epoch": 2.0248060043251495, + "grad_norm": 2.738013505935669, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9068419933319092, + "num_tokens": 617889305.0, + "step": 15917 + }, + { + "epoch": 2.02493321460374, + "grad_norm": 2.625683069229126, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9039629697799683, + "num_tokens": 617931255.0, + "step": 15918 + }, + { + "epoch": 2.0250604248823305, + "grad_norm": 2.481506824493408, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.911599338054657, + "num_tokens": 617975683.0, + "step": 15919 + }, + { + "epoch": 2.025187635160921, + "grad_norm": 2.6815602779388428, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9122689366340637, + "num_tokens": 618013518.0, + "step": 15920 + }, + { + "epoch": 2.0253148454395116, + "grad_norm": 2.807040214538574, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9025089740753174, + "num_tokens": 618053203.0, + "step": 15921 + }, + { + "epoch": 2.025442055718102, + "grad_norm": 2.935659646987915, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9076932072639465, + "num_tokens": 618091081.0, + "step": 15922 + }, + { + "epoch": 2.0255692659966926, + "grad_norm": 2.686117172241211, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9098849892616272, + "num_tokens": 618131500.0, + "step": 15923 + }, + { + "epoch": 2.025696476275283, + "grad_norm": 2.9591643810272217, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.8943469524383545, + "num_tokens": 618169907.0, + "step": 15924 + }, + { + "epoch": 2.0258236865538737, + "grad_norm": 2.446634531021118, + "learning_rate": 1e-06, + "loss": 0.2027, + "mean_token_accuracy": 0.9244604110717773, + "num_tokens": 618209237.0, + "step": 15925 + }, + { + "epoch": 2.025950896832464, + "grad_norm": 2.7509806156158447, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8979196548461914, + "num_tokens": 618249476.0, + "step": 15926 + }, + { + "epoch": 2.0260781071110547, + "grad_norm": 2.985464334487915, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9102039933204651, + "num_tokens": 618283635.0, + "step": 15927 + }, + { + "epoch": 2.0262053173896453, + "grad_norm": 2.761626958847046, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9039852619171143, + "num_tokens": 618326635.0, + "step": 15928 + }, + { + "epoch": 2.026332527668236, + "grad_norm": 2.4869139194488525, + "learning_rate": 1e-06, + "loss": 0.2028, + "mean_token_accuracy": 0.9235323667526245, + "num_tokens": 618369237.0, + "step": 15929 + }, + { + "epoch": 2.0264597379468263, + "grad_norm": 3.120032787322998, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.905962347984314, + "num_tokens": 618400594.0, + "step": 15930 + }, + { + "epoch": 2.0265869482254164, + "grad_norm": 2.6478703022003174, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9017283916473389, + "num_tokens": 618444498.0, + "step": 15931 + }, + { + "epoch": 2.026714158504007, + "grad_norm": 2.9210634231567383, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9047983884811401, + "num_tokens": 618480519.0, + "step": 15932 + }, + { + "epoch": 2.0268413687825975, + "grad_norm": 2.6969258785247803, + "learning_rate": 1e-06, + "loss": 0.2164, + "mean_token_accuracy": 0.9200891256332397, + "num_tokens": 618518474.0, + "step": 15933 + }, + { + "epoch": 2.026968579061188, + "grad_norm": 2.9172468185424805, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9074790477752686, + "num_tokens": 618552454.0, + "step": 15934 + }, + { + "epoch": 2.0270957893397785, + "grad_norm": 2.7998030185699463, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9147157073020935, + "num_tokens": 618590223.0, + "step": 15935 + }, + { + "epoch": 2.027222999618369, + "grad_norm": 2.664094924926758, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9104756116867065, + "num_tokens": 618629588.0, + "step": 15936 + }, + { + "epoch": 2.0273502098969596, + "grad_norm": 2.572598457336426, + "learning_rate": 1e-06, + "loss": 0.2255, + "mean_token_accuracy": 0.9162108898162842, + "num_tokens": 618673624.0, + "step": 15937 + }, + { + "epoch": 2.02747742017555, + "grad_norm": 3.04569673538208, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9082546830177307, + "num_tokens": 618706108.0, + "step": 15938 + }, + { + "epoch": 2.0276046304541406, + "grad_norm": 2.9844322204589844, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9062867164611816, + "num_tokens": 618741139.0, + "step": 15939 + }, + { + "epoch": 2.027731840732731, + "grad_norm": 2.714991331100464, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9070621728897095, + "num_tokens": 618784250.0, + "step": 15940 + }, + { + "epoch": 2.0278590510113217, + "grad_norm": 2.78291916847229, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9000524282455444, + "num_tokens": 618826250.0, + "step": 15941 + }, + { + "epoch": 2.027986261289912, + "grad_norm": 3.127479076385498, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9018732309341431, + "num_tokens": 618861305.0, + "step": 15942 + }, + { + "epoch": 2.0281134715685027, + "grad_norm": 2.6599533557891846, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9051170945167542, + "num_tokens": 618906629.0, + "step": 15943 + }, + { + "epoch": 2.0282406818470933, + "grad_norm": 3.035379409790039, + "learning_rate": 1e-06, + "loss": 0.2105, + "mean_token_accuracy": 0.9247601628303528, + "num_tokens": 618938094.0, + "step": 15944 + }, + { + "epoch": 2.028367892125684, + "grad_norm": 2.6593916416168213, + "learning_rate": 1e-06, + "loss": 0.2164, + "mean_token_accuracy": 0.91881263256073, + "num_tokens": 618977214.0, + "step": 15945 + }, + { + "epoch": 2.0284951024042743, + "grad_norm": 3.154491424560547, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8908180594444275, + "num_tokens": 619016258.0, + "step": 15946 + }, + { + "epoch": 2.028622312682865, + "grad_norm": 2.7559187412261963, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9097229242324829, + "num_tokens": 619057201.0, + "step": 15947 + }, + { + "epoch": 2.0287495229614554, + "grad_norm": 2.78373646736145, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9051628708839417, + "num_tokens": 619096490.0, + "step": 15948 + }, + { + "epoch": 2.028876733240046, + "grad_norm": 2.722698926925659, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.9129751920700073, + "num_tokens": 619138330.0, + "step": 15949 + }, + { + "epoch": 2.0290039435186364, + "grad_norm": 2.628455400466919, + "learning_rate": 1e-06, + "loss": 0.2176, + "mean_token_accuracy": 0.9189979434013367, + "num_tokens": 619176171.0, + "step": 15950 + }, + { + "epoch": 2.029131153797227, + "grad_norm": 2.816877603530884, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.902640700340271, + "num_tokens": 619215988.0, + "step": 15951 + }, + { + "epoch": 2.0292583640758175, + "grad_norm": 3.081843376159668, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9039914011955261, + "num_tokens": 619251992.0, + "step": 15952 + }, + { + "epoch": 2.029385574354408, + "grad_norm": 3.012052297592163, + "learning_rate": 1e-06, + "loss": 0.2262, + "mean_token_accuracy": 0.9153016805648804, + "num_tokens": 619282867.0, + "step": 15953 + }, + { + "epoch": 2.0295127846329986, + "grad_norm": 2.6920793056488037, + "learning_rate": 1e-06, + "loss": 0.2219, + "mean_token_accuracy": 0.916649580001831, + "num_tokens": 619318944.0, + "step": 15954 + }, + { + "epoch": 2.0296399949115886, + "grad_norm": 2.624997615814209, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9009752869606018, + "num_tokens": 619362281.0, + "step": 15955 + }, + { + "epoch": 2.029767205190179, + "grad_norm": 3.00470232963562, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.893399715423584, + "num_tokens": 619403915.0, + "step": 15956 + }, + { + "epoch": 2.0298944154687697, + "grad_norm": 2.874786138534546, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9020794630050659, + "num_tokens": 619444419.0, + "step": 15957 + }, + { + "epoch": 2.0300216257473602, + "grad_norm": 3.060027837753296, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9007922410964966, + "num_tokens": 619482522.0, + "step": 15958 + }, + { + "epoch": 2.0301488360259508, + "grad_norm": 2.9448142051696777, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9057221412658691, + "num_tokens": 619518518.0, + "step": 15959 + }, + { + "epoch": 2.0302760463045413, + "grad_norm": 2.455138683319092, + "learning_rate": 1e-06, + "loss": 0.2237, + "mean_token_accuracy": 0.9138991832733154, + "num_tokens": 619562478.0, + "step": 15960 + }, + { + "epoch": 2.030403256583132, + "grad_norm": 2.7791006565093994, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9024733304977417, + "num_tokens": 619604348.0, + "step": 15961 + }, + { + "epoch": 2.0305304668617223, + "grad_norm": 2.7953357696533203, + "learning_rate": 1e-06, + "loss": 0.2308, + "mean_token_accuracy": 0.9159058928489685, + "num_tokens": 619640400.0, + "step": 15962 + }, + { + "epoch": 2.030657677140313, + "grad_norm": 2.939800977706909, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9016334414482117, + "num_tokens": 619676572.0, + "step": 15963 + }, + { + "epoch": 2.0307848874189034, + "grad_norm": 2.950951099395752, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9040290713310242, + "num_tokens": 619712973.0, + "step": 15964 + }, + { + "epoch": 2.030912097697494, + "grad_norm": 2.3915369510650635, + "learning_rate": 1e-06, + "loss": 0.2273, + "mean_token_accuracy": 0.9193404912948608, + "num_tokens": 619759725.0, + "step": 15965 + }, + { + "epoch": 2.0310393079760845, + "grad_norm": 3.0473225116729736, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8995336294174194, + "num_tokens": 619795722.0, + "step": 15966 + }, + { + "epoch": 2.031166518254675, + "grad_norm": 3.0314078330993652, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.8998907804489136, + "num_tokens": 619831347.0, + "step": 15967 + }, + { + "epoch": 2.0312937285332655, + "grad_norm": 2.89790415763855, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9051034450531006, + "num_tokens": 619867083.0, + "step": 15968 + }, + { + "epoch": 2.031420938811856, + "grad_norm": 3.1040751934051514, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9067744016647339, + "num_tokens": 619898229.0, + "step": 15969 + }, + { + "epoch": 2.0315481490904466, + "grad_norm": 3.63657808303833, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8990316390991211, + "num_tokens": 619938862.0, + "step": 15970 + }, + { + "epoch": 2.031675359369037, + "grad_norm": 3.149181365966797, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9064928889274597, + "num_tokens": 619971081.0, + "step": 15971 + }, + { + "epoch": 2.0318025696476276, + "grad_norm": 2.9260220527648926, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9097824096679688, + "num_tokens": 620008050.0, + "step": 15972 + }, + { + "epoch": 2.031929779926218, + "grad_norm": 2.480168104171753, + "learning_rate": 1e-06, + "loss": 0.2262, + "mean_token_accuracy": 0.914262592792511, + "num_tokens": 620053033.0, + "step": 15973 + }, + { + "epoch": 2.0320569902048087, + "grad_norm": 2.7403621673583984, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8975522518157959, + "num_tokens": 620096491.0, + "step": 15974 + }, + { + "epoch": 2.032184200483399, + "grad_norm": 2.8679678440093994, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9042921662330627, + "num_tokens": 620134734.0, + "step": 15975 + }, + { + "epoch": 2.0323114107619897, + "grad_norm": 2.9776833057403564, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.910321831703186, + "num_tokens": 620172782.0, + "step": 15976 + }, + { + "epoch": 2.0324386210405803, + "grad_norm": 3.0754013061523438, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9049600958824158, + "num_tokens": 620208507.0, + "step": 15977 + }, + { + "epoch": 2.032565831319171, + "grad_norm": 2.570646286010742, + "learning_rate": 1e-06, + "loss": 0.2317, + "mean_token_accuracy": 0.9125310778617859, + "num_tokens": 620250382.0, + "step": 15978 + }, + { + "epoch": 2.032693041597761, + "grad_norm": 2.925031900405884, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.911191999912262, + "num_tokens": 620284720.0, + "step": 15979 + }, + { + "epoch": 2.0328202518763514, + "grad_norm": 2.7584497928619385, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.9127422571182251, + "num_tokens": 620323931.0, + "step": 15980 + }, + { + "epoch": 2.032947462154942, + "grad_norm": 2.929877996444702, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9123052358627319, + "num_tokens": 620362967.0, + "step": 15981 + }, + { + "epoch": 2.0330746724335325, + "grad_norm": 2.8562700748443604, + "learning_rate": 1e-06, + "loss": 0.3144, + "mean_token_accuracy": 0.8878476023674011, + "num_tokens": 620411838.0, + "step": 15982 + }, + { + "epoch": 2.033201882712123, + "grad_norm": 3.12600040435791, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9112862348556519, + "num_tokens": 620446665.0, + "step": 15983 + }, + { + "epoch": 2.0333290929907135, + "grad_norm": 2.6821746826171875, + "learning_rate": 1e-06, + "loss": 0.2178, + "mean_token_accuracy": 0.9177188873291016, + "num_tokens": 620486320.0, + "step": 15984 + }, + { + "epoch": 2.033456303269304, + "grad_norm": 2.743135929107666, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9084619283676147, + "num_tokens": 620526231.0, + "step": 15985 + }, + { + "epoch": 2.0335835135478946, + "grad_norm": 2.7938077449798584, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.907888650894165, + "num_tokens": 620566043.0, + "step": 15986 + }, + { + "epoch": 2.033710723826485, + "grad_norm": 3.084473133087158, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9020734429359436, + "num_tokens": 620605145.0, + "step": 15987 + }, + { + "epoch": 2.0338379341050756, + "grad_norm": 2.870065450668335, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9017103910446167, + "num_tokens": 620643382.0, + "step": 15988 + }, + { + "epoch": 2.033965144383666, + "grad_norm": 2.902519702911377, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9048027396202087, + "num_tokens": 620682202.0, + "step": 15989 + }, + { + "epoch": 2.0340923546622567, + "grad_norm": 3.1432642936706543, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8951981067657471, + "num_tokens": 620715935.0, + "step": 15990 + }, + { + "epoch": 2.034219564940847, + "grad_norm": 2.5752973556518555, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9029019474983215, + "num_tokens": 620760739.0, + "step": 15991 + }, + { + "epoch": 2.0343467752194377, + "grad_norm": 2.8647773265838623, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9073617458343506, + "num_tokens": 620799539.0, + "step": 15992 + }, + { + "epoch": 2.0344739854980283, + "grad_norm": 3.0438761711120605, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8990201950073242, + "num_tokens": 620836087.0, + "step": 15993 + }, + { + "epoch": 2.034601195776619, + "grad_norm": 2.906562089920044, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.900566577911377, + "num_tokens": 620873361.0, + "step": 15994 + }, + { + "epoch": 2.0347284060552093, + "grad_norm": 2.7464301586151123, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.900637686252594, + "num_tokens": 620914326.0, + "step": 15995 + }, + { + "epoch": 2.0348556163338, + "grad_norm": 2.864837646484375, + "learning_rate": 1e-06, + "loss": 0.2416, + "mean_token_accuracy": 0.9089518785476685, + "num_tokens": 620948712.0, + "step": 15996 + }, + { + "epoch": 2.0349828266123904, + "grad_norm": 3.082348346710205, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9074745178222656, + "num_tokens": 620980836.0, + "step": 15997 + }, + { + "epoch": 2.035110036890981, + "grad_norm": 2.7950243949890137, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9127246737480164, + "num_tokens": 621017225.0, + "step": 15998 + }, + { + "epoch": 2.0352372471695714, + "grad_norm": 2.960502862930298, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9096157550811768, + "num_tokens": 621054201.0, + "step": 15999 + }, + { + "epoch": 2.035364457448162, + "grad_norm": 3.0221192836761475, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.8977091908454895, + "num_tokens": 621095014.0, + "step": 16000 + }, + { + "epoch": 2.0354916677267525, + "grad_norm": 2.5759711265563965, + "learning_rate": 1e-06, + "loss": 0.2308, + "mean_token_accuracy": 0.9145313501358032, + "num_tokens": 621138233.0, + "step": 16001 + }, + { + "epoch": 2.035618878005343, + "grad_norm": 2.4848873615264893, + "learning_rate": 1e-06, + "loss": 0.242, + "mean_token_accuracy": 0.9099497199058533, + "num_tokens": 621183719.0, + "step": 16002 + }, + { + "epoch": 2.0357460882839336, + "grad_norm": 3.086069345474243, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9083576202392578, + "num_tokens": 621217230.0, + "step": 16003 + }, + { + "epoch": 2.0358732985625236, + "grad_norm": 2.6327931880950928, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9009844064712524, + "num_tokens": 621259952.0, + "step": 16004 + }, + { + "epoch": 2.036000508841114, + "grad_norm": 2.9308042526245117, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.9006480574607849, + "num_tokens": 621299390.0, + "step": 16005 + }, + { + "epoch": 2.0361277191197047, + "grad_norm": 2.6189346313476562, + "learning_rate": 1e-06, + "loss": 0.2227, + "mean_token_accuracy": 0.9154387712478638, + "num_tokens": 621339666.0, + "step": 16006 + }, + { + "epoch": 2.036254929398295, + "grad_norm": 3.2126128673553467, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.91157466173172, + "num_tokens": 621368995.0, + "step": 16007 + }, + { + "epoch": 2.0363821396768857, + "grad_norm": 2.830585479736328, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9030822515487671, + "num_tokens": 621409217.0, + "step": 16008 + }, + { + "epoch": 2.0365093499554763, + "grad_norm": 2.7943508625030518, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9144949913024902, + "num_tokens": 621448920.0, + "step": 16009 + }, + { + "epoch": 2.036636560234067, + "grad_norm": 2.9962048530578613, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9043681621551514, + "num_tokens": 621488982.0, + "step": 16010 + }, + { + "epoch": 2.0367637705126573, + "grad_norm": 2.6841964721679688, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.8996337652206421, + "num_tokens": 621532089.0, + "step": 16011 + }, + { + "epoch": 2.036890980791248, + "grad_norm": 2.732621431350708, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8942492008209229, + "num_tokens": 621577891.0, + "step": 16012 + }, + { + "epoch": 2.0370181910698384, + "grad_norm": 3.0887346267700195, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9004142880439758, + "num_tokens": 621614350.0, + "step": 16013 + }, + { + "epoch": 2.037145401348429, + "grad_norm": 2.751309871673584, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9033931493759155, + "num_tokens": 621656550.0, + "step": 16014 + }, + { + "epoch": 2.0372726116270194, + "grad_norm": 2.7503602504730225, + "learning_rate": 1e-06, + "loss": 0.2218, + "mean_token_accuracy": 0.9179868102073669, + "num_tokens": 621692246.0, + "step": 16015 + }, + { + "epoch": 2.03739982190561, + "grad_norm": 2.785447835922241, + "learning_rate": 1e-06, + "loss": 0.2305, + "mean_token_accuracy": 0.9144477248191833, + "num_tokens": 621729614.0, + "step": 16016 + }, + { + "epoch": 2.0375270321842005, + "grad_norm": 2.857065439224243, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8980003595352173, + "num_tokens": 621769025.0, + "step": 16017 + }, + { + "epoch": 2.037654242462791, + "grad_norm": 2.9961366653442383, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.9157853126525879, + "num_tokens": 621801149.0, + "step": 16018 + }, + { + "epoch": 2.0377814527413816, + "grad_norm": 3.0816867351531982, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.8976058959960938, + "num_tokens": 621839606.0, + "step": 16019 + }, + { + "epoch": 2.037908663019972, + "grad_norm": 2.749264717102051, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9104674458503723, + "num_tokens": 621878537.0, + "step": 16020 + }, + { + "epoch": 2.0380358732985626, + "grad_norm": 3.0400991439819336, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8964558243751526, + "num_tokens": 621912551.0, + "step": 16021 + }, + { + "epoch": 2.038163083577153, + "grad_norm": 2.981680154800415, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8986384272575378, + "num_tokens": 621949612.0, + "step": 16022 + }, + { + "epoch": 2.0382902938557437, + "grad_norm": 2.7362048625946045, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9130640625953674, + "num_tokens": 621987267.0, + "step": 16023 + }, + { + "epoch": 2.038417504134334, + "grad_norm": 2.9000942707061768, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9151391983032227, + "num_tokens": 622022170.0, + "step": 16024 + }, + { + "epoch": 2.0385447144129247, + "grad_norm": 2.953965425491333, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9115753769874573, + "num_tokens": 622058094.0, + "step": 16025 + }, + { + "epoch": 2.0386719246915153, + "grad_norm": 2.93090558052063, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9010788202285767, + "num_tokens": 622096010.0, + "step": 16026 + }, + { + "epoch": 2.038799134970106, + "grad_norm": 3.2305123805999756, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9042163491249084, + "num_tokens": 622127941.0, + "step": 16027 + }, + { + "epoch": 2.0389263452486963, + "grad_norm": 2.4749648571014404, + "learning_rate": 1e-06, + "loss": 0.2376, + "mean_token_accuracy": 0.9119935035705566, + "num_tokens": 622174203.0, + "step": 16028 + }, + { + "epoch": 2.0390535555272864, + "grad_norm": 2.8424320220947266, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9098629951477051, + "num_tokens": 622211741.0, + "step": 16029 + }, + { + "epoch": 2.039180765805877, + "grad_norm": 3.1320934295654297, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9066336750984192, + "num_tokens": 622244594.0, + "step": 16030 + }, + { + "epoch": 2.0393079760844675, + "grad_norm": 2.737030267715454, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8970929384231567, + "num_tokens": 622291162.0, + "step": 16031 + }, + { + "epoch": 2.039435186363058, + "grad_norm": 2.810767650604248, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9047614336013794, + "num_tokens": 622333029.0, + "step": 16032 + }, + { + "epoch": 2.0395623966416485, + "grad_norm": 2.968881130218506, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9091578722000122, + "num_tokens": 622370892.0, + "step": 16033 + }, + { + "epoch": 2.039689606920239, + "grad_norm": 2.9870097637176514, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9121654033660889, + "num_tokens": 622405817.0, + "step": 16034 + }, + { + "epoch": 2.0398168171988296, + "grad_norm": 2.848426103591919, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9057989716529846, + "num_tokens": 622445307.0, + "step": 16035 + }, + { + "epoch": 2.03994402747742, + "grad_norm": 2.6957297325134277, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9107008576393127, + "num_tokens": 622488415.0, + "step": 16036 + }, + { + "epoch": 2.0400712377560106, + "grad_norm": 2.9443953037261963, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9017200469970703, + "num_tokens": 622527395.0, + "step": 16037 + }, + { + "epoch": 2.040198448034601, + "grad_norm": 2.7801320552825928, + "learning_rate": 1e-06, + "loss": 0.217, + "mean_token_accuracy": 0.9196164608001709, + "num_tokens": 622563904.0, + "step": 16038 + }, + { + "epoch": 2.0403256583131917, + "grad_norm": 2.9063491821289062, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9035629630088806, + "num_tokens": 622603417.0, + "step": 16039 + }, + { + "epoch": 2.040452868591782, + "grad_norm": 2.903266191482544, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9058707356452942, + "num_tokens": 622641036.0, + "step": 16040 + }, + { + "epoch": 2.0405800788703727, + "grad_norm": 2.754592180252075, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9089245796203613, + "num_tokens": 622677197.0, + "step": 16041 + }, + { + "epoch": 2.0407072891489633, + "grad_norm": 2.794952869415283, + "learning_rate": 1e-06, + "loss": 0.2121, + "mean_token_accuracy": 0.9186800122261047, + "num_tokens": 622711030.0, + "step": 16042 + }, + { + "epoch": 2.040834499427554, + "grad_norm": 3.00528621673584, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9065201282501221, + "num_tokens": 622747465.0, + "step": 16043 + }, + { + "epoch": 2.0409617097061443, + "grad_norm": 2.873347043991089, + "learning_rate": 1e-06, + "loss": 0.215, + "mean_token_accuracy": 0.9185466766357422, + "num_tokens": 622784446.0, + "step": 16044 + }, + { + "epoch": 2.041088919984735, + "grad_norm": 2.8808083534240723, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8963561058044434, + "num_tokens": 622826342.0, + "step": 16045 + }, + { + "epoch": 2.0412161302633254, + "grad_norm": 3.4009206295013428, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.907519519329071, + "num_tokens": 622853758.0, + "step": 16046 + }, + { + "epoch": 2.041343340541916, + "grad_norm": 2.924647092819214, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9156047105789185, + "num_tokens": 622890767.0, + "step": 16047 + }, + { + "epoch": 2.0414705508205064, + "grad_norm": 2.8406708240509033, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9059499502182007, + "num_tokens": 622929720.0, + "step": 16048 + }, + { + "epoch": 2.041597761099097, + "grad_norm": 2.6451964378356934, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9145994186401367, + "num_tokens": 622972114.0, + "step": 16049 + }, + { + "epoch": 2.0417249713776875, + "grad_norm": 2.7473771572113037, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9076064825057983, + "num_tokens": 623013767.0, + "step": 16050 + }, + { + "epoch": 2.041852181656278, + "grad_norm": 2.4685323238372803, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9090224504470825, + "num_tokens": 623061281.0, + "step": 16051 + }, + { + "epoch": 2.0419793919348685, + "grad_norm": 2.7273037433624268, + "learning_rate": 1e-06, + "loss": 0.2217, + "mean_token_accuracy": 0.9171192646026611, + "num_tokens": 623097506.0, + "step": 16052 + }, + { + "epoch": 2.0421066022134586, + "grad_norm": 2.755446195602417, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9082521796226501, + "num_tokens": 623134958.0, + "step": 16053 + }, + { + "epoch": 2.042233812492049, + "grad_norm": 3.0069475173950195, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8942013382911682, + "num_tokens": 623175596.0, + "step": 16054 + }, + { + "epoch": 2.0423610227706397, + "grad_norm": 2.9924442768096924, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9022547006607056, + "num_tokens": 623212122.0, + "step": 16055 + }, + { + "epoch": 2.04248823304923, + "grad_norm": 2.997180700302124, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8916344046592712, + "num_tokens": 623250490.0, + "step": 16056 + }, + { + "epoch": 2.0426154433278207, + "grad_norm": 2.8095171451568604, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9132686257362366, + "num_tokens": 623295383.0, + "step": 16057 + }, + { + "epoch": 2.0427426536064113, + "grad_norm": 2.4654972553253174, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9112786054611206, + "num_tokens": 623344963.0, + "step": 16058 + }, + { + "epoch": 2.042869863885002, + "grad_norm": 3.0466806888580322, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9077786207199097, + "num_tokens": 623377709.0, + "step": 16059 + }, + { + "epoch": 2.0429970741635923, + "grad_norm": 2.783046007156372, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9145070314407349, + "num_tokens": 623414085.0, + "step": 16060 + }, + { + "epoch": 2.043124284442183, + "grad_norm": 2.608450412750244, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9111970067024231, + "num_tokens": 623455718.0, + "step": 16061 + }, + { + "epoch": 2.0432514947207734, + "grad_norm": 2.9774036407470703, + "learning_rate": 1e-06, + "loss": 0.2413, + "mean_token_accuracy": 0.9092310667037964, + "num_tokens": 623489046.0, + "step": 16062 + }, + { + "epoch": 2.043378704999364, + "grad_norm": 2.8013575077056885, + "learning_rate": 1e-06, + "loss": 0.2183, + "mean_token_accuracy": 0.9173295497894287, + "num_tokens": 623526831.0, + "step": 16063 + }, + { + "epoch": 2.0435059152779544, + "grad_norm": 2.9027273654937744, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.905788242816925, + "num_tokens": 623567151.0, + "step": 16064 + }, + { + "epoch": 2.043633125556545, + "grad_norm": 2.7689099311828613, + "learning_rate": 1e-06, + "loss": 0.213, + "mean_token_accuracy": 0.9187234044075012, + "num_tokens": 623603073.0, + "step": 16065 + }, + { + "epoch": 2.0437603358351355, + "grad_norm": 2.64890193939209, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.903852105140686, + "num_tokens": 623647043.0, + "step": 16066 + }, + { + "epoch": 2.043887546113726, + "grad_norm": 2.773961305618286, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9101467728614807, + "num_tokens": 623686768.0, + "step": 16067 + }, + { + "epoch": 2.0440147563923166, + "grad_norm": 3.0391476154327393, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9054645299911499, + "num_tokens": 623721775.0, + "step": 16068 + }, + { + "epoch": 2.044141966670907, + "grad_norm": 2.9578962326049805, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.898084282875061, + "num_tokens": 623763169.0, + "step": 16069 + }, + { + "epoch": 2.0442691769494976, + "grad_norm": 3.176752805709839, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9011548161506653, + "num_tokens": 623796089.0, + "step": 16070 + }, + { + "epoch": 2.044396387228088, + "grad_norm": 2.5985093116760254, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9121689796447754, + "num_tokens": 623841369.0, + "step": 16071 + }, + { + "epoch": 2.0445235975066787, + "grad_norm": 2.696690559387207, + "learning_rate": 1e-06, + "loss": 0.2176, + "mean_token_accuracy": 0.9213729500770569, + "num_tokens": 623880324.0, + "step": 16072 + }, + { + "epoch": 2.044650807785269, + "grad_norm": 2.777587652206421, + "learning_rate": 1e-06, + "loss": 0.2442, + "mean_token_accuracy": 0.911404550075531, + "num_tokens": 623920401.0, + "step": 16073 + }, + { + "epoch": 2.0447780180638597, + "grad_norm": 2.845698118209839, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9086843132972717, + "num_tokens": 623958263.0, + "step": 16074 + }, + { + "epoch": 2.0449052283424503, + "grad_norm": 2.7011024951934814, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9109694361686707, + "num_tokens": 624002561.0, + "step": 16075 + }, + { + "epoch": 2.045032438621041, + "grad_norm": 2.9059624671936035, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9030150175094604, + "num_tokens": 624039215.0, + "step": 16076 + }, + { + "epoch": 2.045159648899631, + "grad_norm": 3.8830831050872803, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9070383310317993, + "num_tokens": 624079417.0, + "step": 16077 + }, + { + "epoch": 2.0452868591782214, + "grad_norm": 3.0356829166412354, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8950463533401489, + "num_tokens": 624115585.0, + "step": 16078 + }, + { + "epoch": 2.045414069456812, + "grad_norm": 2.4939775466918945, + "learning_rate": 1e-06, + "loss": 0.2157, + "mean_token_accuracy": 0.9177121520042419, + "num_tokens": 624157782.0, + "step": 16079 + }, + { + "epoch": 2.0455412797354025, + "grad_norm": 3.3587000370025635, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9036846160888672, + "num_tokens": 624192240.0, + "step": 16080 + }, + { + "epoch": 2.045668490013993, + "grad_norm": 3.1752071380615234, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9109053015708923, + "num_tokens": 624224356.0, + "step": 16081 + }, + { + "epoch": 2.0457957002925835, + "grad_norm": 3.139620304107666, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8975880146026611, + "num_tokens": 624260913.0, + "step": 16082 + }, + { + "epoch": 2.045922910571174, + "grad_norm": 2.828766345977783, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9115474224090576, + "num_tokens": 624295190.0, + "step": 16083 + }, + { + "epoch": 2.0460501208497646, + "grad_norm": 2.711977005004883, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9105578660964966, + "num_tokens": 624334622.0, + "step": 16084 + }, + { + "epoch": 2.046177331128355, + "grad_norm": 2.8256919384002686, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9103958606719971, + "num_tokens": 624372651.0, + "step": 16085 + }, + { + "epoch": 2.0463045414069456, + "grad_norm": 2.8872992992401123, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9020833373069763, + "num_tokens": 624414293.0, + "step": 16086 + }, + { + "epoch": 2.046431751685536, + "grad_norm": 2.8597054481506348, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9067258834838867, + "num_tokens": 624449117.0, + "step": 16087 + }, + { + "epoch": 2.0465589619641267, + "grad_norm": 2.921243667602539, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9088966846466064, + "num_tokens": 624485251.0, + "step": 16088 + }, + { + "epoch": 2.046686172242717, + "grad_norm": 2.713400363922119, + "learning_rate": 1e-06, + "loss": 0.2309, + "mean_token_accuracy": 0.9123395681381226, + "num_tokens": 624524292.0, + "step": 16089 + }, + { + "epoch": 2.0468133825213077, + "grad_norm": 2.6468029022216797, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9131492376327515, + "num_tokens": 624570813.0, + "step": 16090 + }, + { + "epoch": 2.0469405927998983, + "grad_norm": 2.7203071117401123, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.903130292892456, + "num_tokens": 624615036.0, + "step": 16091 + }, + { + "epoch": 2.047067803078489, + "grad_norm": 2.6805520057678223, + "learning_rate": 1e-06, + "loss": 0.2243, + "mean_token_accuracy": 0.9155482053756714, + "num_tokens": 624654161.0, + "step": 16092 + }, + { + "epoch": 2.0471950133570793, + "grad_norm": 2.8479063510894775, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9026219248771667, + "num_tokens": 624696594.0, + "step": 16093 + }, + { + "epoch": 2.04732222363567, + "grad_norm": 2.977116584777832, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9067442417144775, + "num_tokens": 624733131.0, + "step": 16094 + }, + { + "epoch": 2.0474494339142604, + "grad_norm": 3.0099315643310547, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9090930223464966, + "num_tokens": 624768331.0, + "step": 16095 + }, + { + "epoch": 2.047576644192851, + "grad_norm": 2.8546969890594482, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9052489995956421, + "num_tokens": 624808734.0, + "step": 16096 + }, + { + "epoch": 2.0477038544714414, + "grad_norm": 2.7812905311584473, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9116944670677185, + "num_tokens": 624847329.0, + "step": 16097 + }, + { + "epoch": 2.047831064750032, + "grad_norm": 2.939039468765259, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.908134937286377, + "num_tokens": 624881702.0, + "step": 16098 + }, + { + "epoch": 2.0479582750286225, + "grad_norm": 2.660022735595703, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9065582156181335, + "num_tokens": 624921725.0, + "step": 16099 + }, + { + "epoch": 2.048085485307213, + "grad_norm": 2.975031614303589, + "learning_rate": 1e-06, + "loss": 0.2241, + "mean_token_accuracy": 0.9180629253387451, + "num_tokens": 624954769.0, + "step": 16100 + }, + { + "epoch": 2.0482126955858035, + "grad_norm": 2.9346420764923096, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.904492974281311, + "num_tokens": 624990370.0, + "step": 16101 + }, + { + "epoch": 2.0483399058643936, + "grad_norm": 2.6937432289123535, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9095091223716736, + "num_tokens": 625030399.0, + "step": 16102 + }, + { + "epoch": 2.048467116142984, + "grad_norm": 2.9173991680145264, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9012699127197266, + "num_tokens": 625070298.0, + "step": 16103 + }, + { + "epoch": 2.0485943264215747, + "grad_norm": 2.858504295349121, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.8990281224250793, + "num_tokens": 625108874.0, + "step": 16104 + }, + { + "epoch": 2.048721536700165, + "grad_norm": 3.0525519847869873, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8977358341217041, + "num_tokens": 625146006.0, + "step": 16105 + }, + { + "epoch": 2.0488487469787557, + "grad_norm": 2.83129620552063, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9120802879333496, + "num_tokens": 625180149.0, + "step": 16106 + }, + { + "epoch": 2.0489759572573463, + "grad_norm": 2.7108380794525146, + "learning_rate": 1e-06, + "loss": 0.2156, + "mean_token_accuracy": 0.9203733801841736, + "num_tokens": 625218403.0, + "step": 16107 + }, + { + "epoch": 2.049103167535937, + "grad_norm": 3.0675017833709717, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9060258865356445, + "num_tokens": 625255046.0, + "step": 16108 + }, + { + "epoch": 2.0492303778145273, + "grad_norm": 3.1517484188079834, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9065912961959839, + "num_tokens": 625290754.0, + "step": 16109 + }, + { + "epoch": 2.049357588093118, + "grad_norm": 3.0777859687805176, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9025940895080566, + "num_tokens": 625326839.0, + "step": 16110 + }, + { + "epoch": 2.0494847983717084, + "grad_norm": 2.7837047576904297, + "learning_rate": 1e-06, + "loss": 0.2442, + "mean_token_accuracy": 0.9103841781616211, + "num_tokens": 625364429.0, + "step": 16111 + }, + { + "epoch": 2.049612008650299, + "grad_norm": 2.574516534805298, + "learning_rate": 1e-06, + "loss": 0.215, + "mean_token_accuracy": 0.9191918969154358, + "num_tokens": 625407048.0, + "step": 16112 + }, + { + "epoch": 2.0497392189288894, + "grad_norm": 2.8927574157714844, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.8998406529426575, + "num_tokens": 625448515.0, + "step": 16113 + }, + { + "epoch": 2.04986642920748, + "grad_norm": 3.3676297664642334, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9018044471740723, + "num_tokens": 625481741.0, + "step": 16114 + }, + { + "epoch": 2.0499936394860705, + "grad_norm": 2.9001259803771973, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9021091461181641, + "num_tokens": 625521536.0, + "step": 16115 + }, + { + "epoch": 2.050120849764661, + "grad_norm": 2.620284080505371, + "learning_rate": 1e-06, + "loss": 0.2174, + "mean_token_accuracy": 0.9186502695083618, + "num_tokens": 625561808.0, + "step": 16116 + }, + { + "epoch": 2.0502480600432516, + "grad_norm": 2.9955360889434814, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9035662412643433, + "num_tokens": 625598710.0, + "step": 16117 + }, + { + "epoch": 2.050375270321842, + "grad_norm": 2.8726253509521484, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.911202609539032, + "num_tokens": 625637903.0, + "step": 16118 + }, + { + "epoch": 2.0505024806004326, + "grad_norm": 2.783438205718994, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9100512266159058, + "num_tokens": 625678265.0, + "step": 16119 + }, + { + "epoch": 2.050629690879023, + "grad_norm": 3.041151285171509, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9003509283065796, + "num_tokens": 625717318.0, + "step": 16120 + }, + { + "epoch": 2.0507569011576137, + "grad_norm": 2.962907314300537, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8932150602340698, + "num_tokens": 625757688.0, + "step": 16121 + }, + { + "epoch": 2.050884111436204, + "grad_norm": 2.865138292312622, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9078620672225952, + "num_tokens": 625796424.0, + "step": 16122 + }, + { + "epoch": 2.0510113217147947, + "grad_norm": 2.634458303451538, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9088486433029175, + "num_tokens": 625835930.0, + "step": 16123 + }, + { + "epoch": 2.0511385319933853, + "grad_norm": 2.913518190383911, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9009213447570801, + "num_tokens": 625873153.0, + "step": 16124 + }, + { + "epoch": 2.051265742271976, + "grad_norm": 2.6903879642486572, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.911690354347229, + "num_tokens": 625913227.0, + "step": 16125 + }, + { + "epoch": 2.0513929525505663, + "grad_norm": 2.4952356815338135, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9110845923423767, + "num_tokens": 625957810.0, + "step": 16126 + }, + { + "epoch": 2.0515201628291564, + "grad_norm": 3.0460832118988037, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.907960057258606, + "num_tokens": 625995002.0, + "step": 16127 + }, + { + "epoch": 2.051647373107747, + "grad_norm": 3.033242702484131, + "learning_rate": 1e-06, + "loss": 0.2255, + "mean_token_accuracy": 0.9129979610443115, + "num_tokens": 626027613.0, + "step": 16128 + }, + { + "epoch": 2.0517745833863374, + "grad_norm": 2.6830451488494873, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9089834094047546, + "num_tokens": 626069092.0, + "step": 16129 + }, + { + "epoch": 2.051901793664928, + "grad_norm": 2.9910290241241455, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.896190881729126, + "num_tokens": 626106319.0, + "step": 16130 + }, + { + "epoch": 2.0520290039435185, + "grad_norm": 2.7527318000793457, + "learning_rate": 1e-06, + "loss": 0.2277, + "mean_token_accuracy": 0.9156301021575928, + "num_tokens": 626145031.0, + "step": 16131 + }, + { + "epoch": 2.052156214222109, + "grad_norm": 2.564934730529785, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9065006971359253, + "num_tokens": 626187892.0, + "step": 16132 + }, + { + "epoch": 2.0522834245006996, + "grad_norm": 2.6176350116729736, + "learning_rate": 1e-06, + "loss": 0.2339, + "mean_token_accuracy": 0.9120135307312012, + "num_tokens": 626233861.0, + "step": 16133 + }, + { + "epoch": 2.05241063477929, + "grad_norm": 2.8022377490997314, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9056280851364136, + "num_tokens": 626272823.0, + "step": 16134 + }, + { + "epoch": 2.0525378450578806, + "grad_norm": 2.865233898162842, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8944917321205139, + "num_tokens": 626314135.0, + "step": 16135 + }, + { + "epoch": 2.052665055336471, + "grad_norm": 2.542548894882202, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.9160498380661011, + "num_tokens": 626355908.0, + "step": 16136 + }, + { + "epoch": 2.0527922656150617, + "grad_norm": 3.236948251724243, + "learning_rate": 1e-06, + "loss": 0.335, + "mean_token_accuracy": 0.8774576783180237, + "num_tokens": 626392238.0, + "step": 16137 + }, + { + "epoch": 2.052919475893652, + "grad_norm": 2.679121255874634, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.909261167049408, + "num_tokens": 626433145.0, + "step": 16138 + }, + { + "epoch": 2.0530466861722427, + "grad_norm": 2.572566270828247, + "learning_rate": 1e-06, + "loss": 0.2222, + "mean_token_accuracy": 0.9182536602020264, + "num_tokens": 626473203.0, + "step": 16139 + }, + { + "epoch": 2.0531738964508333, + "grad_norm": 2.8787362575531006, + "learning_rate": 1e-06, + "loss": 0.2261, + "mean_token_accuracy": 0.9159832000732422, + "num_tokens": 626507906.0, + "step": 16140 + }, + { + "epoch": 2.053301106729424, + "grad_norm": 2.958146572113037, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.8997811079025269, + "num_tokens": 626545729.0, + "step": 16141 + }, + { + "epoch": 2.0534283170080143, + "grad_norm": 3.0205039978027344, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.897588849067688, + "num_tokens": 626582267.0, + "step": 16142 + }, + { + "epoch": 2.053555527286605, + "grad_norm": 2.6332387924194336, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9045072793960571, + "num_tokens": 626625530.0, + "step": 16143 + }, + { + "epoch": 2.0536827375651954, + "grad_norm": 2.9335453510284424, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9040749669075012, + "num_tokens": 626661354.0, + "step": 16144 + }, + { + "epoch": 2.053809947843786, + "grad_norm": 2.7291343212127686, + "learning_rate": 1e-06, + "loss": 0.2309, + "mean_token_accuracy": 0.9142546653747559, + "num_tokens": 626699834.0, + "step": 16145 + }, + { + "epoch": 2.0539371581223764, + "grad_norm": 2.7457118034362793, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9033988118171692, + "num_tokens": 626743467.0, + "step": 16146 + }, + { + "epoch": 2.054064368400967, + "grad_norm": 2.6273674964904785, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9044384956359863, + "num_tokens": 626791334.0, + "step": 16147 + }, + { + "epoch": 2.0541915786795575, + "grad_norm": 2.7217459678649902, + "learning_rate": 1e-06, + "loss": 0.2223, + "mean_token_accuracy": 0.918708324432373, + "num_tokens": 626826617.0, + "step": 16148 + }, + { + "epoch": 2.054318788958148, + "grad_norm": 2.983792304992676, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.8984531164169312, + "num_tokens": 626860861.0, + "step": 16149 + }, + { + "epoch": 2.0544459992367385, + "grad_norm": 2.7859067916870117, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9129699468612671, + "num_tokens": 626903030.0, + "step": 16150 + }, + { + "epoch": 2.0545732095153286, + "grad_norm": 3.0940310955047607, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8977380990982056, + "num_tokens": 626940718.0, + "step": 16151 + }, + { + "epoch": 2.054700419793919, + "grad_norm": 2.832714796066284, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.8962448835372925, + "num_tokens": 626982765.0, + "step": 16152 + }, + { + "epoch": 2.0548276300725097, + "grad_norm": 2.5323853492736816, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9145195484161377, + "num_tokens": 627026657.0, + "step": 16153 + }, + { + "epoch": 2.0549548403511, + "grad_norm": 2.9983015060424805, + "learning_rate": 1e-06, + "loss": 0.2376, + "mean_token_accuracy": 0.915715217590332, + "num_tokens": 627060449.0, + "step": 16154 + }, + { + "epoch": 2.0550820506296907, + "grad_norm": 2.8333427906036377, + "learning_rate": 1e-06, + "loss": 0.2246, + "mean_token_accuracy": 0.9139142036437988, + "num_tokens": 627095210.0, + "step": 16155 + }, + { + "epoch": 2.0552092609082813, + "grad_norm": 3.0801799297332764, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.8990780711174011, + "num_tokens": 627131698.0, + "step": 16156 + }, + { + "epoch": 2.055336471186872, + "grad_norm": 2.7081105709075928, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9118072390556335, + "num_tokens": 627172136.0, + "step": 16157 + }, + { + "epoch": 2.0554636814654623, + "grad_norm": 2.7925446033477783, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8934082984924316, + "num_tokens": 627218343.0, + "step": 16158 + }, + { + "epoch": 2.055590891744053, + "grad_norm": 2.922915458679199, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9124382734298706, + "num_tokens": 627256985.0, + "step": 16159 + }, + { + "epoch": 2.0557181020226434, + "grad_norm": 2.812190532684326, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.8999620676040649, + "num_tokens": 627300042.0, + "step": 16160 + }, + { + "epoch": 2.055845312301234, + "grad_norm": 2.7327163219451904, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9110432863235474, + "num_tokens": 627341653.0, + "step": 16161 + }, + { + "epoch": 2.0559725225798244, + "grad_norm": 2.8825881481170654, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8896302580833435, + "num_tokens": 627386769.0, + "step": 16162 + }, + { + "epoch": 2.056099732858415, + "grad_norm": 2.963416337966919, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9124301671981812, + "num_tokens": 627420832.0, + "step": 16163 + }, + { + "epoch": 2.0562269431370055, + "grad_norm": 2.8775691986083984, + "learning_rate": 1e-06, + "loss": 0.2425, + "mean_token_accuracy": 0.9088419079780579, + "num_tokens": 627453147.0, + "step": 16164 + }, + { + "epoch": 2.056354153415596, + "grad_norm": 2.768087148666382, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9011995792388916, + "num_tokens": 627495628.0, + "step": 16165 + }, + { + "epoch": 2.0564813636941865, + "grad_norm": 2.641165018081665, + "learning_rate": 1e-06, + "loss": 0.229, + "mean_token_accuracy": 0.9153296947479248, + "num_tokens": 627538984.0, + "step": 16166 + }, + { + "epoch": 2.056608573972777, + "grad_norm": 2.798314332962036, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.907066822052002, + "num_tokens": 627577655.0, + "step": 16167 + }, + { + "epoch": 2.0567357842513676, + "grad_norm": 2.8033440113067627, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9132862091064453, + "num_tokens": 627617226.0, + "step": 16168 + }, + { + "epoch": 2.056862994529958, + "grad_norm": 3.1817121505737305, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9016022682189941, + "num_tokens": 627649001.0, + "step": 16169 + }, + { + "epoch": 2.0569902048085487, + "grad_norm": 2.926787853240967, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9121582508087158, + "num_tokens": 627686806.0, + "step": 16170 + }, + { + "epoch": 2.057117415087139, + "grad_norm": 2.912782669067383, + "learning_rate": 1e-06, + "loss": 0.2249, + "mean_token_accuracy": 0.9179275035858154, + "num_tokens": 627722100.0, + "step": 16171 + }, + { + "epoch": 2.0572446253657297, + "grad_norm": 2.7484512329101562, + "learning_rate": 1e-06, + "loss": 0.2124, + "mean_token_accuracy": 0.92045658826828, + "num_tokens": 627754822.0, + "step": 16172 + }, + { + "epoch": 2.0573718356443202, + "grad_norm": 2.73288631439209, + "learning_rate": 1e-06, + "loss": 0.2129, + "mean_token_accuracy": 0.920353889465332, + "num_tokens": 627792365.0, + "step": 16173 + }, + { + "epoch": 2.0574990459229108, + "grad_norm": 3.0172784328460693, + "learning_rate": 1e-06, + "loss": 0.2425, + "mean_token_accuracy": 0.910099446773529, + "num_tokens": 627826735.0, + "step": 16174 + }, + { + "epoch": 2.057626256201501, + "grad_norm": 2.997576951980591, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.898444652557373, + "num_tokens": 627869757.0, + "step": 16175 + }, + { + "epoch": 2.0577534664800914, + "grad_norm": 3.0814406871795654, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9040396809577942, + "num_tokens": 627908178.0, + "step": 16176 + }, + { + "epoch": 2.057880676758682, + "grad_norm": 2.7469964027404785, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9118683338165283, + "num_tokens": 627948515.0, + "step": 16177 + }, + { + "epoch": 2.0580078870372724, + "grad_norm": 2.6269619464874268, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9067330956459045, + "num_tokens": 627990011.0, + "step": 16178 + }, + { + "epoch": 2.058135097315863, + "grad_norm": 3.389821767807007, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9036413431167603, + "num_tokens": 628018922.0, + "step": 16179 + }, + { + "epoch": 2.0582623075944535, + "grad_norm": 2.9526710510253906, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8974650502204895, + "num_tokens": 628057253.0, + "step": 16180 + }, + { + "epoch": 2.058389517873044, + "grad_norm": 3.0824239253997803, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8964108228683472, + "num_tokens": 628094805.0, + "step": 16181 + }, + { + "epoch": 2.0585167281516346, + "grad_norm": 2.7342007160186768, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9002247452735901, + "num_tokens": 628138323.0, + "step": 16182 + }, + { + "epoch": 2.058643938430225, + "grad_norm": 2.964719533920288, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9077476859092712, + "num_tokens": 628176051.0, + "step": 16183 + }, + { + "epoch": 2.0587711487088156, + "grad_norm": 2.92533278465271, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9028375744819641, + "num_tokens": 628212516.0, + "step": 16184 + }, + { + "epoch": 2.058898358987406, + "grad_norm": 2.8729302883148193, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8984265327453613, + "num_tokens": 628255024.0, + "step": 16185 + }, + { + "epoch": 2.0590255692659967, + "grad_norm": 3.1561009883880615, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9042903184890747, + "num_tokens": 628288419.0, + "step": 16186 + }, + { + "epoch": 2.059152779544587, + "grad_norm": 2.794389009475708, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9104428291320801, + "num_tokens": 628329059.0, + "step": 16187 + }, + { + "epoch": 2.0592799898231777, + "grad_norm": 3.084336519241333, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9013310074806213, + "num_tokens": 628365318.0, + "step": 16188 + }, + { + "epoch": 2.0594072001017683, + "grad_norm": 2.751267433166504, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9048925638198853, + "num_tokens": 628408022.0, + "step": 16189 + }, + { + "epoch": 2.059534410380359, + "grad_norm": 2.674623727798462, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9051295518875122, + "num_tokens": 628451867.0, + "step": 16190 + }, + { + "epoch": 2.0596616206589493, + "grad_norm": 2.58135724067688, + "learning_rate": 1e-06, + "loss": 0.2303, + "mean_token_accuracy": 0.9135817885398865, + "num_tokens": 628494856.0, + "step": 16191 + }, + { + "epoch": 2.05978883093754, + "grad_norm": 2.7138540744781494, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.8991153240203857, + "num_tokens": 628536573.0, + "step": 16192 + }, + { + "epoch": 2.0599160412161304, + "grad_norm": 2.7077126502990723, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9122700691223145, + "num_tokens": 628574303.0, + "step": 16193 + }, + { + "epoch": 2.060043251494721, + "grad_norm": 2.8247203826904297, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9105226993560791, + "num_tokens": 628613508.0, + "step": 16194 + }, + { + "epoch": 2.0601704617733114, + "grad_norm": 2.8482487201690674, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9027572274208069, + "num_tokens": 628658748.0, + "step": 16195 + }, + { + "epoch": 2.060297672051902, + "grad_norm": 3.0940327644348145, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9012379050254822, + "num_tokens": 628692665.0, + "step": 16196 + }, + { + "epoch": 2.0604248823304925, + "grad_norm": 2.9454867839813232, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8951319456100464, + "num_tokens": 628732115.0, + "step": 16197 + }, + { + "epoch": 2.060552092609083, + "grad_norm": 2.8589437007904053, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9048926830291748, + "num_tokens": 628771417.0, + "step": 16198 + }, + { + "epoch": 2.0606793028876735, + "grad_norm": 2.949305534362793, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.900853157043457, + "num_tokens": 628810586.0, + "step": 16199 + }, + { + "epoch": 2.0608065131662636, + "grad_norm": 2.6975057125091553, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.911871075630188, + "num_tokens": 628851034.0, + "step": 16200 + }, + { + "epoch": 2.060933723444854, + "grad_norm": 2.5669872760772705, + "learning_rate": 1e-06, + "loss": 0.222, + "mean_token_accuracy": 0.91650390625, + "num_tokens": 628890590.0, + "step": 16201 + }, + { + "epoch": 2.0610609337234447, + "grad_norm": 2.7100934982299805, + "learning_rate": 1e-06, + "loss": 0.2443, + "mean_token_accuracy": 0.9071803092956543, + "num_tokens": 628930927.0, + "step": 16202 + }, + { + "epoch": 2.061188144002035, + "grad_norm": 2.7624671459198, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9040085077285767, + "num_tokens": 628972679.0, + "step": 16203 + }, + { + "epoch": 2.0613153542806257, + "grad_norm": 2.990055561065674, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8965842723846436, + "num_tokens": 629010245.0, + "step": 16204 + }, + { + "epoch": 2.0614425645592163, + "grad_norm": 2.9833221435546875, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9126765727996826, + "num_tokens": 629041980.0, + "step": 16205 + }, + { + "epoch": 2.061569774837807, + "grad_norm": 2.891072988510132, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9045716524124146, + "num_tokens": 629086285.0, + "step": 16206 + }, + { + "epoch": 2.0616969851163973, + "grad_norm": 2.789252996444702, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9047426581382751, + "num_tokens": 629123277.0, + "step": 16207 + }, + { + "epoch": 2.061824195394988, + "grad_norm": 2.947707176208496, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8981560468673706, + "num_tokens": 629162145.0, + "step": 16208 + }, + { + "epoch": 2.0619514056735784, + "grad_norm": 2.6508331298828125, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9095768928527832, + "num_tokens": 629203321.0, + "step": 16209 + }, + { + "epoch": 2.062078615952169, + "grad_norm": 2.8748488426208496, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9090716242790222, + "num_tokens": 629242203.0, + "step": 16210 + }, + { + "epoch": 2.0622058262307594, + "grad_norm": 2.8170864582061768, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8998789191246033, + "num_tokens": 629284436.0, + "step": 16211 + }, + { + "epoch": 2.06233303650935, + "grad_norm": 2.801786422729492, + "learning_rate": 1e-06, + "loss": 0.2289, + "mean_token_accuracy": 0.9134789705276489, + "num_tokens": 629320349.0, + "step": 16212 + }, + { + "epoch": 2.0624602467879405, + "grad_norm": 2.6685619354248047, + "learning_rate": 1e-06, + "loss": 0.2263, + "mean_token_accuracy": 0.9150986671447754, + "num_tokens": 629360995.0, + "step": 16213 + }, + { + "epoch": 2.062587457066531, + "grad_norm": 2.8836092948913574, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8966097831726074, + "num_tokens": 629403189.0, + "step": 16214 + }, + { + "epoch": 2.0627146673451215, + "grad_norm": 2.7316806316375732, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9126039743423462, + "num_tokens": 629438478.0, + "step": 16215 + }, + { + "epoch": 2.062841877623712, + "grad_norm": 2.753812313079834, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8974387049674988, + "num_tokens": 629485056.0, + "step": 16216 + }, + { + "epoch": 2.0629690879023026, + "grad_norm": 2.915642499923706, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9005606174468994, + "num_tokens": 629523834.0, + "step": 16217 + }, + { + "epoch": 2.063096298180893, + "grad_norm": 2.7256104946136475, + "learning_rate": 1e-06, + "loss": 0.2398, + "mean_token_accuracy": 0.9105124473571777, + "num_tokens": 629565884.0, + "step": 16218 + }, + { + "epoch": 2.0632235084594837, + "grad_norm": 2.6308369636535645, + "learning_rate": 1e-06, + "loss": 0.2342, + "mean_token_accuracy": 0.9145042896270752, + "num_tokens": 629606057.0, + "step": 16219 + }, + { + "epoch": 2.063350718738074, + "grad_norm": 3.1880080699920654, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9075808525085449, + "num_tokens": 629636642.0, + "step": 16220 + }, + { + "epoch": 2.0634779290166647, + "grad_norm": 3.172964334487915, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.906638503074646, + "num_tokens": 629669189.0, + "step": 16221 + }, + { + "epoch": 2.0636051392952552, + "grad_norm": 2.7613253593444824, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.90712571144104, + "num_tokens": 629708732.0, + "step": 16222 + }, + { + "epoch": 2.0637323495738458, + "grad_norm": 3.091545343399048, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8999885320663452, + "num_tokens": 629744952.0, + "step": 16223 + }, + { + "epoch": 2.0638595598524363, + "grad_norm": 2.7027926445007324, + "learning_rate": 1e-06, + "loss": 0.2273, + "mean_token_accuracy": 0.9176188707351685, + "num_tokens": 629782928.0, + "step": 16224 + }, + { + "epoch": 2.0639867701310264, + "grad_norm": 2.9204256534576416, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9079597592353821, + "num_tokens": 629816926.0, + "step": 16225 + }, + { + "epoch": 2.064113980409617, + "grad_norm": 3.237511157989502, + "learning_rate": 1e-06, + "loss": 0.2186, + "mean_token_accuracy": 0.9168812036514282, + "num_tokens": 629846730.0, + "step": 16226 + }, + { + "epoch": 2.0642411906882074, + "grad_norm": 3.1940579414367676, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9011818170547485, + "num_tokens": 629880567.0, + "step": 16227 + }, + { + "epoch": 2.064368400966798, + "grad_norm": 3.275304079055786, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.893921434879303, + "num_tokens": 629913871.0, + "step": 16228 + }, + { + "epoch": 2.0644956112453885, + "grad_norm": 2.892632007598877, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9110454320907593, + "num_tokens": 629947253.0, + "step": 16229 + }, + { + "epoch": 2.064622821523979, + "grad_norm": 2.898048162460327, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9033384323120117, + "num_tokens": 629983671.0, + "step": 16230 + }, + { + "epoch": 2.0647500318025696, + "grad_norm": 3.03102707862854, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9084975719451904, + "num_tokens": 630017273.0, + "step": 16231 + }, + { + "epoch": 2.06487724208116, + "grad_norm": 2.8608508110046387, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9064999222755432, + "num_tokens": 630055497.0, + "step": 16232 + }, + { + "epoch": 2.0650044523597506, + "grad_norm": 2.545774459838867, + "learning_rate": 1e-06, + "loss": 0.2273, + "mean_token_accuracy": 0.9156291484832764, + "num_tokens": 630096282.0, + "step": 16233 + }, + { + "epoch": 2.065131662638341, + "grad_norm": 3.057927131652832, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9018820524215698, + "num_tokens": 630130802.0, + "step": 16234 + }, + { + "epoch": 2.0652588729169317, + "grad_norm": 2.9127020835876465, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.914455771446228, + "num_tokens": 630165867.0, + "step": 16235 + }, + { + "epoch": 2.065386083195522, + "grad_norm": 2.970947265625, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9003136157989502, + "num_tokens": 630201938.0, + "step": 16236 + }, + { + "epoch": 2.0655132934741127, + "grad_norm": 2.889836072921753, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9059024453163147, + "num_tokens": 630236467.0, + "step": 16237 + }, + { + "epoch": 2.0656405037527032, + "grad_norm": 3.0538887977600098, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9104404449462891, + "num_tokens": 630271543.0, + "step": 16238 + }, + { + "epoch": 2.065767714031294, + "grad_norm": 2.6620330810546875, + "learning_rate": 1e-06, + "loss": 0.2271, + "mean_token_accuracy": 0.9150452017784119, + "num_tokens": 630311017.0, + "step": 16239 + }, + { + "epoch": 2.0658949243098843, + "grad_norm": 3.030301094055176, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9068694114685059, + "num_tokens": 630344057.0, + "step": 16240 + }, + { + "epoch": 2.066022134588475, + "grad_norm": 2.831348180770874, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9064671993255615, + "num_tokens": 630384682.0, + "step": 16241 + }, + { + "epoch": 2.0661493448670654, + "grad_norm": 2.7351486682891846, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8961502313613892, + "num_tokens": 630428419.0, + "step": 16242 + }, + { + "epoch": 2.066276555145656, + "grad_norm": 2.9412903785705566, + "learning_rate": 1e-06, + "loss": 0.2132, + "mean_token_accuracy": 0.9204035401344299, + "num_tokens": 630459479.0, + "step": 16243 + }, + { + "epoch": 2.0664037654242464, + "grad_norm": 2.697469472885132, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9002818465232849, + "num_tokens": 630502030.0, + "step": 16244 + }, + { + "epoch": 2.066530975702837, + "grad_norm": 2.663233518600464, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9108933210372925, + "num_tokens": 630547472.0, + "step": 16245 + }, + { + "epoch": 2.0666581859814275, + "grad_norm": 2.8000683784484863, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9112285375595093, + "num_tokens": 630584663.0, + "step": 16246 + }, + { + "epoch": 2.066785396260018, + "grad_norm": 2.717043399810791, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9140027165412903, + "num_tokens": 630625068.0, + "step": 16247 + }, + { + "epoch": 2.0669126065386085, + "grad_norm": 2.9183290004730225, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9105422496795654, + "num_tokens": 630661969.0, + "step": 16248 + }, + { + "epoch": 2.0670398168171986, + "grad_norm": 3.2042341232299805, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.8996117115020752, + "num_tokens": 630696060.0, + "step": 16249 + }, + { + "epoch": 2.067167027095789, + "grad_norm": 2.836780309677124, + "learning_rate": 1e-06, + "loss": 0.2443, + "mean_token_accuracy": 0.9119625091552734, + "num_tokens": 630736523.0, + "step": 16250 + }, + { + "epoch": 2.0672942373743797, + "grad_norm": 2.8060710430145264, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9089471697807312, + "num_tokens": 630775498.0, + "step": 16251 + }, + { + "epoch": 2.06742144765297, + "grad_norm": 3.023599863052368, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9014794826507568, + "num_tokens": 630810734.0, + "step": 16252 + }, + { + "epoch": 2.0675486579315607, + "grad_norm": 2.712076187133789, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9051231145858765, + "num_tokens": 630854012.0, + "step": 16253 + }, + { + "epoch": 2.0676758682101513, + "grad_norm": 2.749846935272217, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9059100151062012, + "num_tokens": 630896553.0, + "step": 16254 + }, + { + "epoch": 2.067803078488742, + "grad_norm": 2.67041277885437, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9082967042922974, + "num_tokens": 630942045.0, + "step": 16255 + }, + { + "epoch": 2.0679302887673323, + "grad_norm": 3.3143296241760254, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9010922312736511, + "num_tokens": 630974039.0, + "step": 16256 + }, + { + "epoch": 2.068057499045923, + "grad_norm": 3.3061630725860596, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8865376710891724, + "num_tokens": 631012659.0, + "step": 16257 + }, + { + "epoch": 2.0681847093245134, + "grad_norm": 2.8158740997314453, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9057754874229431, + "num_tokens": 631052784.0, + "step": 16258 + }, + { + "epoch": 2.068311919603104, + "grad_norm": 2.8341760635375977, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9125246405601501, + "num_tokens": 631090710.0, + "step": 16259 + }, + { + "epoch": 2.0684391298816944, + "grad_norm": 2.851160764694214, + "learning_rate": 1e-06, + "loss": 0.2261, + "mean_token_accuracy": 0.9153195023536682, + "num_tokens": 631127050.0, + "step": 16260 + }, + { + "epoch": 2.068566340160285, + "grad_norm": 2.9842476844787598, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9048234224319458, + "num_tokens": 631163492.0, + "step": 16261 + }, + { + "epoch": 2.0686935504388755, + "grad_norm": 2.546888589859009, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9123725891113281, + "num_tokens": 631206570.0, + "step": 16262 + }, + { + "epoch": 2.068820760717466, + "grad_norm": 2.751112937927246, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.911668598651886, + "num_tokens": 631246431.0, + "step": 16263 + }, + { + "epoch": 2.0689479709960565, + "grad_norm": 2.738997220993042, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.916509747505188, + "num_tokens": 631284875.0, + "step": 16264 + }, + { + "epoch": 2.069075181274647, + "grad_norm": 2.8286633491516113, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8962509632110596, + "num_tokens": 631325308.0, + "step": 16265 + }, + { + "epoch": 2.0692023915532376, + "grad_norm": 3.2769525051116943, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8874629735946655, + "num_tokens": 631359552.0, + "step": 16266 + }, + { + "epoch": 2.069329601831828, + "grad_norm": 2.7036287784576416, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9065853953361511, + "num_tokens": 631402714.0, + "step": 16267 + }, + { + "epoch": 2.0694568121104187, + "grad_norm": 2.9913926124572754, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9003485441207886, + "num_tokens": 631441782.0, + "step": 16268 + }, + { + "epoch": 2.069584022389009, + "grad_norm": 2.8802602291107178, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9060940742492676, + "num_tokens": 631479605.0, + "step": 16269 + }, + { + "epoch": 2.0697112326675997, + "grad_norm": 2.804245710372925, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9087684750556946, + "num_tokens": 631518277.0, + "step": 16270 + }, + { + "epoch": 2.0698384429461902, + "grad_norm": 2.6242847442626953, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9013261795043945, + "num_tokens": 631565450.0, + "step": 16271 + }, + { + "epoch": 2.0699656532247808, + "grad_norm": 2.715287923812866, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9113054871559143, + "num_tokens": 631605423.0, + "step": 16272 + }, + { + "epoch": 2.070092863503371, + "grad_norm": 3.381676435470581, + "learning_rate": 1e-06, + "loss": 0.3312, + "mean_token_accuracy": 0.8833823800086975, + "num_tokens": 631638726.0, + "step": 16273 + }, + { + "epoch": 2.0702200737819614, + "grad_norm": 2.563908815383911, + "learning_rate": 1e-06, + "loss": 0.2142, + "mean_token_accuracy": 0.9217114448547363, + "num_tokens": 631680347.0, + "step": 16274 + }, + { + "epoch": 2.070347284060552, + "grad_norm": 2.9707977771759033, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.890274167060852, + "num_tokens": 631720987.0, + "step": 16275 + }, + { + "epoch": 2.0704744943391424, + "grad_norm": 2.7912676334381104, + "learning_rate": 1e-06, + "loss": 0.302, + "mean_token_accuracy": 0.8904446363449097, + "num_tokens": 631764639.0, + "step": 16276 + }, + { + "epoch": 2.070601704617733, + "grad_norm": 3.255110502243042, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8974478840827942, + "num_tokens": 631804628.0, + "step": 16277 + }, + { + "epoch": 2.0707289148963235, + "grad_norm": 2.865540027618408, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8981741666793823, + "num_tokens": 631849315.0, + "step": 16278 + }, + { + "epoch": 2.070856125174914, + "grad_norm": 2.5097134113311768, + "learning_rate": 1e-06, + "loss": 0.2279, + "mean_token_accuracy": 0.9149900674819946, + "num_tokens": 631894572.0, + "step": 16279 + }, + { + "epoch": 2.0709833354535045, + "grad_norm": 2.6173160076141357, + "learning_rate": 1e-06, + "loss": 0.2255, + "mean_token_accuracy": 0.9147325754165649, + "num_tokens": 631937147.0, + "step": 16280 + }, + { + "epoch": 2.071110545732095, + "grad_norm": 2.743210792541504, + "learning_rate": 1e-06, + "loss": 0.2345, + "mean_token_accuracy": 0.9119282960891724, + "num_tokens": 631976015.0, + "step": 16281 + }, + { + "epoch": 2.0712377560106856, + "grad_norm": 2.9630510807037354, + "learning_rate": 1e-06, + "loss": 0.2274, + "mean_token_accuracy": 0.9126639366149902, + "num_tokens": 632010267.0, + "step": 16282 + }, + { + "epoch": 2.071364966289276, + "grad_norm": 2.6849474906921387, + "learning_rate": 1e-06, + "loss": 0.2187, + "mean_token_accuracy": 0.9190065264701843, + "num_tokens": 632048827.0, + "step": 16283 + }, + { + "epoch": 2.0714921765678667, + "grad_norm": 3.4153242111206055, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9041770696640015, + "num_tokens": 632080933.0, + "step": 16284 + }, + { + "epoch": 2.071619386846457, + "grad_norm": 3.103895425796509, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9020256400108337, + "num_tokens": 632114217.0, + "step": 16285 + }, + { + "epoch": 2.0717465971250477, + "grad_norm": 3.022611618041992, + "learning_rate": 1e-06, + "loss": 0.3043, + "mean_token_accuracy": 0.8889581561088562, + "num_tokens": 632151072.0, + "step": 16286 + }, + { + "epoch": 2.0718738074036382, + "grad_norm": 2.7964694499969482, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9026384353637695, + "num_tokens": 632191864.0, + "step": 16287 + }, + { + "epoch": 2.0720010176822288, + "grad_norm": 2.8759186267852783, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9016189575195312, + "num_tokens": 632234226.0, + "step": 16288 + }, + { + "epoch": 2.0721282279608193, + "grad_norm": 3.118896961212158, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9066922664642334, + "num_tokens": 632268818.0, + "step": 16289 + }, + { + "epoch": 2.07225543823941, + "grad_norm": 3.2357349395751953, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8973391056060791, + "num_tokens": 632302418.0, + "step": 16290 + }, + { + "epoch": 2.0723826485180004, + "grad_norm": 2.6604979038238525, + "learning_rate": 1e-06, + "loss": 0.2246, + "mean_token_accuracy": 0.9151620864868164, + "num_tokens": 632343004.0, + "step": 16291 + }, + { + "epoch": 2.072509858796591, + "grad_norm": 2.9092812538146973, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9074759483337402, + "num_tokens": 632379583.0, + "step": 16292 + }, + { + "epoch": 2.0726370690751814, + "grad_norm": 2.8455660343170166, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9130443334579468, + "num_tokens": 632418211.0, + "step": 16293 + }, + { + "epoch": 2.072764279353772, + "grad_norm": 3.1477510929107666, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.8998119235038757, + "num_tokens": 632456739.0, + "step": 16294 + }, + { + "epoch": 2.0728914896323625, + "grad_norm": 2.8034305572509766, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9088767766952515, + "num_tokens": 632493388.0, + "step": 16295 + }, + { + "epoch": 2.073018699910953, + "grad_norm": 3.141451358795166, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9069879055023193, + "num_tokens": 632525885.0, + "step": 16296 + }, + { + "epoch": 2.0731459101895435, + "grad_norm": 2.612553119659424, + "learning_rate": 1e-06, + "loss": 0.2178, + "mean_token_accuracy": 0.919724702835083, + "num_tokens": 632565882.0, + "step": 16297 + }, + { + "epoch": 2.0732731204681336, + "grad_norm": 2.4232358932495117, + "learning_rate": 1e-06, + "loss": 0.2014, + "mean_token_accuracy": 0.9236714243888855, + "num_tokens": 632606776.0, + "step": 16298 + }, + { + "epoch": 2.073400330746724, + "grad_norm": 3.132978677749634, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9049168229103088, + "num_tokens": 632638857.0, + "step": 16299 + }, + { + "epoch": 2.0735275410253147, + "grad_norm": 2.77451491355896, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9073362350463867, + "num_tokens": 632681334.0, + "step": 16300 + }, + { + "epoch": 2.073654751303905, + "grad_norm": 2.813781499862671, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9109048843383789, + "num_tokens": 632719984.0, + "step": 16301 + }, + { + "epoch": 2.0737819615824957, + "grad_norm": 2.4129953384399414, + "learning_rate": 1e-06, + "loss": 0.2063, + "mean_token_accuracy": 0.9215167760848999, + "num_tokens": 632760262.0, + "step": 16302 + }, + { + "epoch": 2.0739091718610863, + "grad_norm": 2.7654244899749756, + "learning_rate": 1e-06, + "loss": 0.2349, + "mean_token_accuracy": 0.9107998013496399, + "num_tokens": 632800558.0, + "step": 16303 + }, + { + "epoch": 2.074036382139677, + "grad_norm": 2.9951417446136475, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8975434899330139, + "num_tokens": 632838224.0, + "step": 16304 + }, + { + "epoch": 2.0741635924182673, + "grad_norm": 3.0998897552490234, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8949490785598755, + "num_tokens": 632874047.0, + "step": 16305 + }, + { + "epoch": 2.074290802696858, + "grad_norm": 3.004612922668457, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8972564935684204, + "num_tokens": 632912043.0, + "step": 16306 + }, + { + "epoch": 2.0744180129754484, + "grad_norm": 2.6666126251220703, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9013742804527283, + "num_tokens": 632955999.0, + "step": 16307 + }, + { + "epoch": 2.074545223254039, + "grad_norm": 2.9491050243377686, + "learning_rate": 1e-06, + "loss": 0.2487, + "mean_token_accuracy": 0.9088829755783081, + "num_tokens": 632989170.0, + "step": 16308 + }, + { + "epoch": 2.0746724335326294, + "grad_norm": 2.8663079738616943, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.905359148979187, + "num_tokens": 633028974.0, + "step": 16309 + }, + { + "epoch": 2.07479964381122, + "grad_norm": 2.864245891571045, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.8997425436973572, + "num_tokens": 633071655.0, + "step": 16310 + }, + { + "epoch": 2.0749268540898105, + "grad_norm": 3.080260753631592, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9116500616073608, + "num_tokens": 633103379.0, + "step": 16311 + }, + { + "epoch": 2.075054064368401, + "grad_norm": 2.85046648979187, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9021192789077759, + "num_tokens": 633140788.0, + "step": 16312 + }, + { + "epoch": 2.0751812746469915, + "grad_norm": 3.0825164318084717, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9079981446266174, + "num_tokens": 633175613.0, + "step": 16313 + }, + { + "epoch": 2.075308484925582, + "grad_norm": 2.5744102001190186, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9016433954238892, + "num_tokens": 633222058.0, + "step": 16314 + }, + { + "epoch": 2.0754356952041726, + "grad_norm": 2.506439685821533, + "learning_rate": 1e-06, + "loss": 0.2236, + "mean_token_accuracy": 0.9182994365692139, + "num_tokens": 633264437.0, + "step": 16315 + }, + { + "epoch": 2.075562905482763, + "grad_norm": 2.851973533630371, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.8993930220603943, + "num_tokens": 633306076.0, + "step": 16316 + }, + { + "epoch": 2.0756901157613536, + "grad_norm": 3.1621508598327637, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8904651403427124, + "num_tokens": 633339804.0, + "step": 16317 + }, + { + "epoch": 2.075817326039944, + "grad_norm": 2.760284185409546, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9065933227539062, + "num_tokens": 633380263.0, + "step": 16318 + }, + { + "epoch": 2.0759445363185347, + "grad_norm": 2.615316867828369, + "learning_rate": 1e-06, + "loss": 0.2222, + "mean_token_accuracy": 0.917726457118988, + "num_tokens": 633421423.0, + "step": 16319 + }, + { + "epoch": 2.0760717465971252, + "grad_norm": 2.614614486694336, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9068203568458557, + "num_tokens": 633465317.0, + "step": 16320 + }, + { + "epoch": 2.0761989568757158, + "grad_norm": 3.228541851043701, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9059843420982361, + "num_tokens": 633501147.0, + "step": 16321 + }, + { + "epoch": 2.0763261671543063, + "grad_norm": 2.5616605281829834, + "learning_rate": 1e-06, + "loss": 0.2194, + "mean_token_accuracy": 0.9196611642837524, + "num_tokens": 633540137.0, + "step": 16322 + }, + { + "epoch": 2.0764533774328964, + "grad_norm": 2.8046860694885254, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9020814895629883, + "num_tokens": 633581925.0, + "step": 16323 + }, + { + "epoch": 2.076580587711487, + "grad_norm": 2.771880865097046, + "learning_rate": 1e-06, + "loss": 0.1875, + "mean_token_accuracy": 0.9277868270874023, + "num_tokens": 633614815.0, + "step": 16324 + }, + { + "epoch": 2.0767077979900774, + "grad_norm": 2.8565878868103027, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.9135896563529968, + "num_tokens": 633651808.0, + "step": 16325 + }, + { + "epoch": 2.076835008268668, + "grad_norm": 2.5436103343963623, + "learning_rate": 1e-06, + "loss": 0.2264, + "mean_token_accuracy": 0.914914608001709, + "num_tokens": 633693018.0, + "step": 16326 + }, + { + "epoch": 2.0769622185472585, + "grad_norm": 2.620821952819824, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9085816740989685, + "num_tokens": 633737450.0, + "step": 16327 + }, + { + "epoch": 2.077089428825849, + "grad_norm": 2.9833767414093018, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9056491255760193, + "num_tokens": 633776240.0, + "step": 16328 + }, + { + "epoch": 2.0772166391044395, + "grad_norm": 2.8966472148895264, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8992037773132324, + "num_tokens": 633814846.0, + "step": 16329 + }, + { + "epoch": 2.07734384938303, + "grad_norm": 2.751507520675659, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9057937264442444, + "num_tokens": 633855349.0, + "step": 16330 + }, + { + "epoch": 2.0774710596616206, + "grad_norm": 2.9296252727508545, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8949489593505859, + "num_tokens": 633892772.0, + "step": 16331 + }, + { + "epoch": 2.077598269940211, + "grad_norm": 2.680791139602661, + "learning_rate": 1e-06, + "loss": 0.22, + "mean_token_accuracy": 0.9177106618881226, + "num_tokens": 633931526.0, + "step": 16332 + }, + { + "epoch": 2.0777254802188017, + "grad_norm": 2.827448606491089, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9081069231033325, + "num_tokens": 633968181.0, + "step": 16333 + }, + { + "epoch": 2.077852690497392, + "grad_norm": 2.9112491607666016, + "learning_rate": 1e-06, + "loss": 0.2297, + "mean_token_accuracy": 0.9131479263305664, + "num_tokens": 634002766.0, + "step": 16334 + }, + { + "epoch": 2.0779799007759827, + "grad_norm": 2.742849588394165, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9093756675720215, + "num_tokens": 634043991.0, + "step": 16335 + }, + { + "epoch": 2.0781071110545732, + "grad_norm": 2.883523941040039, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.8993576765060425, + "num_tokens": 634085874.0, + "step": 16336 + }, + { + "epoch": 2.0782343213331638, + "grad_norm": 3.035809278488159, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.898483157157898, + "num_tokens": 634124352.0, + "step": 16337 + }, + { + "epoch": 2.0783615316117543, + "grad_norm": 3.072861671447754, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9140549898147583, + "num_tokens": 634157486.0, + "step": 16338 + }, + { + "epoch": 2.078488741890345, + "grad_norm": 3.0166192054748535, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9021034240722656, + "num_tokens": 634195908.0, + "step": 16339 + }, + { + "epoch": 2.0786159521689354, + "grad_norm": 2.8492228984832764, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9081727266311646, + "num_tokens": 634234281.0, + "step": 16340 + }, + { + "epoch": 2.078743162447526, + "grad_norm": 3.0696635246276855, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8978203535079956, + "num_tokens": 634267935.0, + "step": 16341 + }, + { + "epoch": 2.0788703727261164, + "grad_norm": 3.0228445529937744, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8922954797744751, + "num_tokens": 634306782.0, + "step": 16342 + }, + { + "epoch": 2.078997583004707, + "grad_norm": 3.118060827255249, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8924347758293152, + "num_tokens": 634344111.0, + "step": 16343 + }, + { + "epoch": 2.0791247932832975, + "grad_norm": 2.9276816844940186, + "learning_rate": 1e-06, + "loss": 0.2308, + "mean_token_accuracy": 0.9123544692993164, + "num_tokens": 634380767.0, + "step": 16344 + }, + { + "epoch": 2.079252003561888, + "grad_norm": 2.877612352371216, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8982998132705688, + "num_tokens": 634425044.0, + "step": 16345 + }, + { + "epoch": 2.0793792138404785, + "grad_norm": 3.029691457748413, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8959305286407471, + "num_tokens": 634464392.0, + "step": 16346 + }, + { + "epoch": 2.0795064241190686, + "grad_norm": 2.9998939037323, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8908453583717346, + "num_tokens": 634506153.0, + "step": 16347 + }, + { + "epoch": 2.079633634397659, + "grad_norm": 2.857663154602051, + "learning_rate": 1e-06, + "loss": 0.2311, + "mean_token_accuracy": 0.9126746654510498, + "num_tokens": 634540041.0, + "step": 16348 + }, + { + "epoch": 2.0797608446762497, + "grad_norm": 2.852510929107666, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9117774963378906, + "num_tokens": 634581491.0, + "step": 16349 + }, + { + "epoch": 2.07988805495484, + "grad_norm": 2.847978353500366, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9072275757789612, + "num_tokens": 634620586.0, + "step": 16350 + }, + { + "epoch": 2.0800152652334307, + "grad_norm": 2.7897567749023438, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.9096909165382385, + "num_tokens": 634660841.0, + "step": 16351 + }, + { + "epoch": 2.0801424755120212, + "grad_norm": 2.734677791595459, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9034112691879272, + "num_tokens": 634708006.0, + "step": 16352 + }, + { + "epoch": 2.0802696857906118, + "grad_norm": 3.0453808307647705, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9050177931785583, + "num_tokens": 634745024.0, + "step": 16353 + }, + { + "epoch": 2.0803968960692023, + "grad_norm": 3.2036478519439697, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8969042897224426, + "num_tokens": 634781603.0, + "step": 16354 + }, + { + "epoch": 2.080524106347793, + "grad_norm": 2.8877413272857666, + "learning_rate": 1e-06, + "loss": 0.2321, + "mean_token_accuracy": 0.913001537322998, + "num_tokens": 634819549.0, + "step": 16355 + }, + { + "epoch": 2.0806513166263834, + "grad_norm": 2.701052188873291, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9078887701034546, + "num_tokens": 634861024.0, + "step": 16356 + }, + { + "epoch": 2.080778526904974, + "grad_norm": 2.8361759185791016, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9036264419555664, + "num_tokens": 634902817.0, + "step": 16357 + }, + { + "epoch": 2.0809057371835644, + "grad_norm": 3.1837189197540283, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9104468822479248, + "num_tokens": 634936702.0, + "step": 16358 + }, + { + "epoch": 2.081032947462155, + "grad_norm": 2.9732320308685303, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9088258743286133, + "num_tokens": 634971942.0, + "step": 16359 + }, + { + "epoch": 2.0811601577407455, + "grad_norm": 2.8535170555114746, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9107722043991089, + "num_tokens": 635009394.0, + "step": 16360 + }, + { + "epoch": 2.081287368019336, + "grad_norm": 3.1084554195404053, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9022920727729797, + "num_tokens": 635042055.0, + "step": 16361 + }, + { + "epoch": 2.0814145782979265, + "grad_norm": 2.700113534927368, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9076768159866333, + "num_tokens": 635084172.0, + "step": 16362 + }, + { + "epoch": 2.081541788576517, + "grad_norm": 2.8405327796936035, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9049520492553711, + "num_tokens": 635122893.0, + "step": 16363 + }, + { + "epoch": 2.0816689988551076, + "grad_norm": 2.6366512775421143, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9064395427703857, + "num_tokens": 635162558.0, + "step": 16364 + }, + { + "epoch": 2.081796209133698, + "grad_norm": 3.107665538787842, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9038683176040649, + "num_tokens": 635197728.0, + "step": 16365 + }, + { + "epoch": 2.0819234194122886, + "grad_norm": 2.9827022552490234, + "learning_rate": 1e-06, + "loss": 0.2221, + "mean_token_accuracy": 0.9160631895065308, + "num_tokens": 635228735.0, + "step": 16366 + }, + { + "epoch": 2.082050629690879, + "grad_norm": 2.7605650424957275, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.90751051902771, + "num_tokens": 635269907.0, + "step": 16367 + }, + { + "epoch": 2.0821778399694697, + "grad_norm": 2.7277567386627197, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9098100662231445, + "num_tokens": 635308938.0, + "step": 16368 + }, + { + "epoch": 2.0823050502480602, + "grad_norm": 2.8594512939453125, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9100611209869385, + "num_tokens": 635351063.0, + "step": 16369 + }, + { + "epoch": 2.0824322605266508, + "grad_norm": 2.9833531379699707, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.905580997467041, + "num_tokens": 635388565.0, + "step": 16370 + }, + { + "epoch": 2.082559470805241, + "grad_norm": 2.895106554031372, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9057567119598389, + "num_tokens": 635425686.0, + "step": 16371 + }, + { + "epoch": 2.0826866810838314, + "grad_norm": 3.0470001697540283, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9087530970573425, + "num_tokens": 635459640.0, + "step": 16372 + }, + { + "epoch": 2.082813891362422, + "grad_norm": 2.697248697280884, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9053676128387451, + "num_tokens": 635504885.0, + "step": 16373 + }, + { + "epoch": 2.0829411016410124, + "grad_norm": 2.9460840225219727, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9022423028945923, + "num_tokens": 635541114.0, + "step": 16374 + }, + { + "epoch": 2.083068311919603, + "grad_norm": 3.183387279510498, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8959078192710876, + "num_tokens": 635577049.0, + "step": 16375 + }, + { + "epoch": 2.0831955221981935, + "grad_norm": 2.842863082885742, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9056963324546814, + "num_tokens": 635613814.0, + "step": 16376 + }, + { + "epoch": 2.083322732476784, + "grad_norm": 2.816648244857788, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9087746143341064, + "num_tokens": 635656303.0, + "step": 16377 + }, + { + "epoch": 2.0834499427553745, + "grad_norm": 2.8420567512512207, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9063729643821716, + "num_tokens": 635700121.0, + "step": 16378 + }, + { + "epoch": 2.083577153033965, + "grad_norm": 2.793031692504883, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9086710214614868, + "num_tokens": 635740385.0, + "step": 16379 + }, + { + "epoch": 2.0837043633125556, + "grad_norm": 3.001326084136963, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.9004161953926086, + "num_tokens": 635778317.0, + "step": 16380 + }, + { + "epoch": 2.083831573591146, + "grad_norm": 2.695880651473999, + "learning_rate": 1e-06, + "loss": 0.2195, + "mean_token_accuracy": 0.9186789989471436, + "num_tokens": 635816614.0, + "step": 16381 + }, + { + "epoch": 2.0839587838697367, + "grad_norm": 3.134761095046997, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8919037580490112, + "num_tokens": 635855703.0, + "step": 16382 + }, + { + "epoch": 2.084085994148327, + "grad_norm": 2.949148178100586, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8969957828521729, + "num_tokens": 635897419.0, + "step": 16383 + }, + { + "epoch": 2.0842132044269177, + "grad_norm": 2.938504219055176, + "learning_rate": 1e-06, + "loss": 0.2099, + "mean_token_accuracy": 0.9209754467010498, + "num_tokens": 635930833.0, + "step": 16384 + }, + { + "epoch": 2.0843404147055082, + "grad_norm": 2.7680516242980957, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9149128794670105, + "num_tokens": 635967509.0, + "step": 16385 + }, + { + "epoch": 2.0844676249840988, + "grad_norm": 2.8984689712524414, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9035016894340515, + "num_tokens": 636006040.0, + "step": 16386 + }, + { + "epoch": 2.0845948352626893, + "grad_norm": 2.933870553970337, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9137210249900818, + "num_tokens": 636041263.0, + "step": 16387 + }, + { + "epoch": 2.08472204554128, + "grad_norm": 3.3258566856384277, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.8999526500701904, + "num_tokens": 636073507.0, + "step": 16388 + }, + { + "epoch": 2.0848492558198704, + "grad_norm": 2.8445041179656982, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9093906879425049, + "num_tokens": 636112042.0, + "step": 16389 + }, + { + "epoch": 2.084976466098461, + "grad_norm": 3.1858482360839844, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9060274958610535, + "num_tokens": 636147157.0, + "step": 16390 + }, + { + "epoch": 2.0851036763770514, + "grad_norm": 2.847414493560791, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8962346911430359, + "num_tokens": 636188482.0, + "step": 16391 + }, + { + "epoch": 2.085230886655642, + "grad_norm": 2.7710251808166504, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9078960418701172, + "num_tokens": 636228724.0, + "step": 16392 + }, + { + "epoch": 2.0853580969342325, + "grad_norm": 2.7089903354644775, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9051221609115601, + "num_tokens": 636272087.0, + "step": 16393 + }, + { + "epoch": 2.085485307212823, + "grad_norm": 2.615745782852173, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9070878028869629, + "num_tokens": 636314659.0, + "step": 16394 + }, + { + "epoch": 2.0856125174914135, + "grad_norm": 2.8185949325561523, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8971885442733765, + "num_tokens": 636356019.0, + "step": 16395 + }, + { + "epoch": 2.0857397277700036, + "grad_norm": 2.7221546173095703, + "learning_rate": 1e-06, + "loss": 0.2243, + "mean_token_accuracy": 0.9156933426856995, + "num_tokens": 636393926.0, + "step": 16396 + }, + { + "epoch": 2.085866938048594, + "grad_norm": 3.2760605812072754, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.90342116355896, + "num_tokens": 636421539.0, + "step": 16397 + }, + { + "epoch": 2.0859941483271847, + "grad_norm": 2.789057493209839, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.8989080786705017, + "num_tokens": 636466685.0, + "step": 16398 + }, + { + "epoch": 2.086121358605775, + "grad_norm": 2.6582043170928955, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9095187187194824, + "num_tokens": 636509459.0, + "step": 16399 + }, + { + "epoch": 2.0862485688843657, + "grad_norm": 2.9841225147247314, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.909224271774292, + "num_tokens": 636546603.0, + "step": 16400 + }, + { + "epoch": 2.0863757791629562, + "grad_norm": 2.8123838901519775, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9046481847763062, + "num_tokens": 636585073.0, + "step": 16401 + }, + { + "epoch": 2.0865029894415468, + "grad_norm": 2.8734123706817627, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9021602869033813, + "num_tokens": 636624143.0, + "step": 16402 + }, + { + "epoch": 2.0866301997201373, + "grad_norm": 2.9700164794921875, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9009132385253906, + "num_tokens": 636660174.0, + "step": 16403 + }, + { + "epoch": 2.086757409998728, + "grad_norm": 3.1632978916168213, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9014564752578735, + "num_tokens": 636694725.0, + "step": 16404 + }, + { + "epoch": 2.0868846202773184, + "grad_norm": 3.010901689529419, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.8978900909423828, + "num_tokens": 636735027.0, + "step": 16405 + }, + { + "epoch": 2.087011830555909, + "grad_norm": 2.6789753437042236, + "learning_rate": 1e-06, + "loss": 0.2208, + "mean_token_accuracy": 0.9180722236633301, + "num_tokens": 636776791.0, + "step": 16406 + }, + { + "epoch": 2.0871390408344994, + "grad_norm": 3.115006446838379, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9046250581741333, + "num_tokens": 636811663.0, + "step": 16407 + }, + { + "epoch": 2.08726625111309, + "grad_norm": 2.8674020767211914, + "learning_rate": 1e-06, + "loss": 0.2274, + "mean_token_accuracy": 0.915804386138916, + "num_tokens": 636848575.0, + "step": 16408 + }, + { + "epoch": 2.0873934613916805, + "grad_norm": 2.593872547149658, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9066593647003174, + "num_tokens": 636890688.0, + "step": 16409 + }, + { + "epoch": 2.087520671670271, + "grad_norm": 2.950002908706665, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9039657711982727, + "num_tokens": 636928972.0, + "step": 16410 + }, + { + "epoch": 2.0876478819488615, + "grad_norm": 3.2313501834869385, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8944061994552612, + "num_tokens": 636967101.0, + "step": 16411 + }, + { + "epoch": 2.087775092227452, + "grad_norm": 2.8485021591186523, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9042847752571106, + "num_tokens": 637004988.0, + "step": 16412 + }, + { + "epoch": 2.0879023025060426, + "grad_norm": 2.628462791442871, + "learning_rate": 1e-06, + "loss": 0.2323, + "mean_token_accuracy": 0.91371750831604, + "num_tokens": 637046805.0, + "step": 16413 + }, + { + "epoch": 2.088029512784633, + "grad_norm": 2.7492072582244873, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9089993834495544, + "num_tokens": 637087248.0, + "step": 16414 + }, + { + "epoch": 2.0881567230632236, + "grad_norm": 2.7255074977874756, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9062238931655884, + "num_tokens": 637127779.0, + "step": 16415 + }, + { + "epoch": 2.088283933341814, + "grad_norm": 2.649247646331787, + "learning_rate": 1e-06, + "loss": 0.2217, + "mean_token_accuracy": 0.9139511585235596, + "num_tokens": 637166954.0, + "step": 16416 + }, + { + "epoch": 2.0884111436204047, + "grad_norm": 3.24996280670166, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.8990572094917297, + "num_tokens": 637199706.0, + "step": 16417 + }, + { + "epoch": 2.0885383538989952, + "grad_norm": 2.927488327026367, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9068934917449951, + "num_tokens": 637234928.0, + "step": 16418 + }, + { + "epoch": 2.0886655641775858, + "grad_norm": 2.807025671005249, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9121098518371582, + "num_tokens": 637272813.0, + "step": 16419 + }, + { + "epoch": 2.0887927744561763, + "grad_norm": 3.1470110416412354, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.8990321159362793, + "num_tokens": 637306526.0, + "step": 16420 + }, + { + "epoch": 2.0889199847347664, + "grad_norm": 2.747579336166382, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.906553328037262, + "num_tokens": 637346772.0, + "step": 16421 + }, + { + "epoch": 2.089047195013357, + "grad_norm": 2.975087881088257, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9091439247131348, + "num_tokens": 637382083.0, + "step": 16422 + }, + { + "epoch": 2.0891744052919474, + "grad_norm": 2.8015575408935547, + "learning_rate": 1e-06, + "loss": 0.2187, + "mean_token_accuracy": 0.9190540909767151, + "num_tokens": 637417068.0, + "step": 16423 + }, + { + "epoch": 2.089301615570538, + "grad_norm": 2.607044219970703, + "learning_rate": 1e-06, + "loss": 0.2186, + "mean_token_accuracy": 0.9184562563896179, + "num_tokens": 637458423.0, + "step": 16424 + }, + { + "epoch": 2.0894288258491285, + "grad_norm": 2.9508283138275146, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9013221263885498, + "num_tokens": 637495106.0, + "step": 16425 + }, + { + "epoch": 2.089556036127719, + "grad_norm": 2.5807580947875977, + "learning_rate": 1e-06, + "loss": 0.224, + "mean_token_accuracy": 0.9151610136032104, + "num_tokens": 637537999.0, + "step": 16426 + }, + { + "epoch": 2.0896832464063095, + "grad_norm": 2.781822919845581, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9082632064819336, + "num_tokens": 637579985.0, + "step": 16427 + }, + { + "epoch": 2.0898104566849, + "grad_norm": 2.5687971115112305, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.9105937480926514, + "num_tokens": 637622421.0, + "step": 16428 + }, + { + "epoch": 2.0899376669634906, + "grad_norm": 2.9200968742370605, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9094422459602356, + "num_tokens": 637659878.0, + "step": 16429 + }, + { + "epoch": 2.090064877242081, + "grad_norm": 2.9028100967407227, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9064846038818359, + "num_tokens": 637697290.0, + "step": 16430 + }, + { + "epoch": 2.0901920875206716, + "grad_norm": 2.848203659057617, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9011606574058533, + "num_tokens": 637735978.0, + "step": 16431 + }, + { + "epoch": 2.090319297799262, + "grad_norm": 2.766818046569824, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9046640396118164, + "num_tokens": 637777312.0, + "step": 16432 + }, + { + "epoch": 2.0904465080778527, + "grad_norm": 2.9129879474639893, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9049686193466187, + "num_tokens": 637816956.0, + "step": 16433 + }, + { + "epoch": 2.0905737183564432, + "grad_norm": 2.7778677940368652, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9117546677589417, + "num_tokens": 637855395.0, + "step": 16434 + }, + { + "epoch": 2.0907009286350338, + "grad_norm": 2.9206833839416504, + "learning_rate": 1e-06, + "loss": 0.2279, + "mean_token_accuracy": 0.9135804176330566, + "num_tokens": 637889977.0, + "step": 16435 + }, + { + "epoch": 2.0908281389136243, + "grad_norm": 2.9155049324035645, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9070600271224976, + "num_tokens": 637926432.0, + "step": 16436 + }, + { + "epoch": 2.090955349192215, + "grad_norm": 2.8672561645507812, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.90932297706604, + "num_tokens": 637964571.0, + "step": 16437 + }, + { + "epoch": 2.0910825594708053, + "grad_norm": 2.949382781982422, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9096146821975708, + "num_tokens": 638000452.0, + "step": 16438 + }, + { + "epoch": 2.091209769749396, + "grad_norm": 2.8411006927490234, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.9171296954154968, + "num_tokens": 638037465.0, + "step": 16439 + }, + { + "epoch": 2.0913369800279864, + "grad_norm": 2.9915988445281982, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9038565754890442, + "num_tokens": 638074344.0, + "step": 16440 + }, + { + "epoch": 2.091464190306577, + "grad_norm": 2.89402174949646, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9065564870834351, + "num_tokens": 638113194.0, + "step": 16441 + }, + { + "epoch": 2.0915914005851675, + "grad_norm": 2.7403759956359863, + "learning_rate": 1e-06, + "loss": 0.2262, + "mean_token_accuracy": 0.9161686301231384, + "num_tokens": 638150751.0, + "step": 16442 + }, + { + "epoch": 2.091718610863758, + "grad_norm": 2.8080832958221436, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9068642854690552, + "num_tokens": 638191912.0, + "step": 16443 + }, + { + "epoch": 2.0918458211423485, + "grad_norm": 2.9558873176574707, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9033458232879639, + "num_tokens": 638228887.0, + "step": 16444 + }, + { + "epoch": 2.0919730314209386, + "grad_norm": 2.674855947494507, + "learning_rate": 1e-06, + "loss": 0.2334, + "mean_token_accuracy": 0.9115798473358154, + "num_tokens": 638269664.0, + "step": 16445 + }, + { + "epoch": 2.092100241699529, + "grad_norm": 2.8154001235961914, + "learning_rate": 1e-06, + "loss": 0.2307, + "mean_token_accuracy": 0.9133825302124023, + "num_tokens": 638306333.0, + "step": 16446 + }, + { + "epoch": 2.0922274519781197, + "grad_norm": 3.094404697418213, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8941007852554321, + "num_tokens": 638343070.0, + "step": 16447 + }, + { + "epoch": 2.09235466225671, + "grad_norm": 3.2909598350524902, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9030863046646118, + "num_tokens": 638372012.0, + "step": 16448 + }, + { + "epoch": 2.0924818725353007, + "grad_norm": 2.7971160411834717, + "learning_rate": 1e-06, + "loss": 0.2299, + "mean_token_accuracy": 0.9154542684555054, + "num_tokens": 638412046.0, + "step": 16449 + }, + { + "epoch": 2.0926090828138912, + "grad_norm": 3.086052656173706, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9062540531158447, + "num_tokens": 638448062.0, + "step": 16450 + }, + { + "epoch": 2.0927362930924818, + "grad_norm": 3.4032416343688965, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9040688872337341, + "num_tokens": 638479738.0, + "step": 16451 + }, + { + "epoch": 2.0928635033710723, + "grad_norm": 2.817319393157959, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9055247902870178, + "num_tokens": 638519679.0, + "step": 16452 + }, + { + "epoch": 2.092990713649663, + "grad_norm": 2.7621235847473145, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9113391041755676, + "num_tokens": 638557630.0, + "step": 16453 + }, + { + "epoch": 2.0931179239282534, + "grad_norm": 3.03253173828125, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9006698727607727, + "num_tokens": 638597022.0, + "step": 16454 + }, + { + "epoch": 2.093245134206844, + "grad_norm": 2.755833148956299, + "learning_rate": 1e-06, + "loss": 0.22, + "mean_token_accuracy": 0.9165964126586914, + "num_tokens": 638632317.0, + "step": 16455 + }, + { + "epoch": 2.0933723444854344, + "grad_norm": 3.064103841781616, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9037739038467407, + "num_tokens": 638670038.0, + "step": 16456 + }, + { + "epoch": 2.093499554764025, + "grad_norm": 3.0865423679351807, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9040266275405884, + "num_tokens": 638704441.0, + "step": 16457 + }, + { + "epoch": 2.0936267650426155, + "grad_norm": 2.7809786796569824, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9092472195625305, + "num_tokens": 638745363.0, + "step": 16458 + }, + { + "epoch": 2.093753975321206, + "grad_norm": 3.1084060668945312, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.906161367893219, + "num_tokens": 638778098.0, + "step": 16459 + }, + { + "epoch": 2.0938811855997965, + "grad_norm": 2.7241876125335693, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.8992804288864136, + "num_tokens": 638820492.0, + "step": 16460 + }, + { + "epoch": 2.094008395878387, + "grad_norm": 3.2670929431915283, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8942540884017944, + "num_tokens": 638857078.0, + "step": 16461 + }, + { + "epoch": 2.0941356061569776, + "grad_norm": 2.926647901535034, + "learning_rate": 1e-06, + "loss": 0.2269, + "mean_token_accuracy": 0.9142113924026489, + "num_tokens": 638891617.0, + "step": 16462 + }, + { + "epoch": 2.094262816435568, + "grad_norm": 3.128810167312622, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9096879959106445, + "num_tokens": 638925561.0, + "step": 16463 + }, + { + "epoch": 2.0943900267141586, + "grad_norm": 2.902553081512451, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9058984518051147, + "num_tokens": 638963138.0, + "step": 16464 + }, + { + "epoch": 2.094517236992749, + "grad_norm": 3.0603179931640625, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8995494842529297, + "num_tokens": 639000648.0, + "step": 16465 + }, + { + "epoch": 2.0946444472713397, + "grad_norm": 2.834993362426758, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9087722301483154, + "num_tokens": 639038816.0, + "step": 16466 + }, + { + "epoch": 2.09477165754993, + "grad_norm": 3.0281758308410645, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9001332521438599, + "num_tokens": 639072888.0, + "step": 16467 + }, + { + "epoch": 2.0948988678285207, + "grad_norm": 2.723637342453003, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.903380274772644, + "num_tokens": 639115315.0, + "step": 16468 + }, + { + "epoch": 2.095026078107111, + "grad_norm": 2.8498289585113525, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9060211181640625, + "num_tokens": 639150004.0, + "step": 16469 + }, + { + "epoch": 2.0951532883857014, + "grad_norm": 2.6398842334747314, + "learning_rate": 1e-06, + "loss": 0.2212, + "mean_token_accuracy": 0.9198981523513794, + "num_tokens": 639191882.0, + "step": 16470 + }, + { + "epoch": 2.095280498664292, + "grad_norm": 2.5328502655029297, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9038501977920532, + "num_tokens": 639237326.0, + "step": 16471 + }, + { + "epoch": 2.0954077089428824, + "grad_norm": 2.675657272338867, + "learning_rate": 1e-06, + "loss": 0.2309, + "mean_token_accuracy": 0.9135941863059998, + "num_tokens": 639274350.0, + "step": 16472 + }, + { + "epoch": 2.095534919221473, + "grad_norm": 2.782188892364502, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9041719436645508, + "num_tokens": 639316554.0, + "step": 16473 + }, + { + "epoch": 2.0956621295000635, + "grad_norm": 2.7740211486816406, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.909183144569397, + "num_tokens": 639354585.0, + "step": 16474 + }, + { + "epoch": 2.095789339778654, + "grad_norm": 2.985903739929199, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.8991247415542603, + "num_tokens": 639393522.0, + "step": 16475 + }, + { + "epoch": 2.0959165500572445, + "grad_norm": 2.955280065536499, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.8994159698486328, + "num_tokens": 639432824.0, + "step": 16476 + }, + { + "epoch": 2.096043760335835, + "grad_norm": 3.0956318378448486, + "learning_rate": 1e-06, + "loss": 0.3207, + "mean_token_accuracy": 0.8848270177841187, + "num_tokens": 639470425.0, + "step": 16477 + }, + { + "epoch": 2.0961709706144256, + "grad_norm": 2.9087204933166504, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8981799483299255, + "num_tokens": 639511240.0, + "step": 16478 + }, + { + "epoch": 2.096298180893016, + "grad_norm": 3.234605073928833, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8917925357818604, + "num_tokens": 639552025.0, + "step": 16479 + }, + { + "epoch": 2.0964253911716066, + "grad_norm": 3.03041672706604, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.902030348777771, + "num_tokens": 639589175.0, + "step": 16480 + }, + { + "epoch": 2.096552601450197, + "grad_norm": 2.819671154022217, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9002875089645386, + "num_tokens": 639628332.0, + "step": 16481 + }, + { + "epoch": 2.0966798117287877, + "grad_norm": 2.775864601135254, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9129172563552856, + "num_tokens": 639669382.0, + "step": 16482 + }, + { + "epoch": 2.0968070220073782, + "grad_norm": 2.9064748287200928, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9082080125808716, + "num_tokens": 639704491.0, + "step": 16483 + }, + { + "epoch": 2.0969342322859688, + "grad_norm": 3.011073112487793, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9097291827201843, + "num_tokens": 639738382.0, + "step": 16484 + }, + { + "epoch": 2.0970614425645593, + "grad_norm": 2.4494693279266357, + "learning_rate": 1e-06, + "loss": 0.2218, + "mean_token_accuracy": 0.9186383485794067, + "num_tokens": 639784676.0, + "step": 16485 + }, + { + "epoch": 2.09718865284315, + "grad_norm": 2.8434348106384277, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9020416736602783, + "num_tokens": 639827795.0, + "step": 16486 + }, + { + "epoch": 2.0973158631217403, + "grad_norm": 2.9246506690979004, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9073781967163086, + "num_tokens": 639868599.0, + "step": 16487 + }, + { + "epoch": 2.097443073400331, + "grad_norm": 2.661848306655884, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9016345739364624, + "num_tokens": 639913126.0, + "step": 16488 + }, + { + "epoch": 2.0975702836789214, + "grad_norm": 2.725123405456543, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9113994836807251, + "num_tokens": 639952395.0, + "step": 16489 + }, + { + "epoch": 2.097697493957512, + "grad_norm": 2.656809091567993, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.910992443561554, + "num_tokens": 639990095.0, + "step": 16490 + }, + { + "epoch": 2.0978247042361025, + "grad_norm": 2.9214835166931152, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9025009274482727, + "num_tokens": 640029415.0, + "step": 16491 + }, + { + "epoch": 2.097951914514693, + "grad_norm": 2.986020565032959, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9042506814002991, + "num_tokens": 640067529.0, + "step": 16492 + }, + { + "epoch": 2.0980791247932835, + "grad_norm": 2.508648157119751, + "learning_rate": 1e-06, + "loss": 0.2317, + "mean_token_accuracy": 0.9126464128494263, + "num_tokens": 640113479.0, + "step": 16493 + }, + { + "epoch": 2.0982063350718736, + "grad_norm": 2.9820549488067627, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9061155319213867, + "num_tokens": 640150744.0, + "step": 16494 + }, + { + "epoch": 2.098333545350464, + "grad_norm": 2.899855136871338, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9003205299377441, + "num_tokens": 640192000.0, + "step": 16495 + }, + { + "epoch": 2.0984607556290547, + "grad_norm": 2.804471254348755, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9023064970970154, + "num_tokens": 640235347.0, + "step": 16496 + }, + { + "epoch": 2.098587965907645, + "grad_norm": 3.23038911819458, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.895854651927948, + "num_tokens": 640269636.0, + "step": 16497 + }, + { + "epoch": 2.0987151761862357, + "grad_norm": 2.8345448970794678, + "learning_rate": 1e-06, + "loss": 0.2215, + "mean_token_accuracy": 0.9168078899383545, + "num_tokens": 640305487.0, + "step": 16498 + }, + { + "epoch": 2.0988423864648262, + "grad_norm": 2.8145289421081543, + "learning_rate": 1e-06, + "loss": 0.2276, + "mean_token_accuracy": 0.9150816798210144, + "num_tokens": 640338140.0, + "step": 16499 + }, + { + "epoch": 2.0989695967434168, + "grad_norm": 3.7444350719451904, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9099744558334351, + "num_tokens": 640371879.0, + "step": 16500 + }, + { + "epoch": 2.0990968070220073, + "grad_norm": 2.9386818408966064, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9113881587982178, + "num_tokens": 640411571.0, + "step": 16501 + }, + { + "epoch": 2.099224017300598, + "grad_norm": 2.7893216609954834, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.90147864818573, + "num_tokens": 640454058.0, + "step": 16502 + }, + { + "epoch": 2.0993512275791884, + "grad_norm": 3.320845603942871, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8936324715614319, + "num_tokens": 640485790.0, + "step": 16503 + }, + { + "epoch": 2.099478437857779, + "grad_norm": 2.5813374519348145, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9126622080802917, + "num_tokens": 640532036.0, + "step": 16504 + }, + { + "epoch": 2.0996056481363694, + "grad_norm": 2.845796585083008, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9004571437835693, + "num_tokens": 640570859.0, + "step": 16505 + }, + { + "epoch": 2.09973285841496, + "grad_norm": 3.165132522583008, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9018267393112183, + "num_tokens": 640607786.0, + "step": 16506 + }, + { + "epoch": 2.0998600686935505, + "grad_norm": 3.0273144245147705, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8910927772521973, + "num_tokens": 640646990.0, + "step": 16507 + }, + { + "epoch": 2.099987278972141, + "grad_norm": 2.741680383682251, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9119581580162048, + "num_tokens": 640686493.0, + "step": 16508 + }, + { + "epoch": 2.1001144892507315, + "grad_norm": 2.601810932159424, + "learning_rate": 1e-06, + "loss": 0.2252, + "mean_token_accuracy": 0.914949357509613, + "num_tokens": 640726821.0, + "step": 16509 + }, + { + "epoch": 2.100241699529322, + "grad_norm": 3.2421298027038574, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9117635488510132, + "num_tokens": 640758649.0, + "step": 16510 + }, + { + "epoch": 2.1003689098079126, + "grad_norm": 2.895051956176758, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9108008742332458, + "num_tokens": 640794649.0, + "step": 16511 + }, + { + "epoch": 2.100496120086503, + "grad_norm": 3.062678098678589, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.895474374294281, + "num_tokens": 640830402.0, + "step": 16512 + }, + { + "epoch": 2.1006233303650936, + "grad_norm": 2.559025526046753, + "learning_rate": 1e-06, + "loss": 0.2145, + "mean_token_accuracy": 0.9173935055732727, + "num_tokens": 640869803.0, + "step": 16513 + }, + { + "epoch": 2.100750540643684, + "grad_norm": 2.7483577728271484, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9094551205635071, + "num_tokens": 640905268.0, + "step": 16514 + }, + { + "epoch": 2.1008777509222747, + "grad_norm": 2.745579957962036, + "learning_rate": 1e-06, + "loss": 0.2112, + "mean_token_accuracy": 0.9218534231185913, + "num_tokens": 640940478.0, + "step": 16515 + }, + { + "epoch": 2.101004961200865, + "grad_norm": 2.62397837638855, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.906230628490448, + "num_tokens": 640985891.0, + "step": 16516 + }, + { + "epoch": 2.1011321714794557, + "grad_norm": 2.8085999488830566, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9173886179924011, + "num_tokens": 641019400.0, + "step": 16517 + }, + { + "epoch": 2.1012593817580463, + "grad_norm": 2.5971426963806152, + "learning_rate": 1e-06, + "loss": 0.2294, + "mean_token_accuracy": 0.9160618782043457, + "num_tokens": 641063190.0, + "step": 16518 + }, + { + "epoch": 2.1013865920366364, + "grad_norm": 3.1683273315429688, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9018441438674927, + "num_tokens": 641099054.0, + "step": 16519 + }, + { + "epoch": 2.101513802315227, + "grad_norm": 2.783799648284912, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9093195796012878, + "num_tokens": 641141036.0, + "step": 16520 + }, + { + "epoch": 2.1016410125938174, + "grad_norm": 2.7453486919403076, + "learning_rate": 1e-06, + "loss": 0.2364, + "mean_token_accuracy": 0.911290168762207, + "num_tokens": 641180207.0, + "step": 16521 + }, + { + "epoch": 2.101768222872408, + "grad_norm": 2.9487099647521973, + "learning_rate": 1e-06, + "loss": 0.2317, + "mean_token_accuracy": 0.9141396284103394, + "num_tokens": 641213607.0, + "step": 16522 + }, + { + "epoch": 2.1018954331509985, + "grad_norm": 2.78913950920105, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8988551497459412, + "num_tokens": 641261813.0, + "step": 16523 + }, + { + "epoch": 2.102022643429589, + "grad_norm": 3.1282143592834473, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.9081931114196777, + "num_tokens": 641295865.0, + "step": 16524 + }, + { + "epoch": 2.1021498537081795, + "grad_norm": 2.5713696479797363, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.909465491771698, + "num_tokens": 641340323.0, + "step": 16525 + }, + { + "epoch": 2.10227706398677, + "grad_norm": 3.0246810913085938, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9029353857040405, + "num_tokens": 641378594.0, + "step": 16526 + }, + { + "epoch": 2.1024042742653606, + "grad_norm": 2.8490242958068848, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9019896984100342, + "num_tokens": 641417992.0, + "step": 16527 + }, + { + "epoch": 2.102531484543951, + "grad_norm": 3.16636061668396, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9051280617713928, + "num_tokens": 641450366.0, + "step": 16528 + }, + { + "epoch": 2.1026586948225416, + "grad_norm": 3.0199506282806396, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.904297947883606, + "num_tokens": 641488191.0, + "step": 16529 + }, + { + "epoch": 2.102785905101132, + "grad_norm": 2.819448709487915, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9036122560501099, + "num_tokens": 641526955.0, + "step": 16530 + }, + { + "epoch": 2.1029131153797227, + "grad_norm": 2.865565776824951, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9028002023696899, + "num_tokens": 641565143.0, + "step": 16531 + }, + { + "epoch": 2.1030403256583132, + "grad_norm": 2.6792969703674316, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9007777571678162, + "num_tokens": 641608928.0, + "step": 16532 + }, + { + "epoch": 2.1031675359369038, + "grad_norm": 2.8608827590942383, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9106134176254272, + "num_tokens": 641650782.0, + "step": 16533 + }, + { + "epoch": 2.1032947462154943, + "grad_norm": 2.907536506652832, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9082506895065308, + "num_tokens": 641685282.0, + "step": 16534 + }, + { + "epoch": 2.103421956494085, + "grad_norm": 2.762462615966797, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9070891737937927, + "num_tokens": 641727942.0, + "step": 16535 + }, + { + "epoch": 2.1035491667726753, + "grad_norm": 2.733194351196289, + "learning_rate": 1e-06, + "loss": 0.2208, + "mean_token_accuracy": 0.9176817536354065, + "num_tokens": 641766393.0, + "step": 16536 + }, + { + "epoch": 2.103676377051266, + "grad_norm": 2.5583336353302, + "learning_rate": 1e-06, + "loss": 0.2224, + "mean_token_accuracy": 0.9150112867355347, + "num_tokens": 641809879.0, + "step": 16537 + }, + { + "epoch": 2.1038035873298564, + "grad_norm": 2.632220983505249, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9044623374938965, + "num_tokens": 641852020.0, + "step": 16538 + }, + { + "epoch": 2.103930797608447, + "grad_norm": 2.93239426612854, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9022644758224487, + "num_tokens": 641890342.0, + "step": 16539 + }, + { + "epoch": 2.1040580078870375, + "grad_norm": 2.83901047706604, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9077658653259277, + "num_tokens": 641926646.0, + "step": 16540 + }, + { + "epoch": 2.104185218165628, + "grad_norm": 3.071378231048584, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9102232456207275, + "num_tokens": 641962987.0, + "step": 16541 + }, + { + "epoch": 2.1043124284442185, + "grad_norm": 2.581402063369751, + "learning_rate": 1e-06, + "loss": 0.2279, + "mean_token_accuracy": 0.9150393009185791, + "num_tokens": 642002042.0, + "step": 16542 + }, + { + "epoch": 2.1044396387228086, + "grad_norm": 2.6863222122192383, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9134006500244141, + "num_tokens": 642041294.0, + "step": 16543 + }, + { + "epoch": 2.104566849001399, + "grad_norm": 2.924710750579834, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.8995789289474487, + "num_tokens": 642083542.0, + "step": 16544 + }, + { + "epoch": 2.1046940592799896, + "grad_norm": 2.87389874458313, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9080846309661865, + "num_tokens": 642123622.0, + "step": 16545 + }, + { + "epoch": 2.10482126955858, + "grad_norm": 2.82973313331604, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9085260629653931, + "num_tokens": 642162150.0, + "step": 16546 + }, + { + "epoch": 2.1049484798371707, + "grad_norm": 2.657412528991699, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9097879528999329, + "num_tokens": 642203650.0, + "step": 16547 + }, + { + "epoch": 2.1050756901157612, + "grad_norm": 3.0794498920440674, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9051179885864258, + "num_tokens": 642236877.0, + "step": 16548 + }, + { + "epoch": 2.1052029003943518, + "grad_norm": 2.9800920486450195, + "learning_rate": 1e-06, + "loss": 0.2163, + "mean_token_accuracy": 0.9200721979141235, + "num_tokens": 642266822.0, + "step": 16549 + }, + { + "epoch": 2.1053301106729423, + "grad_norm": 2.8863284587860107, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9063246250152588, + "num_tokens": 642302130.0, + "step": 16550 + }, + { + "epoch": 2.105457320951533, + "grad_norm": 2.746669054031372, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9026041626930237, + "num_tokens": 642345506.0, + "step": 16551 + }, + { + "epoch": 2.1055845312301233, + "grad_norm": 2.8141045570373535, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.907239556312561, + "num_tokens": 642384478.0, + "step": 16552 + }, + { + "epoch": 2.105711741508714, + "grad_norm": 2.8626742362976074, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9024716019630432, + "num_tokens": 642422455.0, + "step": 16553 + }, + { + "epoch": 2.1058389517873044, + "grad_norm": 3.0818660259246826, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9061503410339355, + "num_tokens": 642455802.0, + "step": 16554 + }, + { + "epoch": 2.105966162065895, + "grad_norm": 2.944411039352417, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8947662711143494, + "num_tokens": 642497272.0, + "step": 16555 + }, + { + "epoch": 2.1060933723444855, + "grad_norm": 3.081663131713867, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8968408107757568, + "num_tokens": 642536988.0, + "step": 16556 + }, + { + "epoch": 2.106220582623076, + "grad_norm": 2.9283814430236816, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9101884961128235, + "num_tokens": 642570821.0, + "step": 16557 + }, + { + "epoch": 2.1063477929016665, + "grad_norm": 2.954050302505493, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9074804186820984, + "num_tokens": 642606921.0, + "step": 16558 + }, + { + "epoch": 2.106475003180257, + "grad_norm": 2.7052621841430664, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9037604331970215, + "num_tokens": 642650775.0, + "step": 16559 + }, + { + "epoch": 2.1066022134588476, + "grad_norm": 2.7506706714630127, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9079294204711914, + "num_tokens": 642688576.0, + "step": 16560 + }, + { + "epoch": 2.106729423737438, + "grad_norm": 2.721057415008545, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9162753820419312, + "num_tokens": 642725888.0, + "step": 16561 + }, + { + "epoch": 2.1068566340160286, + "grad_norm": 2.8023416996002197, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9075391292572021, + "num_tokens": 642764400.0, + "step": 16562 + }, + { + "epoch": 2.106983844294619, + "grad_norm": 2.9562907218933105, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9088696241378784, + "num_tokens": 642799969.0, + "step": 16563 + }, + { + "epoch": 2.1071110545732097, + "grad_norm": 2.8742494583129883, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9126495122909546, + "num_tokens": 642837712.0, + "step": 16564 + }, + { + "epoch": 2.1072382648518, + "grad_norm": 3.016313076019287, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9100795984268188, + "num_tokens": 642868552.0, + "step": 16565 + }, + { + "epoch": 2.1073654751303907, + "grad_norm": 2.8823838233947754, + "learning_rate": 1e-06, + "loss": 0.2422, + "mean_token_accuracy": 0.9077048301696777, + "num_tokens": 642906074.0, + "step": 16566 + }, + { + "epoch": 2.107492685408981, + "grad_norm": 3.0297181606292725, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9018510580062866, + "num_tokens": 642945084.0, + "step": 16567 + }, + { + "epoch": 2.1076198956875714, + "grad_norm": 2.747840404510498, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9082750678062439, + "num_tokens": 642984297.0, + "step": 16568 + }, + { + "epoch": 2.107747105966162, + "grad_norm": 2.8494210243225098, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9110362529754639, + "num_tokens": 643025865.0, + "step": 16569 + }, + { + "epoch": 2.1078743162447524, + "grad_norm": 2.8158912658691406, + "learning_rate": 1e-06, + "loss": 0.2342, + "mean_token_accuracy": 0.909185528755188, + "num_tokens": 643062893.0, + "step": 16570 + }, + { + "epoch": 2.108001526523343, + "grad_norm": 2.879913330078125, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9011150598526001, + "num_tokens": 643106985.0, + "step": 16571 + }, + { + "epoch": 2.1081287368019335, + "grad_norm": 3.0367438793182373, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9080495834350586, + "num_tokens": 643143961.0, + "step": 16572 + }, + { + "epoch": 2.108255947080524, + "grad_norm": 2.813133716583252, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9027575254440308, + "num_tokens": 643183652.0, + "step": 16573 + }, + { + "epoch": 2.1083831573591145, + "grad_norm": 2.9845495223999023, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9090048670768738, + "num_tokens": 643220019.0, + "step": 16574 + }, + { + "epoch": 2.108510367637705, + "grad_norm": 2.465113878250122, + "learning_rate": 1e-06, + "loss": 0.2283, + "mean_token_accuracy": 0.9146149754524231, + "num_tokens": 643266159.0, + "step": 16575 + }, + { + "epoch": 2.1086375779162956, + "grad_norm": 2.7753942012786865, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.903639018535614, + "num_tokens": 643306873.0, + "step": 16576 + }, + { + "epoch": 2.108764788194886, + "grad_norm": 2.862131118774414, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9082925319671631, + "num_tokens": 643346636.0, + "step": 16577 + }, + { + "epoch": 2.1088919984734766, + "grad_norm": 2.8191888332366943, + "learning_rate": 1e-06, + "loss": 0.234, + "mean_token_accuracy": 0.9152498245239258, + "num_tokens": 643382289.0, + "step": 16578 + }, + { + "epoch": 2.109019208752067, + "grad_norm": 2.820341110229492, + "learning_rate": 1e-06, + "loss": 0.2123, + "mean_token_accuracy": 0.9201035499572754, + "num_tokens": 643414714.0, + "step": 16579 + }, + { + "epoch": 2.1091464190306577, + "grad_norm": 2.756810188293457, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.909673273563385, + "num_tokens": 643454302.0, + "step": 16580 + }, + { + "epoch": 2.109273629309248, + "grad_norm": 2.6601548194885254, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9021787643432617, + "num_tokens": 643499453.0, + "step": 16581 + }, + { + "epoch": 2.1094008395878387, + "grad_norm": 2.9414596557617188, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9043228626251221, + "num_tokens": 643537681.0, + "step": 16582 + }, + { + "epoch": 2.1095280498664293, + "grad_norm": 2.7548346519470215, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9108651876449585, + "num_tokens": 643576141.0, + "step": 16583 + }, + { + "epoch": 2.10965526014502, + "grad_norm": 2.9044885635375977, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9026077389717102, + "num_tokens": 643613358.0, + "step": 16584 + }, + { + "epoch": 2.1097824704236103, + "grad_norm": 2.9802565574645996, + "learning_rate": 1e-06, + "loss": 0.2236, + "mean_token_accuracy": 0.9152597784996033, + "num_tokens": 643646613.0, + "step": 16585 + }, + { + "epoch": 2.109909680702201, + "grad_norm": 2.712493896484375, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9087015390396118, + "num_tokens": 643691647.0, + "step": 16586 + }, + { + "epoch": 2.1100368909807914, + "grad_norm": 3.1649301052093506, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9084025621414185, + "num_tokens": 643725385.0, + "step": 16587 + }, + { + "epoch": 2.110164101259382, + "grad_norm": 2.740968704223633, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9028234481811523, + "num_tokens": 643772716.0, + "step": 16588 + }, + { + "epoch": 2.1102913115379724, + "grad_norm": 2.975198745727539, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9035823345184326, + "num_tokens": 643807061.0, + "step": 16589 + }, + { + "epoch": 2.110418521816563, + "grad_norm": 2.863056182861328, + "learning_rate": 1e-06, + "loss": 0.2394, + "mean_token_accuracy": 0.910393476486206, + "num_tokens": 643843802.0, + "step": 16590 + }, + { + "epoch": 2.1105457320951535, + "grad_norm": 2.881991386413574, + "learning_rate": 1e-06, + "loss": 0.2264, + "mean_token_accuracy": 0.9124689102172852, + "num_tokens": 643876588.0, + "step": 16591 + }, + { + "epoch": 2.1106729423737436, + "grad_norm": 3.1803805828094482, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9040833115577698, + "num_tokens": 643910114.0, + "step": 16592 + }, + { + "epoch": 2.110800152652334, + "grad_norm": 2.7671830654144287, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9058071970939636, + "num_tokens": 643954182.0, + "step": 16593 + }, + { + "epoch": 2.1109273629309246, + "grad_norm": 2.7848877906799316, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8958408832550049, + "num_tokens": 643997624.0, + "step": 16594 + }, + { + "epoch": 2.111054573209515, + "grad_norm": 2.740222215652466, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.910551905632019, + "num_tokens": 644035802.0, + "step": 16595 + }, + { + "epoch": 2.1111817834881057, + "grad_norm": 2.7194676399230957, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9076566100120544, + "num_tokens": 644074600.0, + "step": 16596 + }, + { + "epoch": 2.1113089937666962, + "grad_norm": 3.2955167293548584, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9007575511932373, + "num_tokens": 644104549.0, + "step": 16597 + }, + { + "epoch": 2.1114362040452868, + "grad_norm": 2.7136542797088623, + "learning_rate": 1e-06, + "loss": 0.2242, + "mean_token_accuracy": 0.9144622087478638, + "num_tokens": 644142533.0, + "step": 16598 + }, + { + "epoch": 2.1115634143238773, + "grad_norm": 2.63372802734375, + "learning_rate": 1e-06, + "loss": 0.2159, + "mean_token_accuracy": 0.9191699624061584, + "num_tokens": 644185850.0, + "step": 16599 + }, + { + "epoch": 2.111690624602468, + "grad_norm": 2.708660125732422, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9025788307189941, + "num_tokens": 644231916.0, + "step": 16600 + }, + { + "epoch": 2.1118178348810583, + "grad_norm": 2.919548749923706, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9076440930366516, + "num_tokens": 644268722.0, + "step": 16601 + }, + { + "epoch": 2.111945045159649, + "grad_norm": 2.694235324859619, + "learning_rate": 1e-06, + "loss": 0.2243, + "mean_token_accuracy": 0.9148392081260681, + "num_tokens": 644308328.0, + "step": 16602 + }, + { + "epoch": 2.1120722554382394, + "grad_norm": 2.762752056121826, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9001861810684204, + "num_tokens": 644351573.0, + "step": 16603 + }, + { + "epoch": 2.11219946571683, + "grad_norm": 3.255506753921509, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9022380113601685, + "num_tokens": 644390490.0, + "step": 16604 + }, + { + "epoch": 2.1123266759954205, + "grad_norm": 2.787144660949707, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9067149758338928, + "num_tokens": 644430227.0, + "step": 16605 + }, + { + "epoch": 2.112453886274011, + "grad_norm": 2.8257620334625244, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9056572914123535, + "num_tokens": 644470852.0, + "step": 16606 + }, + { + "epoch": 2.1125810965526015, + "grad_norm": 2.7964394092559814, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9088796377182007, + "num_tokens": 644510889.0, + "step": 16607 + }, + { + "epoch": 2.112708306831192, + "grad_norm": 2.7218575477600098, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9038762450218201, + "num_tokens": 644556041.0, + "step": 16608 + }, + { + "epoch": 2.1128355171097826, + "grad_norm": 2.577064275741577, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9070947170257568, + "num_tokens": 644606337.0, + "step": 16609 + }, + { + "epoch": 2.112962727388373, + "grad_norm": 2.9252634048461914, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.904147744178772, + "num_tokens": 644646048.0, + "step": 16610 + }, + { + "epoch": 2.1130899376669636, + "grad_norm": 3.250579833984375, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8959972262382507, + "num_tokens": 644682306.0, + "step": 16611 + }, + { + "epoch": 2.113217147945554, + "grad_norm": 3.051496744155884, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8972245454788208, + "num_tokens": 644720338.0, + "step": 16612 + }, + { + "epoch": 2.1133443582241447, + "grad_norm": 2.667454481124878, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9017626047134399, + "num_tokens": 644766212.0, + "step": 16613 + }, + { + "epoch": 2.113471568502735, + "grad_norm": 2.873338222503662, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8988049626350403, + "num_tokens": 644806926.0, + "step": 16614 + }, + { + "epoch": 2.1135987787813257, + "grad_norm": 2.962722063064575, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9023001790046692, + "num_tokens": 644845597.0, + "step": 16615 + }, + { + "epoch": 2.1137259890599163, + "grad_norm": 3.1976373195648193, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9077008366584778, + "num_tokens": 644877064.0, + "step": 16616 + }, + { + "epoch": 2.1138531993385064, + "grad_norm": 3.0294976234436035, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9077113270759583, + "num_tokens": 644912215.0, + "step": 16617 + }, + { + "epoch": 2.113980409617097, + "grad_norm": 3.1431736946105957, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9032731652259827, + "num_tokens": 644948335.0, + "step": 16618 + }, + { + "epoch": 2.1141076198956874, + "grad_norm": 2.9205174446105957, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9035325050354004, + "num_tokens": 644987584.0, + "step": 16619 + }, + { + "epoch": 2.114234830174278, + "grad_norm": 3.1306257247924805, + "learning_rate": 1e-06, + "loss": 0.2416, + "mean_token_accuracy": 0.9096137285232544, + "num_tokens": 645017207.0, + "step": 16620 + }, + { + "epoch": 2.1143620404528685, + "grad_norm": 2.829695701599121, + "learning_rate": 1e-06, + "loss": 0.2272, + "mean_token_accuracy": 0.9129660129547119, + "num_tokens": 645051960.0, + "step": 16621 + }, + { + "epoch": 2.114489250731459, + "grad_norm": 2.804147720336914, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.898754358291626, + "num_tokens": 645093412.0, + "step": 16622 + }, + { + "epoch": 2.1146164610100495, + "grad_norm": 3.0925354957580566, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8958355784416199, + "num_tokens": 645133933.0, + "step": 16623 + }, + { + "epoch": 2.11474367128864, + "grad_norm": 2.992784023284912, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8950591683387756, + "num_tokens": 645173376.0, + "step": 16624 + }, + { + "epoch": 2.1148708815672306, + "grad_norm": 3.0105693340301514, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.902906060218811, + "num_tokens": 645207648.0, + "step": 16625 + }, + { + "epoch": 2.114998091845821, + "grad_norm": 2.895054817199707, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9034781455993652, + "num_tokens": 645250130.0, + "step": 16626 + }, + { + "epoch": 2.1151253021244116, + "grad_norm": 2.9932022094726562, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8991901874542236, + "num_tokens": 645289079.0, + "step": 16627 + }, + { + "epoch": 2.115252512403002, + "grad_norm": 3.0162720680236816, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9108557105064392, + "num_tokens": 645324522.0, + "step": 16628 + }, + { + "epoch": 2.1153797226815927, + "grad_norm": 2.9066033363342285, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9133416414260864, + "num_tokens": 645359775.0, + "step": 16629 + }, + { + "epoch": 2.115506932960183, + "grad_norm": 3.105107545852661, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9038596153259277, + "num_tokens": 645393098.0, + "step": 16630 + }, + { + "epoch": 2.1156341432387737, + "grad_norm": 2.9291465282440186, + "learning_rate": 1e-06, + "loss": 0.2334, + "mean_token_accuracy": 0.9131478071212769, + "num_tokens": 645428363.0, + "step": 16631 + }, + { + "epoch": 2.1157613535173643, + "grad_norm": 3.0424578189849854, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9073017239570618, + "num_tokens": 645463003.0, + "step": 16632 + }, + { + "epoch": 2.115888563795955, + "grad_norm": 3.110316276550293, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8943777680397034, + "num_tokens": 645499521.0, + "step": 16633 + }, + { + "epoch": 2.1160157740745453, + "grad_norm": 2.9595563411712646, + "learning_rate": 1e-06, + "loss": 0.2272, + "mean_token_accuracy": 0.9146116971969604, + "num_tokens": 645532763.0, + "step": 16634 + }, + { + "epoch": 2.116142984353136, + "grad_norm": 2.979609251022339, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8946972489356995, + "num_tokens": 645574617.0, + "step": 16635 + }, + { + "epoch": 2.1162701946317264, + "grad_norm": 2.8021695613861084, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9026241898536682, + "num_tokens": 645616684.0, + "step": 16636 + }, + { + "epoch": 2.116397404910317, + "grad_norm": 2.934999704360962, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9147181510925293, + "num_tokens": 645654149.0, + "step": 16637 + }, + { + "epoch": 2.1165246151889074, + "grad_norm": 2.72102689743042, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9087381958961487, + "num_tokens": 645695409.0, + "step": 16638 + }, + { + "epoch": 2.116651825467498, + "grad_norm": 2.7665021419525146, + "learning_rate": 1e-06, + "loss": 0.2362, + "mean_token_accuracy": 0.9116290807723999, + "num_tokens": 645736803.0, + "step": 16639 + }, + { + "epoch": 2.116779035746088, + "grad_norm": 3.2326645851135254, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9036780595779419, + "num_tokens": 645768411.0, + "step": 16640 + }, + { + "epoch": 2.1169062460246786, + "grad_norm": 3.0468592643737793, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.895165205001831, + "num_tokens": 645805996.0, + "step": 16641 + }, + { + "epoch": 2.117033456303269, + "grad_norm": 3.0450074672698975, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9049118757247925, + "num_tokens": 645844854.0, + "step": 16642 + }, + { + "epoch": 2.1171606665818596, + "grad_norm": 2.7648634910583496, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.895865797996521, + "num_tokens": 645889657.0, + "step": 16643 + }, + { + "epoch": 2.11728787686045, + "grad_norm": 2.660224199295044, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9066241979598999, + "num_tokens": 645930652.0, + "step": 16644 + }, + { + "epoch": 2.1174150871390407, + "grad_norm": 3.531615734100342, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8969660401344299, + "num_tokens": 645969888.0, + "step": 16645 + }, + { + "epoch": 2.1175422974176312, + "grad_norm": 2.786168336868286, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9052817225456238, + "num_tokens": 646010484.0, + "step": 16646 + }, + { + "epoch": 2.1176695076962218, + "grad_norm": 2.6187198162078857, + "learning_rate": 1e-06, + "loss": 0.2296, + "mean_token_accuracy": 0.9152461290359497, + "num_tokens": 646052182.0, + "step": 16647 + }, + { + "epoch": 2.1177967179748123, + "grad_norm": 2.726691961288452, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9026967883110046, + "num_tokens": 646096064.0, + "step": 16648 + }, + { + "epoch": 2.117923928253403, + "grad_norm": 3.0437893867492676, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9083764553070068, + "num_tokens": 646131204.0, + "step": 16649 + }, + { + "epoch": 2.1180511385319933, + "grad_norm": 2.8019773960113525, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9006912112236023, + "num_tokens": 646170456.0, + "step": 16650 + }, + { + "epoch": 2.118178348810584, + "grad_norm": 2.515984535217285, + "learning_rate": 1e-06, + "loss": 0.2247, + "mean_token_accuracy": 0.9160680770874023, + "num_tokens": 646217220.0, + "step": 16651 + }, + { + "epoch": 2.1183055590891744, + "grad_norm": 2.8370261192321777, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8937093019485474, + "num_tokens": 646258981.0, + "step": 16652 + }, + { + "epoch": 2.118432769367765, + "grad_norm": 2.5801594257354736, + "learning_rate": 1e-06, + "loss": 0.2269, + "mean_token_accuracy": 0.9171258211135864, + "num_tokens": 646302117.0, + "step": 16653 + }, + { + "epoch": 2.1185599796463555, + "grad_norm": 2.9798688888549805, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9031370878219604, + "num_tokens": 646339185.0, + "step": 16654 + }, + { + "epoch": 2.118687189924946, + "grad_norm": 2.7534122467041016, + "learning_rate": 1e-06, + "loss": 0.2317, + "mean_token_accuracy": 0.913489818572998, + "num_tokens": 646376196.0, + "step": 16655 + }, + { + "epoch": 2.1188144002035365, + "grad_norm": 2.597738027572632, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9046643972396851, + "num_tokens": 646425118.0, + "step": 16656 + }, + { + "epoch": 2.118941610482127, + "grad_norm": 3.0399434566497803, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8978190422058105, + "num_tokens": 646462979.0, + "step": 16657 + }, + { + "epoch": 2.1190688207607176, + "grad_norm": 2.707052707672119, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9016927480697632, + "num_tokens": 646506878.0, + "step": 16658 + }, + { + "epoch": 2.119196031039308, + "grad_norm": 2.8851735591888428, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8934304714202881, + "num_tokens": 646550593.0, + "step": 16659 + }, + { + "epoch": 2.1193232413178986, + "grad_norm": 3.0082266330718994, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9135614633560181, + "num_tokens": 646584318.0, + "step": 16660 + }, + { + "epoch": 2.119450451596489, + "grad_norm": 2.7576956748962402, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.905542254447937, + "num_tokens": 646627926.0, + "step": 16661 + }, + { + "epoch": 2.1195776618750797, + "grad_norm": 3.159874677658081, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9148379564285278, + "num_tokens": 646660937.0, + "step": 16662 + }, + { + "epoch": 2.11970487215367, + "grad_norm": 3.0490009784698486, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8989471793174744, + "num_tokens": 646701093.0, + "step": 16663 + }, + { + "epoch": 2.1198320824322607, + "grad_norm": 2.7341341972351074, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.910901665687561, + "num_tokens": 646739887.0, + "step": 16664 + }, + { + "epoch": 2.119959292710851, + "grad_norm": 2.7030487060546875, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9092758893966675, + "num_tokens": 646780399.0, + "step": 16665 + }, + { + "epoch": 2.1200865029894413, + "grad_norm": 2.8660218715667725, + "learning_rate": 1e-06, + "loss": 0.2445, + "mean_token_accuracy": 0.9081334471702576, + "num_tokens": 646819126.0, + "step": 16666 + }, + { + "epoch": 2.120213713268032, + "grad_norm": 2.795107126235962, + "learning_rate": 1e-06, + "loss": 0.2442, + "mean_token_accuracy": 0.9100097417831421, + "num_tokens": 646859791.0, + "step": 16667 + }, + { + "epoch": 2.1203409235466224, + "grad_norm": 3.5225374698638916, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8891605138778687, + "num_tokens": 646891642.0, + "step": 16668 + }, + { + "epoch": 2.120468133825213, + "grad_norm": 2.9428839683532715, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.8978477716445923, + "num_tokens": 646931521.0, + "step": 16669 + }, + { + "epoch": 2.1205953441038035, + "grad_norm": 2.66044545173645, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9077988266944885, + "num_tokens": 646974619.0, + "step": 16670 + }, + { + "epoch": 2.120722554382394, + "grad_norm": 2.7072269916534424, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9089412689208984, + "num_tokens": 647018162.0, + "step": 16671 + }, + { + "epoch": 2.1208497646609845, + "grad_norm": 2.7769124507904053, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9059147834777832, + "num_tokens": 647058268.0, + "step": 16672 + }, + { + "epoch": 2.120976974939575, + "grad_norm": 2.697983741760254, + "learning_rate": 1e-06, + "loss": 0.2337, + "mean_token_accuracy": 0.9120714664459229, + "num_tokens": 647096432.0, + "step": 16673 + }, + { + "epoch": 2.1211041852181656, + "grad_norm": 2.6951799392700195, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9083085060119629, + "num_tokens": 647136296.0, + "step": 16674 + }, + { + "epoch": 2.121231395496756, + "grad_norm": 2.57090163230896, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9056928157806396, + "num_tokens": 647181407.0, + "step": 16675 + }, + { + "epoch": 2.1213586057753466, + "grad_norm": 2.892341136932373, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.8999946713447571, + "num_tokens": 647219268.0, + "step": 16676 + }, + { + "epoch": 2.121485816053937, + "grad_norm": 2.577207326889038, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9119378924369812, + "num_tokens": 647259286.0, + "step": 16677 + }, + { + "epoch": 2.1216130263325277, + "grad_norm": 2.852318048477173, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9062479734420776, + "num_tokens": 647297351.0, + "step": 16678 + }, + { + "epoch": 2.121740236611118, + "grad_norm": 2.7908360958099365, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9051758050918579, + "num_tokens": 647340252.0, + "step": 16679 + }, + { + "epoch": 2.1218674468897087, + "grad_norm": 2.6486499309539795, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9048316478729248, + "num_tokens": 647381035.0, + "step": 16680 + }, + { + "epoch": 2.1219946571682993, + "grad_norm": 2.7642436027526855, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9052106142044067, + "num_tokens": 647420261.0, + "step": 16681 + }, + { + "epoch": 2.12212186744689, + "grad_norm": 2.587559938430786, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9044549465179443, + "num_tokens": 647463248.0, + "step": 16682 + }, + { + "epoch": 2.1222490777254803, + "grad_norm": 2.931654214859009, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8950926065444946, + "num_tokens": 647505296.0, + "step": 16683 + }, + { + "epoch": 2.122376288004071, + "grad_norm": 3.0104942321777344, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8936432600021362, + "num_tokens": 647547827.0, + "step": 16684 + }, + { + "epoch": 2.1225034982826614, + "grad_norm": 2.8090977668762207, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9113267660140991, + "num_tokens": 647588596.0, + "step": 16685 + }, + { + "epoch": 2.122630708561252, + "grad_norm": 2.738632917404175, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9005739688873291, + "num_tokens": 647628408.0, + "step": 16686 + }, + { + "epoch": 2.1227579188398424, + "grad_norm": 3.148109197616577, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8889843225479126, + "num_tokens": 647668646.0, + "step": 16687 + }, + { + "epoch": 2.122885129118433, + "grad_norm": 2.6409108638763428, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.907107412815094, + "num_tokens": 647712412.0, + "step": 16688 + }, + { + "epoch": 2.1230123393970235, + "grad_norm": 2.944662094116211, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8874064683914185, + "num_tokens": 647757704.0, + "step": 16689 + }, + { + "epoch": 2.1231395496756136, + "grad_norm": 2.8554818630218506, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8969367742538452, + "num_tokens": 647798241.0, + "step": 16690 + }, + { + "epoch": 2.123266759954204, + "grad_norm": 3.0667543411254883, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9071281552314758, + "num_tokens": 647832111.0, + "step": 16691 + }, + { + "epoch": 2.1233939702327946, + "grad_norm": 2.9206764698028564, + "learning_rate": 1e-06, + "loss": 0.2283, + "mean_token_accuracy": 0.914689302444458, + "num_tokens": 647866740.0, + "step": 16692 + }, + { + "epoch": 2.123521180511385, + "grad_norm": 2.533154249191284, + "learning_rate": 1e-06, + "loss": 0.2255, + "mean_token_accuracy": 0.9149978160858154, + "num_tokens": 647910515.0, + "step": 16693 + }, + { + "epoch": 2.1236483907899757, + "grad_norm": 2.866966724395752, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9074758291244507, + "num_tokens": 647950460.0, + "step": 16694 + }, + { + "epoch": 2.123775601068566, + "grad_norm": 2.9189019203186035, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9058036804199219, + "num_tokens": 647986789.0, + "step": 16695 + }, + { + "epoch": 2.1239028113471567, + "grad_norm": 2.9163765907287598, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9012482166290283, + "num_tokens": 648024620.0, + "step": 16696 + }, + { + "epoch": 2.1240300216257473, + "grad_norm": 2.7665772438049316, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.909336507320404, + "num_tokens": 648061866.0, + "step": 16697 + }, + { + "epoch": 2.124157231904338, + "grad_norm": 2.8968923091888428, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9028391242027283, + "num_tokens": 648099408.0, + "step": 16698 + }, + { + "epoch": 2.1242844421829283, + "grad_norm": 2.8725478649139404, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9121771454811096, + "num_tokens": 648133807.0, + "step": 16699 + }, + { + "epoch": 2.124411652461519, + "grad_norm": 2.4865877628326416, + "learning_rate": 1e-06, + "loss": 0.219, + "mean_token_accuracy": 0.9200268983840942, + "num_tokens": 648179993.0, + "step": 16700 + }, + { + "epoch": 2.1245388627401094, + "grad_norm": 2.761129379272461, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.914691686630249, + "num_tokens": 648216896.0, + "step": 16701 + }, + { + "epoch": 2.1246660730187, + "grad_norm": 2.766331672668457, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9063664078712463, + "num_tokens": 648255081.0, + "step": 16702 + }, + { + "epoch": 2.1247932832972904, + "grad_norm": 2.7287683486938477, + "learning_rate": 1e-06, + "loss": 0.2289, + "mean_token_accuracy": 0.9155663251876831, + "num_tokens": 648293441.0, + "step": 16703 + }, + { + "epoch": 2.124920493575881, + "grad_norm": 3.1869704723358154, + "learning_rate": 1e-06, + "loss": 0.3098, + "mean_token_accuracy": 0.8867920637130737, + "num_tokens": 648335298.0, + "step": 16704 + }, + { + "epoch": 2.1250477038544715, + "grad_norm": 3.125120162963867, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9009884595870972, + "num_tokens": 648371832.0, + "step": 16705 + }, + { + "epoch": 2.125174914133062, + "grad_norm": 3.121819019317627, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8976346254348755, + "num_tokens": 648409729.0, + "step": 16706 + }, + { + "epoch": 2.1253021244116526, + "grad_norm": 2.675765037536621, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9013766646385193, + "num_tokens": 648451970.0, + "step": 16707 + }, + { + "epoch": 2.125429334690243, + "grad_norm": 2.7027781009674072, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9033688306808472, + "num_tokens": 648494652.0, + "step": 16708 + }, + { + "epoch": 2.1255565449688336, + "grad_norm": 2.838129997253418, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9063888192176819, + "num_tokens": 648533865.0, + "step": 16709 + }, + { + "epoch": 2.125683755247424, + "grad_norm": 3.0565309524536133, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.90513014793396, + "num_tokens": 648569046.0, + "step": 16710 + }, + { + "epoch": 2.1258109655260147, + "grad_norm": 2.805847406387329, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9074629545211792, + "num_tokens": 648605458.0, + "step": 16711 + }, + { + "epoch": 2.125938175804605, + "grad_norm": 2.9680888652801514, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8956447839736938, + "num_tokens": 648644105.0, + "step": 16712 + }, + { + "epoch": 2.1260653860831957, + "grad_norm": 2.8485283851623535, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9046090841293335, + "num_tokens": 648685355.0, + "step": 16713 + }, + { + "epoch": 2.1261925963617863, + "grad_norm": 2.9239516258239746, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9065086841583252, + "num_tokens": 648719766.0, + "step": 16714 + }, + { + "epoch": 2.1263198066403763, + "grad_norm": 2.835725784301758, + "learning_rate": 1e-06, + "loss": 0.242, + "mean_token_accuracy": 0.9099348783493042, + "num_tokens": 648756891.0, + "step": 16715 + }, + { + "epoch": 2.126447016918967, + "grad_norm": 3.0026957988739014, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9046626091003418, + "num_tokens": 648794634.0, + "step": 16716 + }, + { + "epoch": 2.1265742271975574, + "grad_norm": 2.808730363845825, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9053874015808105, + "num_tokens": 648836145.0, + "step": 16717 + }, + { + "epoch": 2.126701437476148, + "grad_norm": 2.8431739807128906, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9101504683494568, + "num_tokens": 648871380.0, + "step": 16718 + }, + { + "epoch": 2.1268286477547385, + "grad_norm": 2.8649954795837402, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9024738073348999, + "num_tokens": 648912482.0, + "step": 16719 + }, + { + "epoch": 2.126955858033329, + "grad_norm": 2.7210421562194824, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9129515886306763, + "num_tokens": 648952247.0, + "step": 16720 + }, + { + "epoch": 2.1270830683119195, + "grad_norm": 2.9176909923553467, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8975436091423035, + "num_tokens": 648991162.0, + "step": 16721 + }, + { + "epoch": 2.12721027859051, + "grad_norm": 2.898893117904663, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8948030471801758, + "num_tokens": 649033165.0, + "step": 16722 + }, + { + "epoch": 2.1273374888691006, + "grad_norm": 2.874690532684326, + "learning_rate": 1e-06, + "loss": 0.225, + "mean_token_accuracy": 0.9141930937767029, + "num_tokens": 649066581.0, + "step": 16723 + }, + { + "epoch": 2.127464699147691, + "grad_norm": 2.7853875160217285, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9081919193267822, + "num_tokens": 649106425.0, + "step": 16724 + }, + { + "epoch": 2.1275919094262816, + "grad_norm": 2.8007571697235107, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9108403921127319, + "num_tokens": 649149421.0, + "step": 16725 + }, + { + "epoch": 2.127719119704872, + "grad_norm": 2.7557578086853027, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9055463075637817, + "num_tokens": 649189158.0, + "step": 16726 + }, + { + "epoch": 2.1278463299834627, + "grad_norm": 2.6862025260925293, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9140613079071045, + "num_tokens": 649230010.0, + "step": 16727 + }, + { + "epoch": 2.127973540262053, + "grad_norm": 2.6765146255493164, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9048923254013062, + "num_tokens": 649268393.0, + "step": 16728 + }, + { + "epoch": 2.1281007505406437, + "grad_norm": 2.9332685470581055, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.8977432250976562, + "num_tokens": 649306194.0, + "step": 16729 + }, + { + "epoch": 2.1282279608192343, + "grad_norm": 2.768242597579956, + "learning_rate": 1e-06, + "loss": 0.2347, + "mean_token_accuracy": 0.9147651195526123, + "num_tokens": 649343451.0, + "step": 16730 + }, + { + "epoch": 2.128355171097825, + "grad_norm": 2.899454116821289, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9095297455787659, + "num_tokens": 649379189.0, + "step": 16731 + }, + { + "epoch": 2.1284823813764153, + "grad_norm": 2.8355236053466797, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.909376323223114, + "num_tokens": 649420235.0, + "step": 16732 + }, + { + "epoch": 2.128609591655006, + "grad_norm": 3.175487518310547, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9016159176826477, + "num_tokens": 649457890.0, + "step": 16733 + }, + { + "epoch": 2.1287368019335964, + "grad_norm": 2.576883554458618, + "learning_rate": 1e-06, + "loss": 0.2134, + "mean_token_accuracy": 0.920871376991272, + "num_tokens": 649497989.0, + "step": 16734 + }, + { + "epoch": 2.128864012212187, + "grad_norm": 2.9441001415252686, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8993808627128601, + "num_tokens": 649539252.0, + "step": 16735 + }, + { + "epoch": 2.1289912224907774, + "grad_norm": 2.917224884033203, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8963029384613037, + "num_tokens": 649580024.0, + "step": 16736 + }, + { + "epoch": 2.129118432769368, + "grad_norm": 2.878206729888916, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8971410989761353, + "num_tokens": 649623782.0, + "step": 16737 + }, + { + "epoch": 2.129245643047958, + "grad_norm": 3.2615549564361572, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9010435342788696, + "num_tokens": 649657192.0, + "step": 16738 + }, + { + "epoch": 2.129372853326549, + "grad_norm": 3.2128419876098633, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9036893248558044, + "num_tokens": 649692951.0, + "step": 16739 + }, + { + "epoch": 2.129500063605139, + "grad_norm": 2.8543851375579834, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9071462154388428, + "num_tokens": 649732078.0, + "step": 16740 + }, + { + "epoch": 2.1296272738837296, + "grad_norm": 2.7387988567352295, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9120202660560608, + "num_tokens": 649771935.0, + "step": 16741 + }, + { + "epoch": 2.12975448416232, + "grad_norm": 2.815305471420288, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9042969942092896, + "num_tokens": 649814333.0, + "step": 16742 + }, + { + "epoch": 2.1298816944409107, + "grad_norm": 2.77937650680542, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9075125455856323, + "num_tokens": 649856150.0, + "step": 16743 + }, + { + "epoch": 2.130008904719501, + "grad_norm": 2.8662123680114746, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9039799571037292, + "num_tokens": 649897011.0, + "step": 16744 + }, + { + "epoch": 2.1301361149980917, + "grad_norm": 2.699437379837036, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9066801071166992, + "num_tokens": 649943612.0, + "step": 16745 + }, + { + "epoch": 2.1302633252766823, + "grad_norm": 2.8010265827178955, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9150755405426025, + "num_tokens": 649985131.0, + "step": 16746 + }, + { + "epoch": 2.130390535555273, + "grad_norm": 3.202382802963257, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8937575817108154, + "num_tokens": 650021991.0, + "step": 16747 + }, + { + "epoch": 2.1305177458338633, + "grad_norm": 3.1832430362701416, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9086834192276001, + "num_tokens": 650055083.0, + "step": 16748 + }, + { + "epoch": 2.130644956112454, + "grad_norm": 2.896524429321289, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9115602970123291, + "num_tokens": 650092299.0, + "step": 16749 + }, + { + "epoch": 2.1307721663910444, + "grad_norm": 2.918240547180176, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.901557445526123, + "num_tokens": 650132367.0, + "step": 16750 + }, + { + "epoch": 2.130899376669635, + "grad_norm": 2.864438533782959, + "learning_rate": 1e-06, + "loss": 0.237, + "mean_token_accuracy": 0.9103745222091675, + "num_tokens": 650167338.0, + "step": 16751 + }, + { + "epoch": 2.1310265869482254, + "grad_norm": 2.8441436290740967, + "learning_rate": 1e-06, + "loss": 0.2284, + "mean_token_accuracy": 0.9163768291473389, + "num_tokens": 650205789.0, + "step": 16752 + }, + { + "epoch": 2.131153797226816, + "grad_norm": 2.6022894382476807, + "learning_rate": 1e-06, + "loss": 0.2347, + "mean_token_accuracy": 0.9118897914886475, + "num_tokens": 650251991.0, + "step": 16753 + }, + { + "epoch": 2.1312810075054065, + "grad_norm": 2.987257480621338, + "learning_rate": 1e-06, + "loss": 0.2966, + "mean_token_accuracy": 0.8926796317100525, + "num_tokens": 650291840.0, + "step": 16754 + }, + { + "epoch": 2.131408217783997, + "grad_norm": 2.959242105484009, + "learning_rate": 1e-06, + "loss": 0.2342, + "mean_token_accuracy": 0.9128187894821167, + "num_tokens": 650325809.0, + "step": 16755 + }, + { + "epoch": 2.1315354280625876, + "grad_norm": 3.042443037033081, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9036481380462646, + "num_tokens": 650363060.0, + "step": 16756 + }, + { + "epoch": 2.131662638341178, + "grad_norm": 2.9379024505615234, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9124443531036377, + "num_tokens": 650402500.0, + "step": 16757 + }, + { + "epoch": 2.1317898486197686, + "grad_norm": 3.227940559387207, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.9080116748809814, + "num_tokens": 650435853.0, + "step": 16758 + }, + { + "epoch": 2.131917058898359, + "grad_norm": 2.796544313430786, + "learning_rate": 1e-06, + "loss": 0.2205, + "mean_token_accuracy": 0.9153778553009033, + "num_tokens": 650472886.0, + "step": 16759 + }, + { + "epoch": 2.1320442691769497, + "grad_norm": 2.83674955368042, + "learning_rate": 1e-06, + "loss": 0.2378, + "mean_token_accuracy": 0.9112162590026855, + "num_tokens": 650511738.0, + "step": 16760 + }, + { + "epoch": 2.13217147945554, + "grad_norm": 2.796905755996704, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9133543372154236, + "num_tokens": 650548961.0, + "step": 16761 + }, + { + "epoch": 2.1322986897341307, + "grad_norm": 3.252815008163452, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9073442220687866, + "num_tokens": 650582435.0, + "step": 16762 + }, + { + "epoch": 2.132425900012721, + "grad_norm": 2.75593638420105, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9155380725860596, + "num_tokens": 650624493.0, + "step": 16763 + }, + { + "epoch": 2.1325531102913113, + "grad_norm": 2.91611647605896, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9055659174919128, + "num_tokens": 650663818.0, + "step": 16764 + }, + { + "epoch": 2.132680320569902, + "grad_norm": 3.072936773300171, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.904409646987915, + "num_tokens": 650698899.0, + "step": 16765 + }, + { + "epoch": 2.1328075308484924, + "grad_norm": 2.7109835147857666, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.911772608757019, + "num_tokens": 650737713.0, + "step": 16766 + }, + { + "epoch": 2.132934741127083, + "grad_norm": 2.8948142528533936, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8997752666473389, + "num_tokens": 650780219.0, + "step": 16767 + }, + { + "epoch": 2.1330619514056735, + "grad_norm": 2.889777421951294, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8979953527450562, + "num_tokens": 650828790.0, + "step": 16768 + }, + { + "epoch": 2.133189161684264, + "grad_norm": 3.033656120300293, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9053938388824463, + "num_tokens": 650865716.0, + "step": 16769 + }, + { + "epoch": 2.1333163719628545, + "grad_norm": 2.9576544761657715, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.901412308216095, + "num_tokens": 650903854.0, + "step": 16770 + }, + { + "epoch": 2.133443582241445, + "grad_norm": 2.846798896789551, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.901418149471283, + "num_tokens": 650944516.0, + "step": 16771 + }, + { + "epoch": 2.1335707925200356, + "grad_norm": 3.1764726638793945, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9016343355178833, + "num_tokens": 650979222.0, + "step": 16772 + }, + { + "epoch": 2.133698002798626, + "grad_norm": 2.7360475063323975, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.8989377021789551, + "num_tokens": 651023312.0, + "step": 16773 + }, + { + "epoch": 2.1338252130772166, + "grad_norm": 2.6202354431152344, + "learning_rate": 1e-06, + "loss": 0.2239, + "mean_token_accuracy": 0.9134336709976196, + "num_tokens": 651063913.0, + "step": 16774 + }, + { + "epoch": 2.133952423355807, + "grad_norm": 3.062633514404297, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9044201374053955, + "num_tokens": 651100090.0, + "step": 16775 + }, + { + "epoch": 2.1340796336343977, + "grad_norm": 3.194138765335083, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.898394763469696, + "num_tokens": 651136608.0, + "step": 16776 + }, + { + "epoch": 2.134206843912988, + "grad_norm": 2.6583850383758545, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.905064582824707, + "num_tokens": 651181898.0, + "step": 16777 + }, + { + "epoch": 2.1343340541915787, + "grad_norm": 2.857287883758545, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.903019905090332, + "num_tokens": 651225113.0, + "step": 16778 + }, + { + "epoch": 2.1344612644701693, + "grad_norm": 2.7572031021118164, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.913596510887146, + "num_tokens": 651260724.0, + "step": 16779 + }, + { + "epoch": 2.13458847474876, + "grad_norm": 3.0081963539123535, + "learning_rate": 1e-06, + "loss": 0.2442, + "mean_token_accuracy": 0.907588005065918, + "num_tokens": 651296852.0, + "step": 16780 + }, + { + "epoch": 2.1347156850273503, + "grad_norm": 2.9544272422790527, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9029337167739868, + "num_tokens": 651335846.0, + "step": 16781 + }, + { + "epoch": 2.134842895305941, + "grad_norm": 2.585484027862549, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9096488356590271, + "num_tokens": 651381217.0, + "step": 16782 + }, + { + "epoch": 2.1349701055845314, + "grad_norm": 3.0952401161193848, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9037647247314453, + "num_tokens": 651416379.0, + "step": 16783 + }, + { + "epoch": 2.135097315863122, + "grad_norm": 2.62617564201355, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9082003235816956, + "num_tokens": 651458371.0, + "step": 16784 + }, + { + "epoch": 2.1352245261417124, + "grad_norm": 3.3957126140594482, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8927879333496094, + "num_tokens": 651500142.0, + "step": 16785 + }, + { + "epoch": 2.135351736420303, + "grad_norm": 2.7907190322875977, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9087178707122803, + "num_tokens": 651542405.0, + "step": 16786 + }, + { + "epoch": 2.1354789466988935, + "grad_norm": 2.832460403442383, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9136843681335449, + "num_tokens": 651580342.0, + "step": 16787 + }, + { + "epoch": 2.1356061569774836, + "grad_norm": 3.0531983375549316, + "learning_rate": 1e-06, + "loss": 0.2254, + "mean_token_accuracy": 0.9145404100418091, + "num_tokens": 651612018.0, + "step": 16788 + }, + { + "epoch": 2.135733367256074, + "grad_norm": 2.7961266040802, + "learning_rate": 1e-06, + "loss": 0.2288, + "mean_token_accuracy": 0.9143700003623962, + "num_tokens": 651649612.0, + "step": 16789 + }, + { + "epoch": 2.1358605775346646, + "grad_norm": 2.7523252964019775, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9012831449508667, + "num_tokens": 651695459.0, + "step": 16790 + }, + { + "epoch": 2.135987787813255, + "grad_norm": 3.1762149333953857, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8891440033912659, + "num_tokens": 651730800.0, + "step": 16791 + }, + { + "epoch": 2.1361149980918457, + "grad_norm": 2.782783031463623, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9098400473594666, + "num_tokens": 651771555.0, + "step": 16792 + }, + { + "epoch": 2.136242208370436, + "grad_norm": 2.7923340797424316, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9070285558700562, + "num_tokens": 651816069.0, + "step": 16793 + }, + { + "epoch": 2.1363694186490267, + "grad_norm": 2.7704367637634277, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9073237180709839, + "num_tokens": 651857397.0, + "step": 16794 + }, + { + "epoch": 2.1364966289276173, + "grad_norm": 3.2626051902770996, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.9002839922904968, + "num_tokens": 651892290.0, + "step": 16795 + }, + { + "epoch": 2.136623839206208, + "grad_norm": 3.0388238430023193, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.8994064331054688, + "num_tokens": 651932887.0, + "step": 16796 + }, + { + "epoch": 2.1367510494847983, + "grad_norm": 2.9623494148254395, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9063460826873779, + "num_tokens": 651973616.0, + "step": 16797 + }, + { + "epoch": 2.136878259763389, + "grad_norm": 3.0018739700317383, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9055473804473877, + "num_tokens": 652011750.0, + "step": 16798 + }, + { + "epoch": 2.1370054700419794, + "grad_norm": 2.9656107425689697, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9136841297149658, + "num_tokens": 652045753.0, + "step": 16799 + }, + { + "epoch": 2.13713268032057, + "grad_norm": 2.7201616764068604, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9057385921478271, + "num_tokens": 652091222.0, + "step": 16800 + }, + { + "epoch": 2.1372598905991604, + "grad_norm": 2.955193042755127, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.901889443397522, + "num_tokens": 652129831.0, + "step": 16801 + }, + { + "epoch": 2.137387100877751, + "grad_norm": 2.6260194778442383, + "learning_rate": 1e-06, + "loss": 0.2195, + "mean_token_accuracy": 0.9207214117050171, + "num_tokens": 652170577.0, + "step": 16802 + }, + { + "epoch": 2.1375143111563415, + "grad_norm": 2.852501630783081, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9074331521987915, + "num_tokens": 652211665.0, + "step": 16803 + }, + { + "epoch": 2.137641521434932, + "grad_norm": 2.85101056098938, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.9099994897842407, + "num_tokens": 652248162.0, + "step": 16804 + }, + { + "epoch": 2.1377687317135226, + "grad_norm": 2.95402193069458, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8946061134338379, + "num_tokens": 652287290.0, + "step": 16805 + }, + { + "epoch": 2.137895941992113, + "grad_norm": 3.3898134231567383, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8916764855384827, + "num_tokens": 652322030.0, + "step": 16806 + }, + { + "epoch": 2.1380231522707036, + "grad_norm": 2.8944482803344727, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.8987998962402344, + "num_tokens": 652361684.0, + "step": 16807 + }, + { + "epoch": 2.138150362549294, + "grad_norm": 3.003200054168701, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9061325788497925, + "num_tokens": 652395045.0, + "step": 16808 + }, + { + "epoch": 2.1382775728278847, + "grad_norm": 2.92081880569458, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9024239778518677, + "num_tokens": 652432087.0, + "step": 16809 + }, + { + "epoch": 2.138404783106475, + "grad_norm": 2.670651435852051, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9061084985733032, + "num_tokens": 652472257.0, + "step": 16810 + }, + { + "epoch": 2.1385319933850653, + "grad_norm": 2.6620240211486816, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9101518392562866, + "num_tokens": 652511645.0, + "step": 16811 + }, + { + "epoch": 2.1386592036636562, + "grad_norm": 2.9198455810546875, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9065701961517334, + "num_tokens": 652546413.0, + "step": 16812 + }, + { + "epoch": 2.1387864139422463, + "grad_norm": 2.645458698272705, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9119782447814941, + "num_tokens": 652588265.0, + "step": 16813 + }, + { + "epoch": 2.138913624220837, + "grad_norm": 2.9958417415618896, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.896874725818634, + "num_tokens": 652628606.0, + "step": 16814 + }, + { + "epoch": 2.1390408344994274, + "grad_norm": 3.056562662124634, + "learning_rate": 1e-06, + "loss": 0.2308, + "mean_token_accuracy": 0.9150388240814209, + "num_tokens": 652659153.0, + "step": 16815 + }, + { + "epoch": 2.139168044778018, + "grad_norm": 2.720430374145508, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9064989686012268, + "num_tokens": 652702346.0, + "step": 16816 + }, + { + "epoch": 2.1392952550566084, + "grad_norm": 3.101517677307129, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9003497362136841, + "num_tokens": 652737031.0, + "step": 16817 + }, + { + "epoch": 2.139422465335199, + "grad_norm": 2.5888748168945312, + "learning_rate": 1e-06, + "loss": 0.2118, + "mean_token_accuracy": 0.9220375418663025, + "num_tokens": 652777964.0, + "step": 16818 + }, + { + "epoch": 2.1395496756137895, + "grad_norm": 2.9448647499084473, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9034798741340637, + "num_tokens": 652818720.0, + "step": 16819 + }, + { + "epoch": 2.13967688589238, + "grad_norm": 2.8995165824890137, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9032899141311646, + "num_tokens": 652860533.0, + "step": 16820 + }, + { + "epoch": 2.1398040961709706, + "grad_norm": 2.88226318359375, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8962663412094116, + "num_tokens": 652900286.0, + "step": 16821 + }, + { + "epoch": 2.139931306449561, + "grad_norm": 3.04178786277771, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9001362919807434, + "num_tokens": 652938538.0, + "step": 16822 + }, + { + "epoch": 2.1400585167281516, + "grad_norm": 3.057878017425537, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.903550386428833, + "num_tokens": 652975869.0, + "step": 16823 + }, + { + "epoch": 2.140185727006742, + "grad_norm": 2.970085859298706, + "learning_rate": 1e-06, + "loss": 0.2305, + "mean_token_accuracy": 0.9157201051712036, + "num_tokens": 653011648.0, + "step": 16824 + }, + { + "epoch": 2.1403129372853327, + "grad_norm": 3.128340721130371, + "learning_rate": 1e-06, + "loss": 0.2438, + "mean_token_accuracy": 0.9096033573150635, + "num_tokens": 653044211.0, + "step": 16825 + }, + { + "epoch": 2.140440147563923, + "grad_norm": 3.125636339187622, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9065502882003784, + "num_tokens": 653076565.0, + "step": 16826 + }, + { + "epoch": 2.1405673578425137, + "grad_norm": 3.072887897491455, + "learning_rate": 1e-06, + "loss": 0.3013, + "mean_token_accuracy": 0.8921428322792053, + "num_tokens": 653114439.0, + "step": 16827 + }, + { + "epoch": 2.1406945681211043, + "grad_norm": 2.9705755710601807, + "learning_rate": 1e-06, + "loss": 0.2313, + "mean_token_accuracy": 0.9146239757537842, + "num_tokens": 653147740.0, + "step": 16828 + }, + { + "epoch": 2.140821778399695, + "grad_norm": 2.805434226989746, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8971453905105591, + "num_tokens": 653195687.0, + "step": 16829 + }, + { + "epoch": 2.1409489886782853, + "grad_norm": 3.1872599124908447, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9051726460456848, + "num_tokens": 653232616.0, + "step": 16830 + }, + { + "epoch": 2.141076198956876, + "grad_norm": 2.8813202381134033, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9013551473617554, + "num_tokens": 653272856.0, + "step": 16831 + }, + { + "epoch": 2.1412034092354664, + "grad_norm": 3.1108944416046143, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8898240327835083, + "num_tokens": 653311635.0, + "step": 16832 + }, + { + "epoch": 2.141330619514057, + "grad_norm": 2.7854673862457275, + "learning_rate": 1e-06, + "loss": 0.2437, + "mean_token_accuracy": 0.9100689888000488, + "num_tokens": 653352271.0, + "step": 16833 + }, + { + "epoch": 2.1414578297926474, + "grad_norm": 2.6266062259674072, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.908959686756134, + "num_tokens": 653392614.0, + "step": 16834 + }, + { + "epoch": 2.141585040071238, + "grad_norm": 2.995448112487793, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9048624038696289, + "num_tokens": 653428703.0, + "step": 16835 + }, + { + "epoch": 2.141712250349828, + "grad_norm": 2.7184159755706787, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9090981483459473, + "num_tokens": 653467429.0, + "step": 16836 + }, + { + "epoch": 2.141839460628419, + "grad_norm": 3.0501086711883545, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.900137186050415, + "num_tokens": 653504370.0, + "step": 16837 + }, + { + "epoch": 2.141966670907009, + "grad_norm": 2.663220167160034, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.907257080078125, + "num_tokens": 653549022.0, + "step": 16838 + }, + { + "epoch": 2.1420938811855996, + "grad_norm": 2.9020612239837646, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9044713973999023, + "num_tokens": 653587966.0, + "step": 16839 + }, + { + "epoch": 2.14222109146419, + "grad_norm": 2.6456663608551025, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9121356010437012, + "num_tokens": 653629432.0, + "step": 16840 + }, + { + "epoch": 2.1423483017427807, + "grad_norm": 2.6296088695526123, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9098093509674072, + "num_tokens": 653670983.0, + "step": 16841 + }, + { + "epoch": 2.142475512021371, + "grad_norm": 2.734776020050049, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9084687829017639, + "num_tokens": 653711166.0, + "step": 16842 + }, + { + "epoch": 2.1426027222999617, + "grad_norm": 2.645643711090088, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9035601615905762, + "num_tokens": 653757844.0, + "step": 16843 + }, + { + "epoch": 2.1427299325785523, + "grad_norm": 2.715050458908081, + "learning_rate": 1e-06, + "loss": 0.2334, + "mean_token_accuracy": 0.9136875867843628, + "num_tokens": 653797882.0, + "step": 16844 + }, + { + "epoch": 2.142857142857143, + "grad_norm": 2.8722355365753174, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8940684795379639, + "num_tokens": 653842099.0, + "step": 16845 + }, + { + "epoch": 2.1429843531357333, + "grad_norm": 3.010439872741699, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9103395342826843, + "num_tokens": 653877868.0, + "step": 16846 + }, + { + "epoch": 2.143111563414324, + "grad_norm": 2.532256841659546, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9105088114738464, + "num_tokens": 653921168.0, + "step": 16847 + }, + { + "epoch": 2.1432387736929144, + "grad_norm": 2.891402006149292, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9126759171485901, + "num_tokens": 653955554.0, + "step": 16848 + }, + { + "epoch": 2.143365983971505, + "grad_norm": 2.8215882778167725, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.907126247882843, + "num_tokens": 653994472.0, + "step": 16849 + }, + { + "epoch": 2.1434931942500954, + "grad_norm": 2.931889057159424, + "learning_rate": 1e-06, + "loss": 0.2203, + "mean_token_accuracy": 0.9186860918998718, + "num_tokens": 654033070.0, + "step": 16850 + }, + { + "epoch": 2.143620404528686, + "grad_norm": 2.908385992050171, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8953495025634766, + "num_tokens": 654074546.0, + "step": 16851 + }, + { + "epoch": 2.1437476148072765, + "grad_norm": 3.0904855728149414, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8949498534202576, + "num_tokens": 654113189.0, + "step": 16852 + }, + { + "epoch": 2.143874825085867, + "grad_norm": 3.421572208404541, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8929234147071838, + "num_tokens": 654148097.0, + "step": 16853 + }, + { + "epoch": 2.1440020353644575, + "grad_norm": 2.755852222442627, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9112823009490967, + "num_tokens": 654186528.0, + "step": 16854 + }, + { + "epoch": 2.144129245643048, + "grad_norm": 2.90147066116333, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9073481559753418, + "num_tokens": 654225475.0, + "step": 16855 + }, + { + "epoch": 2.1442564559216386, + "grad_norm": 2.88586163520813, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9027928113937378, + "num_tokens": 654268976.0, + "step": 16856 + }, + { + "epoch": 2.144383666200229, + "grad_norm": 2.8539164066314697, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9064328074455261, + "num_tokens": 654307542.0, + "step": 16857 + }, + { + "epoch": 2.1445108764788197, + "grad_norm": 3.012190580368042, + "learning_rate": 1e-06, + "loss": 0.2331, + "mean_token_accuracy": 0.9143102169036865, + "num_tokens": 654342470.0, + "step": 16858 + }, + { + "epoch": 2.14463808675741, + "grad_norm": 2.939694404602051, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9072505831718445, + "num_tokens": 654380524.0, + "step": 16859 + }, + { + "epoch": 2.1447652970360007, + "grad_norm": 2.8220577239990234, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.896719217300415, + "num_tokens": 654420001.0, + "step": 16860 + }, + { + "epoch": 2.144892507314591, + "grad_norm": 2.6512420177459717, + "learning_rate": 1e-06, + "loss": 0.2398, + "mean_token_accuracy": 0.911884069442749, + "num_tokens": 654464597.0, + "step": 16861 + }, + { + "epoch": 2.1450197175931813, + "grad_norm": 3.053929090499878, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8982027173042297, + "num_tokens": 654503220.0, + "step": 16862 + }, + { + "epoch": 2.145146927871772, + "grad_norm": 3.15791654586792, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.897992730140686, + "num_tokens": 654536528.0, + "step": 16863 + }, + { + "epoch": 2.1452741381503624, + "grad_norm": 3.015605926513672, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9074624180793762, + "num_tokens": 654570407.0, + "step": 16864 + }, + { + "epoch": 2.145401348428953, + "grad_norm": 2.940974473953247, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.9115523099899292, + "num_tokens": 654610116.0, + "step": 16865 + }, + { + "epoch": 2.1455285587075434, + "grad_norm": 2.7856271266937256, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8963401913642883, + "num_tokens": 654653336.0, + "step": 16866 + }, + { + "epoch": 2.145655768986134, + "grad_norm": 2.974811553955078, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.8983900547027588, + "num_tokens": 654691402.0, + "step": 16867 + }, + { + "epoch": 2.1457829792647245, + "grad_norm": 2.6712448596954346, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9103905558586121, + "num_tokens": 654737142.0, + "step": 16868 + }, + { + "epoch": 2.145910189543315, + "grad_norm": 2.7796685695648193, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9095149040222168, + "num_tokens": 654776317.0, + "step": 16869 + }, + { + "epoch": 2.1460373998219056, + "grad_norm": 2.679960250854492, + "learning_rate": 1e-06, + "loss": 0.2325, + "mean_token_accuracy": 0.9118368029594421, + "num_tokens": 654816282.0, + "step": 16870 + }, + { + "epoch": 2.146164610100496, + "grad_norm": 2.6853535175323486, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8989067077636719, + "num_tokens": 654862138.0, + "step": 16871 + }, + { + "epoch": 2.1462918203790866, + "grad_norm": 2.892228841781616, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9031549096107483, + "num_tokens": 654900709.0, + "step": 16872 + }, + { + "epoch": 2.146419030657677, + "grad_norm": 3.1437602043151855, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9028683304786682, + "num_tokens": 654935660.0, + "step": 16873 + }, + { + "epoch": 2.1465462409362677, + "grad_norm": 2.8301045894622803, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9080127477645874, + "num_tokens": 654975794.0, + "step": 16874 + }, + { + "epoch": 2.146673451214858, + "grad_norm": 2.945924758911133, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9043446779251099, + "num_tokens": 655012939.0, + "step": 16875 + }, + { + "epoch": 2.1468006614934487, + "grad_norm": 2.9548733234405518, + "learning_rate": 1e-06, + "loss": 0.2277, + "mean_token_accuracy": 0.9153749346733093, + "num_tokens": 655049594.0, + "step": 16876 + }, + { + "epoch": 2.1469278717720393, + "grad_norm": 3.001096725463867, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8946599364280701, + "num_tokens": 655088659.0, + "step": 16877 + }, + { + "epoch": 2.14705508205063, + "grad_norm": 3.06705379486084, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9026861190795898, + "num_tokens": 655124505.0, + "step": 16878 + }, + { + "epoch": 2.1471822923292203, + "grad_norm": 2.554046154022217, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9099801182746887, + "num_tokens": 655168318.0, + "step": 16879 + }, + { + "epoch": 2.147309502607811, + "grad_norm": 2.7328059673309326, + "learning_rate": 1e-06, + "loss": 0.2405, + "mean_token_accuracy": 0.9098390936851501, + "num_tokens": 655209844.0, + "step": 16880 + }, + { + "epoch": 2.1474367128864014, + "grad_norm": 2.821772575378418, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9007209539413452, + "num_tokens": 655250055.0, + "step": 16881 + }, + { + "epoch": 2.147563923164992, + "grad_norm": 2.9190707206726074, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9062982201576233, + "num_tokens": 655285167.0, + "step": 16882 + }, + { + "epoch": 2.1476911334435824, + "grad_norm": 3.123797655105591, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.910358190536499, + "num_tokens": 655320604.0, + "step": 16883 + }, + { + "epoch": 2.147818343722173, + "grad_norm": 3.021836996078491, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9109016060829163, + "num_tokens": 655351827.0, + "step": 16884 + }, + { + "epoch": 2.1479455540007635, + "grad_norm": 3.0891168117523193, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.907764732837677, + "num_tokens": 655389114.0, + "step": 16885 + }, + { + "epoch": 2.1480727642793536, + "grad_norm": 2.577655792236328, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.9139106869697571, + "num_tokens": 655433106.0, + "step": 16886 + }, + { + "epoch": 2.148199974557944, + "grad_norm": 2.8408823013305664, + "learning_rate": 1e-06, + "loss": 0.2165, + "mean_token_accuracy": 0.9206531047821045, + "num_tokens": 655470132.0, + "step": 16887 + }, + { + "epoch": 2.1483271848365346, + "grad_norm": 2.644956588745117, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9054194688796997, + "num_tokens": 655511741.0, + "step": 16888 + }, + { + "epoch": 2.148454395115125, + "grad_norm": 2.888024091720581, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9033825397491455, + "num_tokens": 655549789.0, + "step": 16889 + }, + { + "epoch": 2.1485816053937157, + "grad_norm": 2.772951364517212, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.910951554775238, + "num_tokens": 655588599.0, + "step": 16890 + }, + { + "epoch": 2.148708815672306, + "grad_norm": 3.085606575012207, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9125721454620361, + "num_tokens": 655620480.0, + "step": 16891 + }, + { + "epoch": 2.1488360259508967, + "grad_norm": 2.7655117511749268, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.898476243019104, + "num_tokens": 655666822.0, + "step": 16892 + }, + { + "epoch": 2.1489632362294873, + "grad_norm": 2.8125061988830566, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9069350957870483, + "num_tokens": 655706771.0, + "step": 16893 + }, + { + "epoch": 2.149090446508078, + "grad_norm": 3.1938588619232178, + "learning_rate": 1e-06, + "loss": 0.2347, + "mean_token_accuracy": 0.9130337238311768, + "num_tokens": 655735204.0, + "step": 16894 + }, + { + "epoch": 2.1492176567866683, + "grad_norm": 2.8854594230651855, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9028319716453552, + "num_tokens": 655778695.0, + "step": 16895 + }, + { + "epoch": 2.149344867065259, + "grad_norm": 2.8185641765594482, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.906470775604248, + "num_tokens": 655818514.0, + "step": 16896 + }, + { + "epoch": 2.1494720773438494, + "grad_norm": 2.918029308319092, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9109443426132202, + "num_tokens": 655853102.0, + "step": 16897 + }, + { + "epoch": 2.14959928762244, + "grad_norm": 2.809446096420288, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9097527265548706, + "num_tokens": 655893536.0, + "step": 16898 + }, + { + "epoch": 2.1497264979010304, + "grad_norm": 3.1647255420684814, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.9018625020980835, + "num_tokens": 655929065.0, + "step": 16899 + }, + { + "epoch": 2.149853708179621, + "grad_norm": 2.707935333251953, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.905994176864624, + "num_tokens": 655969037.0, + "step": 16900 + }, + { + "epoch": 2.1499809184582115, + "grad_norm": 2.8188440799713135, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.912514328956604, + "num_tokens": 656006686.0, + "step": 16901 + }, + { + "epoch": 2.150108128736802, + "grad_norm": 2.83465838432312, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9102903604507446, + "num_tokens": 656042905.0, + "step": 16902 + }, + { + "epoch": 2.1502353390153925, + "grad_norm": 2.9666504859924316, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.902662456035614, + "num_tokens": 656079814.0, + "step": 16903 + }, + { + "epoch": 2.150362549293983, + "grad_norm": 3.4061694145202637, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8954291939735413, + "num_tokens": 656112640.0, + "step": 16904 + }, + { + "epoch": 2.1504897595725736, + "grad_norm": 2.9061553478240967, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9063621759414673, + "num_tokens": 656148174.0, + "step": 16905 + }, + { + "epoch": 2.150616969851164, + "grad_norm": 3.0248820781707764, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9012570381164551, + "num_tokens": 656183457.0, + "step": 16906 + }, + { + "epoch": 2.1507441801297547, + "grad_norm": 2.69446063041687, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9119933843612671, + "num_tokens": 656222170.0, + "step": 16907 + }, + { + "epoch": 2.150871390408345, + "grad_norm": 2.717008113861084, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9094098210334778, + "num_tokens": 656262099.0, + "step": 16908 + }, + { + "epoch": 2.1509986006869353, + "grad_norm": 2.7819769382476807, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9076098799705505, + "num_tokens": 656303906.0, + "step": 16909 + }, + { + "epoch": 2.1511258109655262, + "grad_norm": 2.6690144538879395, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9101465940475464, + "num_tokens": 656348799.0, + "step": 16910 + }, + { + "epoch": 2.1512530212441163, + "grad_norm": 2.7970664501190186, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9064648151397705, + "num_tokens": 656390602.0, + "step": 16911 + }, + { + "epoch": 2.151380231522707, + "grad_norm": 3.0351622104644775, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8941217660903931, + "num_tokens": 656430260.0, + "step": 16912 + }, + { + "epoch": 2.1515074418012974, + "grad_norm": 3.1421470642089844, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9035963416099548, + "num_tokens": 656465922.0, + "step": 16913 + }, + { + "epoch": 2.151634652079888, + "grad_norm": 2.660360813140869, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9092947840690613, + "num_tokens": 656511731.0, + "step": 16914 + }, + { + "epoch": 2.1517618623584784, + "grad_norm": 2.736717939376831, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9049177765846252, + "num_tokens": 656556010.0, + "step": 16915 + }, + { + "epoch": 2.151889072637069, + "grad_norm": 2.817976951599121, + "learning_rate": 1e-06, + "loss": 0.2443, + "mean_token_accuracy": 0.9085697531700134, + "num_tokens": 656596993.0, + "step": 16916 + }, + { + "epoch": 2.1520162829156595, + "grad_norm": 2.6684184074401855, + "learning_rate": 1e-06, + "loss": 0.2338, + "mean_token_accuracy": 0.911757230758667, + "num_tokens": 656636376.0, + "step": 16917 + }, + { + "epoch": 2.15214349319425, + "grad_norm": 3.176110029220581, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8924767971038818, + "num_tokens": 656674717.0, + "step": 16918 + }, + { + "epoch": 2.1522707034728406, + "grad_norm": 2.9994630813598633, + "learning_rate": 1e-06, + "loss": 0.2176, + "mean_token_accuracy": 0.9199718236923218, + "num_tokens": 656707316.0, + "step": 16919 + }, + { + "epoch": 2.152397913751431, + "grad_norm": 2.7860989570617676, + "learning_rate": 1e-06, + "loss": 0.2394, + "mean_token_accuracy": 0.9134459495544434, + "num_tokens": 656745167.0, + "step": 16920 + }, + { + "epoch": 2.1525251240300216, + "grad_norm": 2.7265474796295166, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9064540863037109, + "num_tokens": 656789280.0, + "step": 16921 + }, + { + "epoch": 2.152652334308612, + "grad_norm": 2.98103928565979, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.8990683555603027, + "num_tokens": 656829076.0, + "step": 16922 + }, + { + "epoch": 2.1527795445872027, + "grad_norm": 2.7963976860046387, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9121246933937073, + "num_tokens": 656867065.0, + "step": 16923 + }, + { + "epoch": 2.152906754865793, + "grad_norm": 2.6613636016845703, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9118843674659729, + "num_tokens": 656904670.0, + "step": 16924 + }, + { + "epoch": 2.1530339651443837, + "grad_norm": 2.918884754180908, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9023642539978027, + "num_tokens": 656942644.0, + "step": 16925 + }, + { + "epoch": 2.1531611754229742, + "grad_norm": 2.7301242351531982, + "learning_rate": 1e-06, + "loss": 0.2259, + "mean_token_accuracy": 0.9155261516571045, + "num_tokens": 656978985.0, + "step": 16926 + }, + { + "epoch": 2.1532883857015648, + "grad_norm": 2.555549144744873, + "learning_rate": 1e-06, + "loss": 0.2072, + "mean_token_accuracy": 0.9225375652313232, + "num_tokens": 657021091.0, + "step": 16927 + }, + { + "epoch": 2.1534155959801553, + "grad_norm": 2.6272287368774414, + "learning_rate": 1e-06, + "loss": 0.2301, + "mean_token_accuracy": 0.9111775755882263, + "num_tokens": 657062726.0, + "step": 16928 + }, + { + "epoch": 2.153542806258746, + "grad_norm": 2.4482884407043457, + "learning_rate": 1e-06, + "loss": 0.2309, + "mean_token_accuracy": 0.9152213335037231, + "num_tokens": 657106429.0, + "step": 16929 + }, + { + "epoch": 2.1536700165373364, + "grad_norm": 2.760085105895996, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9104044437408447, + "num_tokens": 657142018.0, + "step": 16930 + }, + { + "epoch": 2.153797226815927, + "grad_norm": 3.2038753032684326, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9071186780929565, + "num_tokens": 657173679.0, + "step": 16931 + }, + { + "epoch": 2.1539244370945174, + "grad_norm": 2.632802724838257, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9138343334197998, + "num_tokens": 657214931.0, + "step": 16932 + }, + { + "epoch": 2.154051647373108, + "grad_norm": 3.420060873031616, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8909472823143005, + "num_tokens": 657249026.0, + "step": 16933 + }, + { + "epoch": 2.154178857651698, + "grad_norm": 3.245094060897827, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9032679796218872, + "num_tokens": 657281446.0, + "step": 16934 + }, + { + "epoch": 2.1543060679302886, + "grad_norm": 2.766432285308838, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9070091843605042, + "num_tokens": 657322351.0, + "step": 16935 + }, + { + "epoch": 2.154433278208879, + "grad_norm": 3.2325146198272705, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9055794477462769, + "num_tokens": 657353989.0, + "step": 16936 + }, + { + "epoch": 2.1545604884874696, + "grad_norm": 2.7560505867004395, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9109789133071899, + "num_tokens": 657393793.0, + "step": 16937 + }, + { + "epoch": 2.15468769876606, + "grad_norm": 3.002058744430542, + "learning_rate": 1e-06, + "loss": 0.2337, + "mean_token_accuracy": 0.9114152193069458, + "num_tokens": 657427959.0, + "step": 16938 + }, + { + "epoch": 2.1548149090446507, + "grad_norm": 2.9167559146881104, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9055718183517456, + "num_tokens": 657466175.0, + "step": 16939 + }, + { + "epoch": 2.154942119323241, + "grad_norm": 2.884246349334717, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9087613821029663, + "num_tokens": 657504670.0, + "step": 16940 + }, + { + "epoch": 2.1550693296018317, + "grad_norm": 3.1380980014801025, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9074535369873047, + "num_tokens": 657539940.0, + "step": 16941 + }, + { + "epoch": 2.1551965398804223, + "grad_norm": 2.8106603622436523, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9036633968353271, + "num_tokens": 657582306.0, + "step": 16942 + }, + { + "epoch": 2.155323750159013, + "grad_norm": 2.9682397842407227, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9074198007583618, + "num_tokens": 657618176.0, + "step": 16943 + }, + { + "epoch": 2.1554509604376033, + "grad_norm": 2.989672899246216, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8999941349029541, + "num_tokens": 657656913.0, + "step": 16944 + }, + { + "epoch": 2.155578170716194, + "grad_norm": 2.8919317722320557, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9010113477706909, + "num_tokens": 657696148.0, + "step": 16945 + }, + { + "epoch": 2.1557053809947844, + "grad_norm": 2.8061435222625732, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9076813459396362, + "num_tokens": 657734812.0, + "step": 16946 + }, + { + "epoch": 2.155832591273375, + "grad_norm": 2.58937931060791, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9069129228591919, + "num_tokens": 657778777.0, + "step": 16947 + }, + { + "epoch": 2.1559598015519654, + "grad_norm": 2.8808653354644775, + "learning_rate": 1e-06, + "loss": 0.2342, + "mean_token_accuracy": 0.9132912158966064, + "num_tokens": 657814126.0, + "step": 16948 + }, + { + "epoch": 2.156087011830556, + "grad_norm": 2.4603471755981445, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9151383638381958, + "num_tokens": 657857283.0, + "step": 16949 + }, + { + "epoch": 2.1562142221091465, + "grad_norm": 2.9476470947265625, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9035509824752808, + "num_tokens": 657893880.0, + "step": 16950 + }, + { + "epoch": 2.156341432387737, + "grad_norm": 2.777329921722412, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9057308435440063, + "num_tokens": 657935210.0, + "step": 16951 + }, + { + "epoch": 2.1564686426663275, + "grad_norm": 2.844010829925537, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8946160078048706, + "num_tokens": 657975434.0, + "step": 16952 + }, + { + "epoch": 2.156595852944918, + "grad_norm": 2.841718912124634, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9064069986343384, + "num_tokens": 658014991.0, + "step": 16953 + }, + { + "epoch": 2.1567230632235086, + "grad_norm": 2.7193996906280518, + "learning_rate": 1e-06, + "loss": 0.2315, + "mean_token_accuracy": 0.913972795009613, + "num_tokens": 658052170.0, + "step": 16954 + }, + { + "epoch": 2.156850273502099, + "grad_norm": 2.8883373737335205, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8947365283966064, + "num_tokens": 658096615.0, + "step": 16955 + }, + { + "epoch": 2.1569774837806897, + "grad_norm": 3.1464991569519043, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9030949473381042, + "num_tokens": 658128401.0, + "step": 16956 + }, + { + "epoch": 2.15710469405928, + "grad_norm": 2.847097635269165, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8980672955513, + "num_tokens": 658169630.0, + "step": 16957 + }, + { + "epoch": 2.1572319043378707, + "grad_norm": 2.7415030002593994, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9101377725601196, + "num_tokens": 658209902.0, + "step": 16958 + }, + { + "epoch": 2.157359114616461, + "grad_norm": 2.8285107612609863, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.8981859683990479, + "num_tokens": 658250274.0, + "step": 16959 + }, + { + "epoch": 2.1574863248950513, + "grad_norm": 3.1447253227233887, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9080698490142822, + "num_tokens": 658283454.0, + "step": 16960 + }, + { + "epoch": 2.157613535173642, + "grad_norm": 2.7860803604125977, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9072979092597961, + "num_tokens": 658327944.0, + "step": 16961 + }, + { + "epoch": 2.1577407454522324, + "grad_norm": 3.0863144397735596, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9065166115760803, + "num_tokens": 658365828.0, + "step": 16962 + }, + { + "epoch": 2.157867955730823, + "grad_norm": 2.6734297275543213, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9106312990188599, + "num_tokens": 658404653.0, + "step": 16963 + }, + { + "epoch": 2.1579951660094134, + "grad_norm": 2.756910800933838, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.911404550075531, + "num_tokens": 658445106.0, + "step": 16964 + }, + { + "epoch": 2.158122376288004, + "grad_norm": 3.2637312412261963, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8895832300186157, + "num_tokens": 658483196.0, + "step": 16965 + }, + { + "epoch": 2.1582495865665945, + "grad_norm": 2.880875587463379, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9005815982818604, + "num_tokens": 658525087.0, + "step": 16966 + }, + { + "epoch": 2.158376796845185, + "grad_norm": 3.1668126583099365, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.8995318412780762, + "num_tokens": 658561070.0, + "step": 16967 + }, + { + "epoch": 2.1585040071237755, + "grad_norm": 2.9068219661712646, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9075540900230408, + "num_tokens": 658602350.0, + "step": 16968 + }, + { + "epoch": 2.158631217402366, + "grad_norm": 2.878744602203369, + "learning_rate": 1e-06, + "loss": 0.2261, + "mean_token_accuracy": 0.9158570766448975, + "num_tokens": 658638003.0, + "step": 16969 + }, + { + "epoch": 2.1587584276809566, + "grad_norm": 2.7964305877685547, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9052982926368713, + "num_tokens": 658678233.0, + "step": 16970 + }, + { + "epoch": 2.158885637959547, + "grad_norm": 2.916271924972534, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9059842824935913, + "num_tokens": 658715211.0, + "step": 16971 + }, + { + "epoch": 2.1590128482381377, + "grad_norm": 2.8251805305480957, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9133346080780029, + "num_tokens": 658751577.0, + "step": 16972 + }, + { + "epoch": 2.159140058516728, + "grad_norm": 2.6962637901306152, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.910615086555481, + "num_tokens": 658791510.0, + "step": 16973 + }, + { + "epoch": 2.1592672687953187, + "grad_norm": 2.922074794769287, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9088936448097229, + "num_tokens": 658827467.0, + "step": 16974 + }, + { + "epoch": 2.1593944790739092, + "grad_norm": 2.86102557182312, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9053357839584351, + "num_tokens": 658866449.0, + "step": 16975 + }, + { + "epoch": 2.1595216893524998, + "grad_norm": 2.593456506729126, + "learning_rate": 1e-06, + "loss": 0.2248, + "mean_token_accuracy": 0.9136772155761719, + "num_tokens": 658907460.0, + "step": 16976 + }, + { + "epoch": 2.1596488996310903, + "grad_norm": 2.781064987182617, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9110842943191528, + "num_tokens": 658945231.0, + "step": 16977 + }, + { + "epoch": 2.159776109909681, + "grad_norm": 2.977327585220337, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9040520191192627, + "num_tokens": 658979091.0, + "step": 16978 + }, + { + "epoch": 2.1599033201882714, + "grad_norm": 2.971588134765625, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8965731859207153, + "num_tokens": 659017819.0, + "step": 16979 + }, + { + "epoch": 2.160030530466862, + "grad_norm": 2.84362530708313, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9052173495292664, + "num_tokens": 659056524.0, + "step": 16980 + }, + { + "epoch": 2.1601577407454524, + "grad_norm": 3.339759349822998, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8956707119941711, + "num_tokens": 659093723.0, + "step": 16981 + }, + { + "epoch": 2.160284951024043, + "grad_norm": 2.8454110622406006, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9059761166572571, + "num_tokens": 659133870.0, + "step": 16982 + }, + { + "epoch": 2.1604121613026335, + "grad_norm": 2.833966016769409, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9046623706817627, + "num_tokens": 659172917.0, + "step": 16983 + }, + { + "epoch": 2.1605393715812236, + "grad_norm": 3.0354464054107666, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8961093425750732, + "num_tokens": 659210161.0, + "step": 16984 + }, + { + "epoch": 2.160666581859814, + "grad_norm": 2.843939781188965, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9054254293441772, + "num_tokens": 659249707.0, + "step": 16985 + }, + { + "epoch": 2.1607937921384046, + "grad_norm": 3.0577738285064697, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.8991992473602295, + "num_tokens": 659284573.0, + "step": 16986 + }, + { + "epoch": 2.160921002416995, + "grad_norm": 2.957655668258667, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.8967863321304321, + "num_tokens": 659323824.0, + "step": 16987 + }, + { + "epoch": 2.1610482126955857, + "grad_norm": 2.6588456630706787, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9081676602363586, + "num_tokens": 659365463.0, + "step": 16988 + }, + { + "epoch": 2.161175422974176, + "grad_norm": 3.091696262359619, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8995473980903625, + "num_tokens": 659403746.0, + "step": 16989 + }, + { + "epoch": 2.1613026332527667, + "grad_norm": 2.8638060092926025, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9018243551254272, + "num_tokens": 659442458.0, + "step": 16990 + }, + { + "epoch": 2.1614298435313573, + "grad_norm": 3.239600658416748, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8956958055496216, + "num_tokens": 659480628.0, + "step": 16991 + }, + { + "epoch": 2.161557053809948, + "grad_norm": 2.985351800918579, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.8999847173690796, + "num_tokens": 659517237.0, + "step": 16992 + }, + { + "epoch": 2.1616842640885383, + "grad_norm": 2.5964131355285645, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9083883762359619, + "num_tokens": 659563894.0, + "step": 16993 + }, + { + "epoch": 2.161811474367129, + "grad_norm": 2.7007553577423096, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9057083129882812, + "num_tokens": 659604458.0, + "step": 16994 + }, + { + "epoch": 2.1619386846457194, + "grad_norm": 2.8310234546661377, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9102783799171448, + "num_tokens": 659645215.0, + "step": 16995 + }, + { + "epoch": 2.16206589492431, + "grad_norm": 2.743802070617676, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.8998158574104309, + "num_tokens": 659691094.0, + "step": 16996 + }, + { + "epoch": 2.1621931052029004, + "grad_norm": 2.7383580207824707, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.9033037424087524, + "num_tokens": 659734828.0, + "step": 16997 + }, + { + "epoch": 2.162320315481491, + "grad_norm": 2.840610980987549, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8937201499938965, + "num_tokens": 659775615.0, + "step": 16998 + }, + { + "epoch": 2.1624475257600815, + "grad_norm": 2.575732469558716, + "learning_rate": 1e-06, + "loss": 0.2194, + "mean_token_accuracy": 0.9199573993682861, + "num_tokens": 659813280.0, + "step": 16999 + }, + { + "epoch": 2.162574736038672, + "grad_norm": 2.9320127964019775, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9031655788421631, + "num_tokens": 659850469.0, + "step": 17000 + }, + { + "epoch": 2.1627019463172625, + "grad_norm": 2.7047340869903564, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.9097861051559448, + "num_tokens": 659890677.0, + "step": 17001 + }, + { + "epoch": 2.162829156595853, + "grad_norm": 2.762941837310791, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9036598801612854, + "num_tokens": 659932026.0, + "step": 17002 + }, + { + "epoch": 2.1629563668744436, + "grad_norm": 2.850703239440918, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9062082767486572, + "num_tokens": 659968810.0, + "step": 17003 + }, + { + "epoch": 2.163083577153034, + "grad_norm": 3.045762062072754, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9031331539154053, + "num_tokens": 660003035.0, + "step": 17004 + }, + { + "epoch": 2.1632107874316246, + "grad_norm": 2.6981289386749268, + "learning_rate": 1e-06, + "loss": 0.2109, + "mean_token_accuracy": 0.9192667007446289, + "num_tokens": 660038582.0, + "step": 17005 + }, + { + "epoch": 2.163337997710215, + "grad_norm": 2.5259108543395996, + "learning_rate": 1e-06, + "loss": 0.237, + "mean_token_accuracy": 0.9118742942810059, + "num_tokens": 660083058.0, + "step": 17006 + }, + { + "epoch": 2.1634652079888053, + "grad_norm": 2.9576144218444824, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9105278253555298, + "num_tokens": 660116704.0, + "step": 17007 + }, + { + "epoch": 2.1635924182673962, + "grad_norm": 2.9719042778015137, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.8984696865081787, + "num_tokens": 660155259.0, + "step": 17008 + }, + { + "epoch": 2.1637196285459863, + "grad_norm": 2.8562145233154297, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9040029048919678, + "num_tokens": 660196114.0, + "step": 17009 + }, + { + "epoch": 2.163846838824577, + "grad_norm": 2.748335123062134, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9107298851013184, + "num_tokens": 660235791.0, + "step": 17010 + }, + { + "epoch": 2.1639740491031674, + "grad_norm": 2.770268201828003, + "learning_rate": 1e-06, + "loss": 0.232, + "mean_token_accuracy": 0.9120181798934937, + "num_tokens": 660273316.0, + "step": 17011 + }, + { + "epoch": 2.164101259381758, + "grad_norm": 2.7239644527435303, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9061036705970764, + "num_tokens": 660316628.0, + "step": 17012 + }, + { + "epoch": 2.1642284696603484, + "grad_norm": 3.0209927558898926, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8982653617858887, + "num_tokens": 660355418.0, + "step": 17013 + }, + { + "epoch": 2.164355679938939, + "grad_norm": 2.7362613677978516, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9139592051506042, + "num_tokens": 660393313.0, + "step": 17014 + }, + { + "epoch": 2.1644828902175295, + "grad_norm": 2.591547966003418, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9173867106437683, + "num_tokens": 660431205.0, + "step": 17015 + }, + { + "epoch": 2.16461010049612, + "grad_norm": 2.8652050495147705, + "learning_rate": 1e-06, + "loss": 0.2362, + "mean_token_accuracy": 0.9122058749198914, + "num_tokens": 660468188.0, + "step": 17016 + }, + { + "epoch": 2.1647373107747105, + "grad_norm": 2.811260938644409, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9074001908302307, + "num_tokens": 660508618.0, + "step": 17017 + }, + { + "epoch": 2.164864521053301, + "grad_norm": 2.7364554405212402, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9036655426025391, + "num_tokens": 660551251.0, + "step": 17018 + }, + { + "epoch": 2.1649917313318916, + "grad_norm": 3.308436393737793, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9026436805725098, + "num_tokens": 660581701.0, + "step": 17019 + }, + { + "epoch": 2.165118941610482, + "grad_norm": 2.826791763305664, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9127335548400879, + "num_tokens": 660621879.0, + "step": 17020 + }, + { + "epoch": 2.1652461518890727, + "grad_norm": 2.9010188579559326, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9131083488464355, + "num_tokens": 660662187.0, + "step": 17021 + }, + { + "epoch": 2.165373362167663, + "grad_norm": 2.7904720306396484, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9058430790901184, + "num_tokens": 660700905.0, + "step": 17022 + }, + { + "epoch": 2.1655005724462537, + "grad_norm": 2.974273920059204, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8950833082199097, + "num_tokens": 660744019.0, + "step": 17023 + }, + { + "epoch": 2.1656277827248442, + "grad_norm": 2.679213523864746, + "learning_rate": 1e-06, + "loss": 0.2227, + "mean_token_accuracy": 0.9159505367279053, + "num_tokens": 660782921.0, + "step": 17024 + }, + { + "epoch": 2.1657549930034348, + "grad_norm": 2.8363864421844482, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9065922498703003, + "num_tokens": 660820032.0, + "step": 17025 + }, + { + "epoch": 2.1658822032820253, + "grad_norm": 2.8785219192504883, + "learning_rate": 1e-06, + "loss": 0.2254, + "mean_token_accuracy": 0.9155917167663574, + "num_tokens": 660856133.0, + "step": 17026 + }, + { + "epoch": 2.166009413560616, + "grad_norm": 2.880152463912964, + "learning_rate": 1e-06, + "loss": 0.2364, + "mean_token_accuracy": 0.9129828214645386, + "num_tokens": 660890541.0, + "step": 17027 + }, + { + "epoch": 2.1661366238392064, + "grad_norm": 2.839708089828491, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.902781069278717, + "num_tokens": 660930600.0, + "step": 17028 + }, + { + "epoch": 2.166263834117797, + "grad_norm": 2.8453917503356934, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9106356501579285, + "num_tokens": 660971090.0, + "step": 17029 + }, + { + "epoch": 2.1663910443963874, + "grad_norm": 3.254934310913086, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.8969987630844116, + "num_tokens": 661004695.0, + "step": 17030 + }, + { + "epoch": 2.166518254674978, + "grad_norm": 2.6238808631896973, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.9122182726860046, + "num_tokens": 661050226.0, + "step": 17031 + }, + { + "epoch": 2.166645464953568, + "grad_norm": 2.8966591358184814, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9008322358131409, + "num_tokens": 661090273.0, + "step": 17032 + }, + { + "epoch": 2.1667726752321586, + "grad_norm": 2.927539348602295, + "learning_rate": 1e-06, + "loss": 0.2138, + "mean_token_accuracy": 0.9192783832550049, + "num_tokens": 661123017.0, + "step": 17033 + }, + { + "epoch": 2.166899885510749, + "grad_norm": 2.8013246059417725, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9060632586479187, + "num_tokens": 661162748.0, + "step": 17034 + }, + { + "epoch": 2.1670270957893396, + "grad_norm": 2.8075265884399414, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9063407182693481, + "num_tokens": 661202657.0, + "step": 17035 + }, + { + "epoch": 2.16715430606793, + "grad_norm": 2.778061628341675, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9058374762535095, + "num_tokens": 661242347.0, + "step": 17036 + }, + { + "epoch": 2.1672815163465207, + "grad_norm": 3.1041860580444336, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9053114056587219, + "num_tokens": 661279587.0, + "step": 17037 + }, + { + "epoch": 2.167408726625111, + "grad_norm": 2.8235275745391846, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9059399962425232, + "num_tokens": 661315875.0, + "step": 17038 + }, + { + "epoch": 2.1675359369037017, + "grad_norm": 2.744870662689209, + "learning_rate": 1e-06, + "loss": 0.2338, + "mean_token_accuracy": 0.9124712944030762, + "num_tokens": 661355987.0, + "step": 17039 + }, + { + "epoch": 2.1676631471822922, + "grad_norm": 2.752732992172241, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.908573567867279, + "num_tokens": 661391665.0, + "step": 17040 + }, + { + "epoch": 2.1677903574608828, + "grad_norm": 2.9027323722839355, + "learning_rate": 1e-06, + "loss": 0.2376, + "mean_token_accuracy": 0.9107770919799805, + "num_tokens": 661424477.0, + "step": 17041 + }, + { + "epoch": 2.1679175677394733, + "grad_norm": 2.642949342727661, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9124070405960083, + "num_tokens": 661467732.0, + "step": 17042 + }, + { + "epoch": 2.168044778018064, + "grad_norm": 2.7764430046081543, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9138393402099609, + "num_tokens": 661510139.0, + "step": 17043 + }, + { + "epoch": 2.1681719882966544, + "grad_norm": 2.7580223083496094, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9054889678955078, + "num_tokens": 661551803.0, + "step": 17044 + }, + { + "epoch": 2.168299198575245, + "grad_norm": 2.9245312213897705, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9092068672180176, + "num_tokens": 661588925.0, + "step": 17045 + }, + { + "epoch": 2.1684264088538354, + "grad_norm": 2.8374693393707275, + "learning_rate": 1e-06, + "loss": 0.2296, + "mean_token_accuracy": 0.9138025045394897, + "num_tokens": 661623414.0, + "step": 17046 + }, + { + "epoch": 2.168553619132426, + "grad_norm": 2.6499531269073486, + "learning_rate": 1e-06, + "loss": 0.2334, + "mean_token_accuracy": 0.9124985337257385, + "num_tokens": 661661903.0, + "step": 17047 + }, + { + "epoch": 2.1686808294110165, + "grad_norm": 2.8280529975891113, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9021233320236206, + "num_tokens": 661703620.0, + "step": 17048 + }, + { + "epoch": 2.168808039689607, + "grad_norm": 3.01157546043396, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9027035236358643, + "num_tokens": 661741403.0, + "step": 17049 + }, + { + "epoch": 2.1689352499681975, + "grad_norm": 2.9637863636016846, + "learning_rate": 1e-06, + "loss": 0.2333, + "mean_token_accuracy": 0.9127002358436584, + "num_tokens": 661774732.0, + "step": 17050 + }, + { + "epoch": 2.169062460246788, + "grad_norm": 2.786893844604492, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9096362590789795, + "num_tokens": 661811931.0, + "step": 17051 + }, + { + "epoch": 2.1691896705253786, + "grad_norm": 2.946366310119629, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9118572473526001, + "num_tokens": 661846898.0, + "step": 17052 + }, + { + "epoch": 2.169316880803969, + "grad_norm": 2.8260316848754883, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.901607871055603, + "num_tokens": 661889976.0, + "step": 17053 + }, + { + "epoch": 2.1694440910825596, + "grad_norm": 2.639946937561035, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9148117303848267, + "num_tokens": 661929784.0, + "step": 17054 + }, + { + "epoch": 2.16957130136115, + "grad_norm": 2.774409532546997, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9050925970077515, + "num_tokens": 661972429.0, + "step": 17055 + }, + { + "epoch": 2.1696985116397407, + "grad_norm": 3.0306859016418457, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.905184268951416, + "num_tokens": 662004890.0, + "step": 17056 + }, + { + "epoch": 2.169825721918331, + "grad_norm": 2.8737404346466064, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9046419262886047, + "num_tokens": 662043701.0, + "step": 17057 + }, + { + "epoch": 2.1699529321969213, + "grad_norm": 3.0360255241394043, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9031872749328613, + "num_tokens": 662079699.0, + "step": 17058 + }, + { + "epoch": 2.170080142475512, + "grad_norm": 2.8828468322753906, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9032938480377197, + "num_tokens": 662116522.0, + "step": 17059 + }, + { + "epoch": 2.1702073527541024, + "grad_norm": 2.687396287918091, + "learning_rate": 1e-06, + "loss": 0.2053, + "mean_token_accuracy": 0.923284649848938, + "num_tokens": 662151752.0, + "step": 17060 + }, + { + "epoch": 2.170334563032693, + "grad_norm": 2.8842220306396484, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9009029269218445, + "num_tokens": 662191082.0, + "step": 17061 + }, + { + "epoch": 2.1704617733112834, + "grad_norm": 2.5785071849823, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9088670015335083, + "num_tokens": 662236668.0, + "step": 17062 + }, + { + "epoch": 2.170588983589874, + "grad_norm": 3.1041011810302734, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9071870446205139, + "num_tokens": 662269626.0, + "step": 17063 + }, + { + "epoch": 2.1707161938684645, + "grad_norm": 2.605639696121216, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.8990076184272766, + "num_tokens": 662321061.0, + "step": 17064 + }, + { + "epoch": 2.170843404147055, + "grad_norm": 3.0809314250946045, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.903678834438324, + "num_tokens": 662356917.0, + "step": 17065 + }, + { + "epoch": 2.1709706144256455, + "grad_norm": 2.854250907897949, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.909294605255127, + "num_tokens": 662395580.0, + "step": 17066 + }, + { + "epoch": 2.171097824704236, + "grad_norm": 2.7957260608673096, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9088664054870605, + "num_tokens": 662438902.0, + "step": 17067 + }, + { + "epoch": 2.1712250349828266, + "grad_norm": 2.7762956619262695, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9010698795318604, + "num_tokens": 662482655.0, + "step": 17068 + }, + { + "epoch": 2.171352245261417, + "grad_norm": 2.8719594478607178, + "learning_rate": 1e-06, + "loss": 0.2266, + "mean_token_accuracy": 0.9181787371635437, + "num_tokens": 662515608.0, + "step": 17069 + }, + { + "epoch": 2.1714794555400077, + "grad_norm": 2.946378707885742, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8996808528900146, + "num_tokens": 662554929.0, + "step": 17070 + }, + { + "epoch": 2.171606665818598, + "grad_norm": 2.7134132385253906, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8905466794967651, + "num_tokens": 662606993.0, + "step": 17071 + }, + { + "epoch": 2.1717338760971887, + "grad_norm": 2.8849270343780518, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9073745012283325, + "num_tokens": 662647075.0, + "step": 17072 + }, + { + "epoch": 2.1718610863757792, + "grad_norm": 2.6493301391601562, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9077339172363281, + "num_tokens": 662686901.0, + "step": 17073 + }, + { + "epoch": 2.1719882966543698, + "grad_norm": 2.672065496444702, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9155462384223938, + "num_tokens": 662727804.0, + "step": 17074 + }, + { + "epoch": 2.1721155069329603, + "grad_norm": 2.7897562980651855, + "learning_rate": 1e-06, + "loss": 0.2413, + "mean_token_accuracy": 0.9118199348449707, + "num_tokens": 662766863.0, + "step": 17075 + }, + { + "epoch": 2.172242717211551, + "grad_norm": 2.9332096576690674, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9031728506088257, + "num_tokens": 662806743.0, + "step": 17076 + }, + { + "epoch": 2.1723699274901414, + "grad_norm": 3.0278196334838867, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9077575206756592, + "num_tokens": 662840492.0, + "step": 17077 + }, + { + "epoch": 2.172497137768732, + "grad_norm": 2.5471599102020264, + "learning_rate": 1e-06, + "loss": 0.2266, + "mean_token_accuracy": 0.914276123046875, + "num_tokens": 662882840.0, + "step": 17078 + }, + { + "epoch": 2.1726243480473224, + "grad_norm": 2.951781749725342, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8935592174530029, + "num_tokens": 662925189.0, + "step": 17079 + }, + { + "epoch": 2.172751558325913, + "grad_norm": 3.0790953636169434, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8972420692443848, + "num_tokens": 662962465.0, + "step": 17080 + }, + { + "epoch": 2.1728787686045035, + "grad_norm": 2.8790440559387207, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.890231728553772, + "num_tokens": 663006976.0, + "step": 17081 + }, + { + "epoch": 2.1730059788830935, + "grad_norm": 2.925999879837036, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9034641981124878, + "num_tokens": 663046279.0, + "step": 17082 + }, + { + "epoch": 2.173133189161684, + "grad_norm": 2.8263514041900635, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9011722803115845, + "num_tokens": 663089836.0, + "step": 17083 + }, + { + "epoch": 2.1732603994402746, + "grad_norm": 2.7242510318756104, + "learning_rate": 1e-06, + "loss": 0.237, + "mean_token_accuracy": 0.9128029346466064, + "num_tokens": 663127511.0, + "step": 17084 + }, + { + "epoch": 2.173387609718865, + "grad_norm": 2.780909299850464, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9035894870758057, + "num_tokens": 663168238.0, + "step": 17085 + }, + { + "epoch": 2.1735148199974557, + "grad_norm": 2.786430597305298, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.901253879070282, + "num_tokens": 663210972.0, + "step": 17086 + }, + { + "epoch": 2.173642030276046, + "grad_norm": 2.7472939491271973, + "learning_rate": 1e-06, + "loss": 0.2376, + "mean_token_accuracy": 0.9125244617462158, + "num_tokens": 663247397.0, + "step": 17087 + }, + { + "epoch": 2.1737692405546367, + "grad_norm": 3.160339832305908, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9031906127929688, + "num_tokens": 663283042.0, + "step": 17088 + }, + { + "epoch": 2.1738964508332272, + "grad_norm": 2.8654160499572754, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9042922258377075, + "num_tokens": 663322293.0, + "step": 17089 + }, + { + "epoch": 2.1740236611118178, + "grad_norm": 2.8779022693634033, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9085363149642944, + "num_tokens": 663358405.0, + "step": 17090 + }, + { + "epoch": 2.1741508713904083, + "grad_norm": 2.9014229774475098, + "learning_rate": 1e-06, + "loss": 0.3117, + "mean_token_accuracy": 0.8845157623291016, + "num_tokens": 663403035.0, + "step": 17091 + }, + { + "epoch": 2.174278081668999, + "grad_norm": 3.124601125717163, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.898903489112854, + "num_tokens": 663445838.0, + "step": 17092 + }, + { + "epoch": 2.1744052919475894, + "grad_norm": 2.892277479171753, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9041632413864136, + "num_tokens": 663485581.0, + "step": 17093 + }, + { + "epoch": 2.17453250222618, + "grad_norm": 2.8088953495025635, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9074674248695374, + "num_tokens": 663524064.0, + "step": 17094 + }, + { + "epoch": 2.1746597125047704, + "grad_norm": 2.9809186458587646, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9098515510559082, + "num_tokens": 663561146.0, + "step": 17095 + }, + { + "epoch": 2.174786922783361, + "grad_norm": 2.8776044845581055, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8938347697257996, + "num_tokens": 663602614.0, + "step": 17096 + }, + { + "epoch": 2.1749141330619515, + "grad_norm": 2.7727506160736084, + "learning_rate": 1e-06, + "loss": 0.2309, + "mean_token_accuracy": 0.9130287170410156, + "num_tokens": 663637347.0, + "step": 17097 + }, + { + "epoch": 2.175041343340542, + "grad_norm": 2.6723647117614746, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9062601327896118, + "num_tokens": 663677242.0, + "step": 17098 + }, + { + "epoch": 2.1751685536191325, + "grad_norm": 3.180718421936035, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.8996263742446899, + "num_tokens": 663711107.0, + "step": 17099 + }, + { + "epoch": 2.175295763897723, + "grad_norm": 3.0764684677124023, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.901390790939331, + "num_tokens": 663747100.0, + "step": 17100 + }, + { + "epoch": 2.1754229741763136, + "grad_norm": 2.8151211738586426, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8996009826660156, + "num_tokens": 663791188.0, + "step": 17101 + }, + { + "epoch": 2.175550184454904, + "grad_norm": 3.3498942852020264, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9041395783424377, + "num_tokens": 663820831.0, + "step": 17102 + }, + { + "epoch": 2.1756773947334946, + "grad_norm": 2.8320446014404297, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.8995956182479858, + "num_tokens": 663859780.0, + "step": 17103 + }, + { + "epoch": 2.175804605012085, + "grad_norm": 2.6960232257843018, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9094271063804626, + "num_tokens": 663897581.0, + "step": 17104 + }, + { + "epoch": 2.1759318152906753, + "grad_norm": 2.7376744747161865, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.902549147605896, + "num_tokens": 663941067.0, + "step": 17105 + }, + { + "epoch": 2.1760590255692662, + "grad_norm": 3.0848872661590576, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8913857936859131, + "num_tokens": 663982089.0, + "step": 17106 + }, + { + "epoch": 2.1761862358478563, + "grad_norm": 3.0636558532714844, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8980712890625, + "num_tokens": 664018707.0, + "step": 17107 + }, + { + "epoch": 2.176313446126447, + "grad_norm": 2.6630666255950928, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9082872867584229, + "num_tokens": 664061242.0, + "step": 17108 + }, + { + "epoch": 2.1764406564050374, + "grad_norm": 2.582322835922241, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.9094099998474121, + "num_tokens": 664100660.0, + "step": 17109 + }, + { + "epoch": 2.176567866683628, + "grad_norm": 2.9626169204711914, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8933194875717163, + "num_tokens": 664141896.0, + "step": 17110 + }, + { + "epoch": 2.1766950769622184, + "grad_norm": 2.83664870262146, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.904320240020752, + "num_tokens": 664181212.0, + "step": 17111 + }, + { + "epoch": 2.176822287240809, + "grad_norm": 2.779776096343994, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8978391289710999, + "num_tokens": 664227530.0, + "step": 17112 + }, + { + "epoch": 2.1769494975193995, + "grad_norm": 2.886718988418579, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9071604013442993, + "num_tokens": 664264235.0, + "step": 17113 + }, + { + "epoch": 2.17707670779799, + "grad_norm": 3.0647144317626953, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9003002643585205, + "num_tokens": 664303092.0, + "step": 17114 + }, + { + "epoch": 2.1772039180765805, + "grad_norm": 2.9151971340179443, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9111011624336243, + "num_tokens": 664337859.0, + "step": 17115 + }, + { + "epoch": 2.177331128355171, + "grad_norm": 3.1993701457977295, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8963271975517273, + "num_tokens": 664374060.0, + "step": 17116 + }, + { + "epoch": 2.1774583386337616, + "grad_norm": 3.212968111038208, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9116232991218567, + "num_tokens": 664407065.0, + "step": 17117 + }, + { + "epoch": 2.177585548912352, + "grad_norm": 2.9349143505096436, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9097808003425598, + "num_tokens": 664446083.0, + "step": 17118 + }, + { + "epoch": 2.1777127591909426, + "grad_norm": 2.788902997970581, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9037583470344543, + "num_tokens": 664491821.0, + "step": 17119 + }, + { + "epoch": 2.177839969469533, + "grad_norm": 2.9912796020507812, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9094972610473633, + "num_tokens": 664527941.0, + "step": 17120 + }, + { + "epoch": 2.1779671797481237, + "grad_norm": 2.787600517272949, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9100402593612671, + "num_tokens": 664562961.0, + "step": 17121 + }, + { + "epoch": 2.1780943900267142, + "grad_norm": 3.309112787246704, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8924093246459961, + "num_tokens": 664596294.0, + "step": 17122 + }, + { + "epoch": 2.1782216003053048, + "grad_norm": 2.8111135959625244, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9018661975860596, + "num_tokens": 664637820.0, + "step": 17123 + }, + { + "epoch": 2.1783488105838953, + "grad_norm": 2.9352593421936035, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9006893634796143, + "num_tokens": 664676123.0, + "step": 17124 + }, + { + "epoch": 2.178476020862486, + "grad_norm": 2.8147473335266113, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9142174124717712, + "num_tokens": 664717274.0, + "step": 17125 + }, + { + "epoch": 2.1786032311410763, + "grad_norm": 2.762810230255127, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9034975171089172, + "num_tokens": 664758949.0, + "step": 17126 + }, + { + "epoch": 2.178730441419667, + "grad_norm": 2.885850191116333, + "learning_rate": 1e-06, + "loss": 0.2315, + "mean_token_accuracy": 0.9134901165962219, + "num_tokens": 664795161.0, + "step": 17127 + }, + { + "epoch": 2.1788576516982574, + "grad_norm": 2.6887929439544678, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9130880236625671, + "num_tokens": 664834421.0, + "step": 17128 + }, + { + "epoch": 2.178984861976848, + "grad_norm": 2.7524046897888184, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9110485911369324, + "num_tokens": 664874226.0, + "step": 17129 + }, + { + "epoch": 2.179112072255438, + "grad_norm": 2.8856654167175293, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.8999706506729126, + "num_tokens": 664916794.0, + "step": 17130 + }, + { + "epoch": 2.1792392825340285, + "grad_norm": 2.7479212284088135, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9085921049118042, + "num_tokens": 664954310.0, + "step": 17131 + }, + { + "epoch": 2.179366492812619, + "grad_norm": 2.641819715499878, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9119850397109985, + "num_tokens": 664996750.0, + "step": 17132 + }, + { + "epoch": 2.1794937030912096, + "grad_norm": 2.645623207092285, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9105643033981323, + "num_tokens": 665036386.0, + "step": 17133 + }, + { + "epoch": 2.1796209133698, + "grad_norm": 2.845869302749634, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.903335690498352, + "num_tokens": 665080386.0, + "step": 17134 + }, + { + "epoch": 2.1797481236483907, + "grad_norm": 2.8983874320983887, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8897300958633423, + "num_tokens": 665120684.0, + "step": 17135 + }, + { + "epoch": 2.179875333926981, + "grad_norm": 2.7424259185791016, + "learning_rate": 1e-06, + "loss": 0.2323, + "mean_token_accuracy": 0.9129168391227722, + "num_tokens": 665159670.0, + "step": 17136 + }, + { + "epoch": 2.1800025442055717, + "grad_norm": 3.0467822551727295, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8878326416015625, + "num_tokens": 665201380.0, + "step": 17137 + }, + { + "epoch": 2.1801297544841622, + "grad_norm": 3.0527448654174805, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9108351469039917, + "num_tokens": 665235328.0, + "step": 17138 + }, + { + "epoch": 2.1802569647627528, + "grad_norm": 3.268265724182129, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.8983846306800842, + "num_tokens": 665268102.0, + "step": 17139 + }, + { + "epoch": 2.1803841750413433, + "grad_norm": 2.5457963943481445, + "learning_rate": 1e-06, + "loss": 0.2094, + "mean_token_accuracy": 0.9221029877662659, + "num_tokens": 665307218.0, + "step": 17140 + }, + { + "epoch": 2.180511385319934, + "grad_norm": 2.8856101036071777, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9019680023193359, + "num_tokens": 665348814.0, + "step": 17141 + }, + { + "epoch": 2.1806385955985244, + "grad_norm": 3.014364719390869, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9033939838409424, + "num_tokens": 665384611.0, + "step": 17142 + }, + { + "epoch": 2.180765805877115, + "grad_norm": 2.8519585132598877, + "learning_rate": 1e-06, + "loss": 0.2313, + "mean_token_accuracy": 0.9116295576095581, + "num_tokens": 665420616.0, + "step": 17143 + }, + { + "epoch": 2.1808930161557054, + "grad_norm": 2.95516300201416, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.8984678983688354, + "num_tokens": 665460686.0, + "step": 17144 + }, + { + "epoch": 2.181020226434296, + "grad_norm": 2.9047772884368896, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9096581339836121, + "num_tokens": 665492573.0, + "step": 17145 + }, + { + "epoch": 2.1811474367128865, + "grad_norm": 2.7873573303222656, + "learning_rate": 1e-06, + "loss": 0.2126, + "mean_token_accuracy": 0.9191614389419556, + "num_tokens": 665527313.0, + "step": 17146 + }, + { + "epoch": 2.181274646991477, + "grad_norm": 2.63342547416687, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9024445414543152, + "num_tokens": 665570749.0, + "step": 17147 + }, + { + "epoch": 2.1814018572700675, + "grad_norm": 2.987783432006836, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9042611122131348, + "num_tokens": 665605109.0, + "step": 17148 + }, + { + "epoch": 2.181529067548658, + "grad_norm": 2.896061658859253, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8961696624755859, + "num_tokens": 665646806.0, + "step": 17149 + }, + { + "epoch": 2.1816562778272486, + "grad_norm": 2.742554187774658, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.901364266872406, + "num_tokens": 665690029.0, + "step": 17150 + }, + { + "epoch": 2.181783488105839, + "grad_norm": 2.5979700088500977, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8952687382698059, + "num_tokens": 665736695.0, + "step": 17151 + }, + { + "epoch": 2.1819106983844296, + "grad_norm": 2.8575944900512695, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.905785322189331, + "num_tokens": 665775342.0, + "step": 17152 + }, + { + "epoch": 2.18203790866302, + "grad_norm": 2.6233670711517334, + "learning_rate": 1e-06, + "loss": 0.223, + "mean_token_accuracy": 0.9167575836181641, + "num_tokens": 665814541.0, + "step": 17153 + }, + { + "epoch": 2.1821651189416107, + "grad_norm": 2.887653112411499, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9124953746795654, + "num_tokens": 665849563.0, + "step": 17154 + }, + { + "epoch": 2.1822923292202008, + "grad_norm": 2.7645134925842285, + "learning_rate": 1e-06, + "loss": 0.2209, + "mean_token_accuracy": 0.918875515460968, + "num_tokens": 665885117.0, + "step": 17155 + }, + { + "epoch": 2.1824195394987913, + "grad_norm": 2.7823679447174072, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9016447067260742, + "num_tokens": 665926119.0, + "step": 17156 + }, + { + "epoch": 2.182546749777382, + "grad_norm": 2.887746572494507, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9046890735626221, + "num_tokens": 665964256.0, + "step": 17157 + }, + { + "epoch": 2.1826739600559724, + "grad_norm": 2.955411195755005, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.904420018196106, + "num_tokens": 666002516.0, + "step": 17158 + }, + { + "epoch": 2.182801170334563, + "grad_norm": 2.867168426513672, + "learning_rate": 1e-06, + "loss": 0.2299, + "mean_token_accuracy": 0.9151110649108887, + "num_tokens": 666038943.0, + "step": 17159 + }, + { + "epoch": 2.1829283806131534, + "grad_norm": 2.883887767791748, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9058809876441956, + "num_tokens": 666081026.0, + "step": 17160 + }, + { + "epoch": 2.183055590891744, + "grad_norm": 2.567396402359009, + "learning_rate": 1e-06, + "loss": 0.2284, + "mean_token_accuracy": 0.9166648387908936, + "num_tokens": 666121911.0, + "step": 17161 + }, + { + "epoch": 2.1831828011703345, + "grad_norm": 2.98305344581604, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8959925174713135, + "num_tokens": 666163084.0, + "step": 17162 + }, + { + "epoch": 2.183310011448925, + "grad_norm": 3.3737010955810547, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.899292528629303, + "num_tokens": 666199753.0, + "step": 17163 + }, + { + "epoch": 2.1834372217275155, + "grad_norm": 2.798363208770752, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.9146731495857239, + "num_tokens": 666237255.0, + "step": 17164 + }, + { + "epoch": 2.183564432006106, + "grad_norm": 2.613699436187744, + "learning_rate": 1e-06, + "loss": 0.2378, + "mean_token_accuracy": 0.9105944037437439, + "num_tokens": 666282662.0, + "step": 17165 + }, + { + "epoch": 2.1836916422846966, + "grad_norm": 2.8505001068115234, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9025901556015015, + "num_tokens": 666323871.0, + "step": 17166 + }, + { + "epoch": 2.183818852563287, + "grad_norm": 2.7889482975006104, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9063577651977539, + "num_tokens": 666367061.0, + "step": 17167 + }, + { + "epoch": 2.1839460628418776, + "grad_norm": 2.73844051361084, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9066774845123291, + "num_tokens": 666410887.0, + "step": 17168 + }, + { + "epoch": 2.184073273120468, + "grad_norm": 2.5604407787323, + "learning_rate": 1e-06, + "loss": 0.2285, + "mean_token_accuracy": 0.916765570640564, + "num_tokens": 666454804.0, + "step": 17169 + }, + { + "epoch": 2.1842004833990587, + "grad_norm": 3.060272693634033, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8995197415351868, + "num_tokens": 666493135.0, + "step": 17170 + }, + { + "epoch": 2.1843276936776492, + "grad_norm": 2.7326884269714355, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9073522090911865, + "num_tokens": 666532404.0, + "step": 17171 + }, + { + "epoch": 2.1844549039562398, + "grad_norm": 2.849283218383789, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9041573405265808, + "num_tokens": 666571127.0, + "step": 17172 + }, + { + "epoch": 2.1845821142348303, + "grad_norm": 2.872387170791626, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9051624536514282, + "num_tokens": 666610961.0, + "step": 17173 + }, + { + "epoch": 2.184709324513421, + "grad_norm": 2.9590961933135986, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9046036005020142, + "num_tokens": 666646230.0, + "step": 17174 + }, + { + "epoch": 2.1848365347920113, + "grad_norm": 3.1150362491607666, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9046890139579773, + "num_tokens": 666679258.0, + "step": 17175 + }, + { + "epoch": 2.184963745070602, + "grad_norm": 3.0670554637908936, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8977125883102417, + "num_tokens": 666720420.0, + "step": 17176 + }, + { + "epoch": 2.1850909553491924, + "grad_norm": 2.8571667671203613, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9084089994430542, + "num_tokens": 666759314.0, + "step": 17177 + }, + { + "epoch": 2.185218165627783, + "grad_norm": 2.7952823638916016, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9058090448379517, + "num_tokens": 666800654.0, + "step": 17178 + }, + { + "epoch": 2.1853453759063735, + "grad_norm": 3.0637481212615967, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.8998813629150391, + "num_tokens": 666838841.0, + "step": 17179 + }, + { + "epoch": 2.1854725861849635, + "grad_norm": 2.730170726776123, + "learning_rate": 1e-06, + "loss": 0.2276, + "mean_token_accuracy": 0.9145219326019287, + "num_tokens": 666878329.0, + "step": 17180 + }, + { + "epoch": 2.185599796463554, + "grad_norm": 2.906773090362549, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9084334373474121, + "num_tokens": 666914083.0, + "step": 17181 + }, + { + "epoch": 2.1857270067421446, + "grad_norm": 3.0051772594451904, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9035896062850952, + "num_tokens": 666949714.0, + "step": 17182 + }, + { + "epoch": 2.185854217020735, + "grad_norm": 2.6703217029571533, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9110771417617798, + "num_tokens": 666996433.0, + "step": 17183 + }, + { + "epoch": 2.1859814272993257, + "grad_norm": 2.6921534538269043, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9092225432395935, + "num_tokens": 667037535.0, + "step": 17184 + }, + { + "epoch": 2.186108637577916, + "grad_norm": 3.0237526893615723, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8945582509040833, + "num_tokens": 667076982.0, + "step": 17185 + }, + { + "epoch": 2.1862358478565067, + "grad_norm": 3.0829248428344727, + "learning_rate": 1e-06, + "loss": 0.2292, + "mean_token_accuracy": 0.914741039276123, + "num_tokens": 667110674.0, + "step": 17186 + }, + { + "epoch": 2.1863630581350972, + "grad_norm": 2.864077091217041, + "learning_rate": 1e-06, + "loss": 0.228, + "mean_token_accuracy": 0.9129728674888611, + "num_tokens": 667149512.0, + "step": 17187 + }, + { + "epoch": 2.1864902684136878, + "grad_norm": 3.0224101543426514, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9029790759086609, + "num_tokens": 667186511.0, + "step": 17188 + }, + { + "epoch": 2.1866174786922783, + "grad_norm": 2.7969820499420166, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9031321406364441, + "num_tokens": 667228457.0, + "step": 17189 + }, + { + "epoch": 2.186744688970869, + "grad_norm": 2.746022939682007, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9028943777084351, + "num_tokens": 667271397.0, + "step": 17190 + }, + { + "epoch": 2.1868718992494594, + "grad_norm": 2.9558188915252686, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.902400553226471, + "num_tokens": 667309773.0, + "step": 17191 + }, + { + "epoch": 2.18699910952805, + "grad_norm": 2.78635573387146, + "learning_rate": 1e-06, + "loss": 0.2285, + "mean_token_accuracy": 0.9153263568878174, + "num_tokens": 667347099.0, + "step": 17192 + }, + { + "epoch": 2.1871263198066404, + "grad_norm": 3.03747820854187, + "learning_rate": 1e-06, + "loss": 0.2964, + "mean_token_accuracy": 0.8931518793106079, + "num_tokens": 667386122.0, + "step": 17193 + }, + { + "epoch": 2.187253530085231, + "grad_norm": 3.1260132789611816, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9015049338340759, + "num_tokens": 667425105.0, + "step": 17194 + }, + { + "epoch": 2.1873807403638215, + "grad_norm": 2.667292356491089, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9077447652816772, + "num_tokens": 667468475.0, + "step": 17195 + }, + { + "epoch": 2.187507950642412, + "grad_norm": 2.8795969486236572, + "learning_rate": 1e-06, + "loss": 0.2265, + "mean_token_accuracy": 0.9133491516113281, + "num_tokens": 667502134.0, + "step": 17196 + }, + { + "epoch": 2.1876351609210025, + "grad_norm": 3.076352834701538, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9055080413818359, + "num_tokens": 667540615.0, + "step": 17197 + }, + { + "epoch": 2.187762371199593, + "grad_norm": 2.8084537982940674, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9105366468429565, + "num_tokens": 667579909.0, + "step": 17198 + }, + { + "epoch": 2.1878895814781836, + "grad_norm": 3.04240083694458, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9052025079727173, + "num_tokens": 667615422.0, + "step": 17199 + }, + { + "epoch": 2.188016791756774, + "grad_norm": 2.89554500579834, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9037345051765442, + "num_tokens": 667653096.0, + "step": 17200 + }, + { + "epoch": 2.1881440020353646, + "grad_norm": 2.8185460567474365, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9060738682746887, + "num_tokens": 667693077.0, + "step": 17201 + }, + { + "epoch": 2.188271212313955, + "grad_norm": 2.6530275344848633, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9117007255554199, + "num_tokens": 667733174.0, + "step": 17202 + }, + { + "epoch": 2.1883984225925452, + "grad_norm": 2.8296923637390137, + "learning_rate": 1e-06, + "loss": 0.2363, + "mean_token_accuracy": 0.9104620218276978, + "num_tokens": 667767444.0, + "step": 17203 + }, + { + "epoch": 2.188525632871136, + "grad_norm": 2.785747528076172, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9149882793426514, + "num_tokens": 667803462.0, + "step": 17204 + }, + { + "epoch": 2.1886528431497263, + "grad_norm": 2.568718671798706, + "learning_rate": 1e-06, + "loss": 0.2168, + "mean_token_accuracy": 0.9201112985610962, + "num_tokens": 667844590.0, + "step": 17205 + }, + { + "epoch": 2.188780053428317, + "grad_norm": 2.881021499633789, + "learning_rate": 1e-06, + "loss": 0.3009, + "mean_token_accuracy": 0.8911300897598267, + "num_tokens": 667887159.0, + "step": 17206 + }, + { + "epoch": 2.1889072637069074, + "grad_norm": 2.7414512634277344, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.900982141494751, + "num_tokens": 667933875.0, + "step": 17207 + }, + { + "epoch": 2.189034473985498, + "grad_norm": 2.7772419452667236, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.8998352885246277, + "num_tokens": 667973533.0, + "step": 17208 + }, + { + "epoch": 2.1891616842640884, + "grad_norm": 2.8495121002197266, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.8992519378662109, + "num_tokens": 668016383.0, + "step": 17209 + }, + { + "epoch": 2.189288894542679, + "grad_norm": 2.5944557189941406, + "learning_rate": 1e-06, + "loss": 0.214, + "mean_token_accuracy": 0.9196081161499023, + "num_tokens": 668056628.0, + "step": 17210 + }, + { + "epoch": 2.1894161048212695, + "grad_norm": 2.9102466106414795, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.895397961139679, + "num_tokens": 668096040.0, + "step": 17211 + }, + { + "epoch": 2.18954331509986, + "grad_norm": 2.9246904850006104, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9007346630096436, + "num_tokens": 668134682.0, + "step": 17212 + }, + { + "epoch": 2.1896705253784505, + "grad_norm": 3.741253137588501, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9059549570083618, + "num_tokens": 668163842.0, + "step": 17213 + }, + { + "epoch": 2.189797735657041, + "grad_norm": 2.7065231800079346, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9003774523735046, + "num_tokens": 668206926.0, + "step": 17214 + }, + { + "epoch": 2.1899249459356316, + "grad_norm": 2.669214963912964, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9041775465011597, + "num_tokens": 668251598.0, + "step": 17215 + }, + { + "epoch": 2.190052156214222, + "grad_norm": 2.9977059364318848, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9003410339355469, + "num_tokens": 668289519.0, + "step": 17216 + }, + { + "epoch": 2.1901793664928126, + "grad_norm": 2.812122106552124, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.8989508748054504, + "num_tokens": 668332555.0, + "step": 17217 + }, + { + "epoch": 2.190306576771403, + "grad_norm": 3.004619598388672, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.8991682529449463, + "num_tokens": 668367882.0, + "step": 17218 + }, + { + "epoch": 2.1904337870499937, + "grad_norm": 2.4883837699890137, + "learning_rate": 1e-06, + "loss": 0.2131, + "mean_token_accuracy": 0.9182631969451904, + "num_tokens": 668409602.0, + "step": 17219 + }, + { + "epoch": 2.1905609973285842, + "grad_norm": 3.1664295196533203, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9053785800933838, + "num_tokens": 668443455.0, + "step": 17220 + }, + { + "epoch": 2.1906882076071748, + "grad_norm": 2.6666200160980225, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9122406840324402, + "num_tokens": 668481841.0, + "step": 17221 + }, + { + "epoch": 2.1908154178857653, + "grad_norm": 3.325835704803467, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9012219905853271, + "num_tokens": 668512766.0, + "step": 17222 + }, + { + "epoch": 2.190942628164356, + "grad_norm": 2.8095009326934814, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9091982841491699, + "num_tokens": 668555144.0, + "step": 17223 + }, + { + "epoch": 2.1910698384429463, + "grad_norm": 2.9656078815460205, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9096837043762207, + "num_tokens": 668589710.0, + "step": 17224 + }, + { + "epoch": 2.191197048721537, + "grad_norm": 2.7838797569274902, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9058810472488403, + "num_tokens": 668628794.0, + "step": 17225 + }, + { + "epoch": 2.1913242590001274, + "grad_norm": 2.826482057571411, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9076659679412842, + "num_tokens": 668668645.0, + "step": 17226 + }, + { + "epoch": 2.191451469278718, + "grad_norm": 3.0373477935791016, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8929551243782043, + "num_tokens": 668705030.0, + "step": 17227 + }, + { + "epoch": 2.191578679557308, + "grad_norm": 2.890801429748535, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9003101587295532, + "num_tokens": 668743362.0, + "step": 17228 + }, + { + "epoch": 2.1917058898358985, + "grad_norm": 2.843027114868164, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.8991712331771851, + "num_tokens": 668785989.0, + "step": 17229 + }, + { + "epoch": 2.191833100114489, + "grad_norm": 2.8232693672180176, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.9136238694190979, + "num_tokens": 668821559.0, + "step": 17230 + }, + { + "epoch": 2.1919603103930796, + "grad_norm": 2.6584956645965576, + "learning_rate": 1e-06, + "loss": 0.2295, + "mean_token_accuracy": 0.9135309457778931, + "num_tokens": 668860762.0, + "step": 17231 + }, + { + "epoch": 2.19208752067167, + "grad_norm": 2.8751108646392822, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.905348539352417, + "num_tokens": 668899052.0, + "step": 17232 + }, + { + "epoch": 2.1922147309502606, + "grad_norm": 2.783118486404419, + "learning_rate": 1e-06, + "loss": 0.2416, + "mean_token_accuracy": 0.9099609851837158, + "num_tokens": 668936867.0, + "step": 17233 + }, + { + "epoch": 2.192341941228851, + "grad_norm": 2.8563437461853027, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.900704026222229, + "num_tokens": 668974456.0, + "step": 17234 + }, + { + "epoch": 2.1924691515074417, + "grad_norm": 3.162818193435669, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9067754745483398, + "num_tokens": 669007830.0, + "step": 17235 + }, + { + "epoch": 2.1925963617860322, + "grad_norm": 2.7017710208892822, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9111688733100891, + "num_tokens": 669047807.0, + "step": 17236 + }, + { + "epoch": 2.1927235720646228, + "grad_norm": 2.6224377155303955, + "learning_rate": 1e-06, + "loss": 0.2201, + "mean_token_accuracy": 0.9175959229469299, + "num_tokens": 669090650.0, + "step": 17237 + }, + { + "epoch": 2.1928507823432133, + "grad_norm": 2.9436678886413574, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8955100774765015, + "num_tokens": 669133039.0, + "step": 17238 + }, + { + "epoch": 2.192977992621804, + "grad_norm": 2.7792460918426514, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.903093695640564, + "num_tokens": 669173811.0, + "step": 17239 + }, + { + "epoch": 2.1931052029003943, + "grad_norm": 2.709304094314575, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9114507436752319, + "num_tokens": 669214093.0, + "step": 17240 + }, + { + "epoch": 2.193232413178985, + "grad_norm": 3.0326812267303467, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9127076864242554, + "num_tokens": 669246366.0, + "step": 17241 + }, + { + "epoch": 2.1933596234575754, + "grad_norm": 2.856229066848755, + "learning_rate": 1e-06, + "loss": 0.2314, + "mean_token_accuracy": 0.9137922525405884, + "num_tokens": 669283036.0, + "step": 17242 + }, + { + "epoch": 2.193486833736166, + "grad_norm": 2.6991584300994873, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9083649516105652, + "num_tokens": 669325807.0, + "step": 17243 + }, + { + "epoch": 2.1936140440147565, + "grad_norm": 3.011794090270996, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9081956148147583, + "num_tokens": 669361801.0, + "step": 17244 + }, + { + "epoch": 2.193741254293347, + "grad_norm": 2.87581729888916, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9049319624900818, + "num_tokens": 669400651.0, + "step": 17245 + }, + { + "epoch": 2.1938684645719375, + "grad_norm": 2.7858951091766357, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9076241850852966, + "num_tokens": 669440269.0, + "step": 17246 + }, + { + "epoch": 2.193995674850528, + "grad_norm": 2.88297963142395, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9060243964195251, + "num_tokens": 669478594.0, + "step": 17247 + }, + { + "epoch": 2.1941228851291186, + "grad_norm": 2.7210471630096436, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.907457709312439, + "num_tokens": 669520037.0, + "step": 17248 + }, + { + "epoch": 2.194250095407709, + "grad_norm": 3.0584723949432373, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9036473631858826, + "num_tokens": 669556744.0, + "step": 17249 + }, + { + "epoch": 2.1943773056862996, + "grad_norm": 2.8723716735839844, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8942815065383911, + "num_tokens": 669598044.0, + "step": 17250 + }, + { + "epoch": 2.19450451596489, + "grad_norm": 2.8319618701934814, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9016141295433044, + "num_tokens": 669637680.0, + "step": 17251 + }, + { + "epoch": 2.1946317262434807, + "grad_norm": 2.833674192428589, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9025731086730957, + "num_tokens": 669674342.0, + "step": 17252 + }, + { + "epoch": 2.1947589365220708, + "grad_norm": 3.0357003211975098, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9050760865211487, + "num_tokens": 669712593.0, + "step": 17253 + }, + { + "epoch": 2.1948861468006613, + "grad_norm": 3.132821559906006, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8899103403091431, + "num_tokens": 669749877.0, + "step": 17254 + }, + { + "epoch": 2.195013357079252, + "grad_norm": 2.640099287033081, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9108221530914307, + "num_tokens": 669793162.0, + "step": 17255 + }, + { + "epoch": 2.1951405673578424, + "grad_norm": 2.6735832691192627, + "learning_rate": 1e-06, + "loss": 0.2275, + "mean_token_accuracy": 0.9167247414588928, + "num_tokens": 669830641.0, + "step": 17256 + }, + { + "epoch": 2.195267777636433, + "grad_norm": 2.8933568000793457, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8925619125366211, + "num_tokens": 669873389.0, + "step": 17257 + }, + { + "epoch": 2.1953949879150234, + "grad_norm": 2.8330087661743164, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8989821076393127, + "num_tokens": 669915138.0, + "step": 17258 + }, + { + "epoch": 2.195522198193614, + "grad_norm": 3.1253819465637207, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.898385226726532, + "num_tokens": 669948357.0, + "step": 17259 + }, + { + "epoch": 2.1956494084722045, + "grad_norm": 2.6121647357940674, + "learning_rate": 1e-06, + "loss": 0.2147, + "mean_token_accuracy": 0.9203047752380371, + "num_tokens": 669985681.0, + "step": 17260 + }, + { + "epoch": 2.195776618750795, + "grad_norm": 2.7869484424591064, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.9092303514480591, + "num_tokens": 670026346.0, + "step": 17261 + }, + { + "epoch": 2.1959038290293855, + "grad_norm": 3.0716657638549805, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8953847289085388, + "num_tokens": 670066227.0, + "step": 17262 + }, + { + "epoch": 2.196031039307976, + "grad_norm": 2.9895877838134766, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9013033509254456, + "num_tokens": 670104442.0, + "step": 17263 + }, + { + "epoch": 2.1961582495865666, + "grad_norm": 2.944387674331665, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9071106314659119, + "num_tokens": 670138427.0, + "step": 17264 + }, + { + "epoch": 2.196285459865157, + "grad_norm": 2.6935713291168213, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9078050851821899, + "num_tokens": 670183090.0, + "step": 17265 + }, + { + "epoch": 2.1964126701437476, + "grad_norm": 2.970219850540161, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9090991616249084, + "num_tokens": 670219081.0, + "step": 17266 + }, + { + "epoch": 2.196539880422338, + "grad_norm": 2.8084638118743896, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9046231508255005, + "num_tokens": 670258797.0, + "step": 17267 + }, + { + "epoch": 2.1966670907009287, + "grad_norm": 2.5612902641296387, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9130316376686096, + "num_tokens": 670300161.0, + "step": 17268 + }, + { + "epoch": 2.196794300979519, + "grad_norm": 2.883949041366577, + "learning_rate": 1e-06, + "loss": 0.2394, + "mean_token_accuracy": 0.9123082160949707, + "num_tokens": 670337441.0, + "step": 17269 + }, + { + "epoch": 2.1969215112581097, + "grad_norm": 2.666635751724243, + "learning_rate": 1e-06, + "loss": 0.2261, + "mean_token_accuracy": 0.9149508476257324, + "num_tokens": 670378465.0, + "step": 17270 + }, + { + "epoch": 2.1970487215367003, + "grad_norm": 2.8999698162078857, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9088566303253174, + "num_tokens": 670415990.0, + "step": 17271 + }, + { + "epoch": 2.197175931815291, + "grad_norm": 2.875283718109131, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.8998915553092957, + "num_tokens": 670458044.0, + "step": 17272 + }, + { + "epoch": 2.1973031420938813, + "grad_norm": 2.87176775932312, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.900080144405365, + "num_tokens": 670497192.0, + "step": 17273 + }, + { + "epoch": 2.197430352372472, + "grad_norm": 2.8579421043395996, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9039487242698669, + "num_tokens": 670537171.0, + "step": 17274 + }, + { + "epoch": 2.1975575626510624, + "grad_norm": 2.6737871170043945, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.91033935546875, + "num_tokens": 670578942.0, + "step": 17275 + }, + { + "epoch": 2.197684772929653, + "grad_norm": 2.6944403648376465, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9012718200683594, + "num_tokens": 670623732.0, + "step": 17276 + }, + { + "epoch": 2.1978119832082434, + "grad_norm": 2.8159892559051514, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9016720056533813, + "num_tokens": 670666564.0, + "step": 17277 + }, + { + "epoch": 2.1979391934868335, + "grad_norm": 2.9703707695007324, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9006670713424683, + "num_tokens": 670709818.0, + "step": 17278 + }, + { + "epoch": 2.198066403765424, + "grad_norm": 2.9564433097839355, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9048452377319336, + "num_tokens": 670746701.0, + "step": 17279 + }, + { + "epoch": 2.1981936140440146, + "grad_norm": 3.1012930870056152, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9046229720115662, + "num_tokens": 670782096.0, + "step": 17280 + }, + { + "epoch": 2.198320824322605, + "grad_norm": 2.874554395675659, + "learning_rate": 1e-06, + "loss": 0.2347, + "mean_token_accuracy": 0.913054347038269, + "num_tokens": 670817826.0, + "step": 17281 + }, + { + "epoch": 2.1984480346011956, + "grad_norm": 2.7686681747436523, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9043166637420654, + "num_tokens": 670856573.0, + "step": 17282 + }, + { + "epoch": 2.198575244879786, + "grad_norm": 2.796207904815674, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.9108467102050781, + "num_tokens": 670895592.0, + "step": 17283 + }, + { + "epoch": 2.1987024551583767, + "grad_norm": 2.9317567348480225, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9066615104675293, + "num_tokens": 670935960.0, + "step": 17284 + }, + { + "epoch": 2.1988296654369672, + "grad_norm": 2.7776708602905273, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9065432548522949, + "num_tokens": 670976514.0, + "step": 17285 + }, + { + "epoch": 2.1989568757155578, + "grad_norm": 2.9809465408325195, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9114841222763062, + "num_tokens": 671011958.0, + "step": 17286 + }, + { + "epoch": 2.1990840859941483, + "grad_norm": 2.8094990253448486, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9061069488525391, + "num_tokens": 671052172.0, + "step": 17287 + }, + { + "epoch": 2.199211296272739, + "grad_norm": 2.976043939590454, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.8958253860473633, + "num_tokens": 671090888.0, + "step": 17288 + }, + { + "epoch": 2.1993385065513293, + "grad_norm": 2.644253730773926, + "learning_rate": 1e-06, + "loss": 0.219, + "mean_token_accuracy": 0.9157868027687073, + "num_tokens": 671128927.0, + "step": 17289 + }, + { + "epoch": 2.19946571682992, + "grad_norm": 2.515566349029541, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.9121463894844055, + "num_tokens": 671171008.0, + "step": 17290 + }, + { + "epoch": 2.1995929271085104, + "grad_norm": 2.7669317722320557, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.904667854309082, + "num_tokens": 671214096.0, + "step": 17291 + }, + { + "epoch": 2.199720137387101, + "grad_norm": 2.752206325531006, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9103294610977173, + "num_tokens": 671251204.0, + "step": 17292 + }, + { + "epoch": 2.1998473476656915, + "grad_norm": 2.927009105682373, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.8988276720046997, + "num_tokens": 671290934.0, + "step": 17293 + }, + { + "epoch": 2.199974557944282, + "grad_norm": 2.670844554901123, + "learning_rate": 1e-06, + "loss": 0.2318, + "mean_token_accuracy": 0.9138281345367432, + "num_tokens": 671331354.0, + "step": 17294 + }, + { + "epoch": 2.2001017682228725, + "grad_norm": 3.051487922668457, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9069316387176514, + "num_tokens": 671366165.0, + "step": 17295 + }, + { + "epoch": 2.200228978501463, + "grad_norm": 2.7288057804107666, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9043790102005005, + "num_tokens": 671405128.0, + "step": 17296 + }, + { + "epoch": 2.2003561887800536, + "grad_norm": 2.8965537548065186, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9096857309341431, + "num_tokens": 671444617.0, + "step": 17297 + }, + { + "epoch": 2.200483399058644, + "grad_norm": 2.9217774868011475, + "learning_rate": 1e-06, + "loss": 0.2295, + "mean_token_accuracy": 0.9137238264083862, + "num_tokens": 671478847.0, + "step": 17298 + }, + { + "epoch": 2.2006106093372346, + "grad_norm": 2.7963922023773193, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9082733392715454, + "num_tokens": 671518316.0, + "step": 17299 + }, + { + "epoch": 2.200737819615825, + "grad_norm": 3.0179104804992676, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9095080494880676, + "num_tokens": 671551706.0, + "step": 17300 + }, + { + "epoch": 2.2008650298944152, + "grad_norm": 2.8793487548828125, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9074134230613708, + "num_tokens": 671587068.0, + "step": 17301 + }, + { + "epoch": 2.200992240173006, + "grad_norm": 2.993586778640747, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.900395393371582, + "num_tokens": 671626666.0, + "step": 17302 + }, + { + "epoch": 2.2011194504515963, + "grad_norm": 2.5362625122070312, + "learning_rate": 1e-06, + "loss": 0.2236, + "mean_token_accuracy": 0.9156399369239807, + "num_tokens": 671671118.0, + "step": 17303 + }, + { + "epoch": 2.201246660730187, + "grad_norm": 2.939065456390381, + "learning_rate": 1e-06, + "loss": 0.2333, + "mean_token_accuracy": 0.9138760566711426, + "num_tokens": 671706134.0, + "step": 17304 + }, + { + "epoch": 2.2013738710087773, + "grad_norm": 2.7355334758758545, + "learning_rate": 1e-06, + "loss": 0.2089, + "mean_token_accuracy": 0.9213717579841614, + "num_tokens": 671742820.0, + "step": 17305 + }, + { + "epoch": 2.201501081287368, + "grad_norm": 2.885214328765869, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9065346121788025, + "num_tokens": 671781096.0, + "step": 17306 + }, + { + "epoch": 2.2016282915659584, + "grad_norm": 2.7959299087524414, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8986469507217407, + "num_tokens": 671823586.0, + "step": 17307 + }, + { + "epoch": 2.201755501844549, + "grad_norm": 2.9569344520568848, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.8981829285621643, + "num_tokens": 671861490.0, + "step": 17308 + }, + { + "epoch": 2.2018827121231395, + "grad_norm": 3.0864510536193848, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9081282615661621, + "num_tokens": 671895768.0, + "step": 17309 + }, + { + "epoch": 2.20200992240173, + "grad_norm": 2.786346912384033, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9114057421684265, + "num_tokens": 671934848.0, + "step": 17310 + }, + { + "epoch": 2.2021371326803205, + "grad_norm": 3.104649782180786, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8968725204467773, + "num_tokens": 671973298.0, + "step": 17311 + }, + { + "epoch": 2.202264342958911, + "grad_norm": 2.963505744934082, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.908240556716919, + "num_tokens": 672012669.0, + "step": 17312 + }, + { + "epoch": 2.2023915532375016, + "grad_norm": 2.9434990882873535, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9048398733139038, + "num_tokens": 672048797.0, + "step": 17313 + }, + { + "epoch": 2.202518763516092, + "grad_norm": 2.747253894805908, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.912365198135376, + "num_tokens": 672087368.0, + "step": 17314 + }, + { + "epoch": 2.2026459737946826, + "grad_norm": 2.8240103721618652, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9067224860191345, + "num_tokens": 672127845.0, + "step": 17315 + }, + { + "epoch": 2.202773184073273, + "grad_norm": 2.9601187705993652, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8947645425796509, + "num_tokens": 672165948.0, + "step": 17316 + }, + { + "epoch": 2.2029003943518637, + "grad_norm": 3.063279390335083, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9119906425476074, + "num_tokens": 672203179.0, + "step": 17317 + }, + { + "epoch": 2.203027604630454, + "grad_norm": 2.6678285598754883, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.9116595387458801, + "num_tokens": 672246792.0, + "step": 17318 + }, + { + "epoch": 2.2031548149090447, + "grad_norm": 2.7315213680267334, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9010580778121948, + "num_tokens": 672290460.0, + "step": 17319 + }, + { + "epoch": 2.2032820251876353, + "grad_norm": 2.8423593044281006, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9038639068603516, + "num_tokens": 672332017.0, + "step": 17320 + }, + { + "epoch": 2.203409235466226, + "grad_norm": 3.031064748764038, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8902828097343445, + "num_tokens": 672374830.0, + "step": 17321 + }, + { + "epoch": 2.2035364457448163, + "grad_norm": 2.7503156661987305, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9116677641868591, + "num_tokens": 672412432.0, + "step": 17322 + }, + { + "epoch": 2.203663656023407, + "grad_norm": 2.6409852504730225, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8982677459716797, + "num_tokens": 672461090.0, + "step": 17323 + }, + { + "epoch": 2.2037908663019974, + "grad_norm": 2.993882656097412, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9120197296142578, + "num_tokens": 672495904.0, + "step": 17324 + }, + { + "epoch": 2.203918076580588, + "grad_norm": 3.189293384552002, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9056630730628967, + "num_tokens": 672529409.0, + "step": 17325 + }, + { + "epoch": 2.204045286859178, + "grad_norm": 2.9617574214935303, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9066556692123413, + "num_tokens": 672568218.0, + "step": 17326 + }, + { + "epoch": 2.2041724971377685, + "grad_norm": 3.108527660369873, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9065935611724854, + "num_tokens": 672604691.0, + "step": 17327 + }, + { + "epoch": 2.204299707416359, + "grad_norm": 3.014871597290039, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8984271287918091, + "num_tokens": 672643289.0, + "step": 17328 + }, + { + "epoch": 2.2044269176949496, + "grad_norm": 2.8054146766662598, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9062228202819824, + "num_tokens": 672680587.0, + "step": 17329 + }, + { + "epoch": 2.20455412797354, + "grad_norm": 2.9347712993621826, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9083632230758667, + "num_tokens": 672716901.0, + "step": 17330 + }, + { + "epoch": 2.2046813382521306, + "grad_norm": 3.0673329830169678, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9032958745956421, + "num_tokens": 672751049.0, + "step": 17331 + }, + { + "epoch": 2.204808548530721, + "grad_norm": 2.6693637371063232, + "learning_rate": 1e-06, + "loss": 0.2115, + "mean_token_accuracy": 0.9206864833831787, + "num_tokens": 672788160.0, + "step": 17332 + }, + { + "epoch": 2.2049357588093117, + "grad_norm": 3.209839344024658, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.8967881202697754, + "num_tokens": 672824005.0, + "step": 17333 + }, + { + "epoch": 2.2050629690879022, + "grad_norm": 3.041569709777832, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.902258038520813, + "num_tokens": 672863333.0, + "step": 17334 + }, + { + "epoch": 2.2051901793664928, + "grad_norm": 3.132749557495117, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9041389226913452, + "num_tokens": 672902625.0, + "step": 17335 + }, + { + "epoch": 2.2053173896450833, + "grad_norm": 2.712557554244995, + "learning_rate": 1e-06, + "loss": 0.2398, + "mean_token_accuracy": 0.9107190370559692, + "num_tokens": 672942101.0, + "step": 17336 + }, + { + "epoch": 2.205444599923674, + "grad_norm": 3.043581008911133, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9020267724990845, + "num_tokens": 672978233.0, + "step": 17337 + }, + { + "epoch": 2.2055718102022643, + "grad_norm": 2.965564250946045, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9025344848632812, + "num_tokens": 673017136.0, + "step": 17338 + }, + { + "epoch": 2.205699020480855, + "grad_norm": 3.1005117893218994, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8933895826339722, + "num_tokens": 673056743.0, + "step": 17339 + }, + { + "epoch": 2.2058262307594454, + "grad_norm": 2.787365436553955, + "learning_rate": 1e-06, + "loss": 0.2236, + "mean_token_accuracy": 0.9161210060119629, + "num_tokens": 673093932.0, + "step": 17340 + }, + { + "epoch": 2.205953441038036, + "grad_norm": 2.936244010925293, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9082293510437012, + "num_tokens": 673130892.0, + "step": 17341 + }, + { + "epoch": 2.2060806513166265, + "grad_norm": 2.9985454082489014, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9160208106040955, + "num_tokens": 673163424.0, + "step": 17342 + }, + { + "epoch": 2.206207861595217, + "grad_norm": 3.0587825775146484, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.901847779750824, + "num_tokens": 673200789.0, + "step": 17343 + }, + { + "epoch": 2.2063350718738075, + "grad_norm": 3.169220447540283, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8953806757926941, + "num_tokens": 673236861.0, + "step": 17344 + }, + { + "epoch": 2.206462282152398, + "grad_norm": 3.149600028991699, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9064202308654785, + "num_tokens": 673272814.0, + "step": 17345 + }, + { + "epoch": 2.2065894924309886, + "grad_norm": 2.9620838165283203, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.901215672492981, + "num_tokens": 673310852.0, + "step": 17346 + }, + { + "epoch": 2.206716702709579, + "grad_norm": 2.9574697017669678, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.9133856892585754, + "num_tokens": 673346119.0, + "step": 17347 + }, + { + "epoch": 2.2068439129881696, + "grad_norm": 2.7380480766296387, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9081099033355713, + "num_tokens": 673386536.0, + "step": 17348 + }, + { + "epoch": 2.20697112326676, + "grad_norm": 2.804741859436035, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9078890681266785, + "num_tokens": 673429767.0, + "step": 17349 + }, + { + "epoch": 2.2070983335453507, + "grad_norm": 3.1032378673553467, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.8988410234451294, + "num_tokens": 673465835.0, + "step": 17350 + }, + { + "epoch": 2.2072255438239408, + "grad_norm": 3.2438971996307373, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9016329050064087, + "num_tokens": 673499900.0, + "step": 17351 + }, + { + "epoch": 2.2073527541025313, + "grad_norm": 2.6738948822021484, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9105238914489746, + "num_tokens": 673543234.0, + "step": 17352 + }, + { + "epoch": 2.207479964381122, + "grad_norm": 2.7668440341949463, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.8959400653839111, + "num_tokens": 673585272.0, + "step": 17353 + }, + { + "epoch": 2.2076071746597123, + "grad_norm": 2.8045177459716797, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9096734523773193, + "num_tokens": 673622064.0, + "step": 17354 + }, + { + "epoch": 2.207734384938303, + "grad_norm": 2.8528053760528564, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.8985859155654907, + "num_tokens": 673664708.0, + "step": 17355 + }, + { + "epoch": 2.2078615952168934, + "grad_norm": 2.942685842514038, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8918311595916748, + "num_tokens": 673704728.0, + "step": 17356 + }, + { + "epoch": 2.207988805495484, + "grad_norm": 3.156778573989868, + "learning_rate": 1e-06, + "loss": 0.3217, + "mean_token_accuracy": 0.8872488737106323, + "num_tokens": 673742533.0, + "step": 17357 + }, + { + "epoch": 2.2081160157740745, + "grad_norm": 2.3951966762542725, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9122500419616699, + "num_tokens": 673790957.0, + "step": 17358 + }, + { + "epoch": 2.208243226052665, + "grad_norm": 3.151292324066162, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8933638334274292, + "num_tokens": 673830894.0, + "step": 17359 + }, + { + "epoch": 2.2083704363312555, + "grad_norm": 3.2918386459350586, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9002733826637268, + "num_tokens": 673862135.0, + "step": 17360 + }, + { + "epoch": 2.208497646609846, + "grad_norm": 2.8504185676574707, + "learning_rate": 1e-06, + "loss": 0.2269, + "mean_token_accuracy": 0.9140474200248718, + "num_tokens": 673897452.0, + "step": 17361 + }, + { + "epoch": 2.2086248568884366, + "grad_norm": 2.764134168624878, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9014961123466492, + "num_tokens": 673938417.0, + "step": 17362 + }, + { + "epoch": 2.208752067167027, + "grad_norm": 2.5729153156280518, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9085009098052979, + "num_tokens": 673985056.0, + "step": 17363 + }, + { + "epoch": 2.2088792774456176, + "grad_norm": 2.8620986938476562, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9040528535842896, + "num_tokens": 674022836.0, + "step": 17364 + }, + { + "epoch": 2.209006487724208, + "grad_norm": 2.7082037925720215, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9095554947853088, + "num_tokens": 674060541.0, + "step": 17365 + }, + { + "epoch": 2.2091336980027987, + "grad_norm": 2.9673545360565186, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8943616151809692, + "num_tokens": 674099491.0, + "step": 17366 + }, + { + "epoch": 2.209260908281389, + "grad_norm": 2.5779709815979004, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9056993722915649, + "num_tokens": 674143343.0, + "step": 17367 + }, + { + "epoch": 2.2093881185599797, + "grad_norm": 2.8457884788513184, + "learning_rate": 1e-06, + "loss": 0.2318, + "mean_token_accuracy": 0.914872944355011, + "num_tokens": 674182153.0, + "step": 17368 + }, + { + "epoch": 2.2095153288385703, + "grad_norm": 2.922858953475952, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.901300847530365, + "num_tokens": 674221179.0, + "step": 17369 + }, + { + "epoch": 2.209642539117161, + "grad_norm": 2.7349061965942383, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9064649343490601, + "num_tokens": 674261532.0, + "step": 17370 + }, + { + "epoch": 2.2097697493957513, + "grad_norm": 2.8962457180023193, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8976503014564514, + "num_tokens": 674305250.0, + "step": 17371 + }, + { + "epoch": 2.209896959674342, + "grad_norm": 2.6626813411712646, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9069067239761353, + "num_tokens": 674346167.0, + "step": 17372 + }, + { + "epoch": 2.2100241699529324, + "grad_norm": 2.576348304748535, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.909417450428009, + "num_tokens": 674391776.0, + "step": 17373 + }, + { + "epoch": 2.210151380231523, + "grad_norm": 3.0334837436676025, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8961620330810547, + "num_tokens": 674428034.0, + "step": 17374 + }, + { + "epoch": 2.2102785905101134, + "grad_norm": 2.66684889793396, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9042178988456726, + "num_tokens": 674473395.0, + "step": 17375 + }, + { + "epoch": 2.2104058007887035, + "grad_norm": 2.9836506843566895, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9043130874633789, + "num_tokens": 674506246.0, + "step": 17376 + }, + { + "epoch": 2.210533011067294, + "grad_norm": 2.8797595500946045, + "learning_rate": 1e-06, + "loss": 0.2338, + "mean_token_accuracy": 0.910999059677124, + "num_tokens": 674539444.0, + "step": 17377 + }, + { + "epoch": 2.2106602213458846, + "grad_norm": 2.510411500930786, + "learning_rate": 1e-06, + "loss": 0.2222, + "mean_token_accuracy": 0.9165575504302979, + "num_tokens": 674581925.0, + "step": 17378 + }, + { + "epoch": 2.210787431624475, + "grad_norm": 2.7881128787994385, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9010143876075745, + "num_tokens": 674624019.0, + "step": 17379 + }, + { + "epoch": 2.2109146419030656, + "grad_norm": 2.9615230560302734, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8955405354499817, + "num_tokens": 674664163.0, + "step": 17380 + }, + { + "epoch": 2.211041852181656, + "grad_norm": 2.700629234313965, + "learning_rate": 1e-06, + "loss": 0.2398, + "mean_token_accuracy": 0.9123668670654297, + "num_tokens": 674702472.0, + "step": 17381 + }, + { + "epoch": 2.2111690624602467, + "grad_norm": 3.053215265274048, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9100204706192017, + "num_tokens": 674736217.0, + "step": 17382 + }, + { + "epoch": 2.211296272738837, + "grad_norm": 2.9309909343719482, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9084293842315674, + "num_tokens": 674772423.0, + "step": 17383 + }, + { + "epoch": 2.2114234830174277, + "grad_norm": 3.092719078063965, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8947346210479736, + "num_tokens": 674806085.0, + "step": 17384 + }, + { + "epoch": 2.2115506932960183, + "grad_norm": 2.683567523956299, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9135175943374634, + "num_tokens": 674846526.0, + "step": 17385 + }, + { + "epoch": 2.211677903574609, + "grad_norm": 2.9911487102508545, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8950838446617126, + "num_tokens": 674884583.0, + "step": 17386 + }, + { + "epoch": 2.2118051138531993, + "grad_norm": 2.888015031814575, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8977051973342896, + "num_tokens": 674924230.0, + "step": 17387 + }, + { + "epoch": 2.21193232413179, + "grad_norm": 2.8225786685943604, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8971034288406372, + "num_tokens": 674968442.0, + "step": 17388 + }, + { + "epoch": 2.2120595344103804, + "grad_norm": 2.8842785358428955, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9046573638916016, + "num_tokens": 675004125.0, + "step": 17389 + }, + { + "epoch": 2.212186744688971, + "grad_norm": 3.0146889686584473, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9035980105400085, + "num_tokens": 675038943.0, + "step": 17390 + }, + { + "epoch": 2.2123139549675614, + "grad_norm": 2.682112216949463, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9064061641693115, + "num_tokens": 675081561.0, + "step": 17391 + }, + { + "epoch": 2.212441165246152, + "grad_norm": 3.0930252075195312, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.904952347278595, + "num_tokens": 675116966.0, + "step": 17392 + }, + { + "epoch": 2.2125683755247425, + "grad_norm": 3.0616722106933594, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.8976993560791016, + "num_tokens": 675152746.0, + "step": 17393 + }, + { + "epoch": 2.212695585803333, + "grad_norm": 3.16583251953125, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8994395136833191, + "num_tokens": 675186895.0, + "step": 17394 + }, + { + "epoch": 2.2128227960819236, + "grad_norm": 2.7157015800476074, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9062095880508423, + "num_tokens": 675229525.0, + "step": 17395 + }, + { + "epoch": 2.212950006360514, + "grad_norm": 2.7111456394195557, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.9003028869628906, + "num_tokens": 675270958.0, + "step": 17396 + }, + { + "epoch": 2.2130772166391046, + "grad_norm": 2.8335442543029785, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9100416898727417, + "num_tokens": 675307033.0, + "step": 17397 + }, + { + "epoch": 2.213204426917695, + "grad_norm": 2.9150471687316895, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9005085229873657, + "num_tokens": 675345972.0, + "step": 17398 + }, + { + "epoch": 2.2133316371962852, + "grad_norm": 2.898427724838257, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.903367280960083, + "num_tokens": 675383168.0, + "step": 17399 + }, + { + "epoch": 2.213458847474876, + "grad_norm": 2.7426319122314453, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9042655825614929, + "num_tokens": 675426089.0, + "step": 17400 + }, + { + "epoch": 2.2135860577534663, + "grad_norm": 2.6182148456573486, + "learning_rate": 1e-06, + "loss": 0.2254, + "mean_token_accuracy": 0.9165617227554321, + "num_tokens": 675468541.0, + "step": 17401 + }, + { + "epoch": 2.213713268032057, + "grad_norm": 2.8468337059020996, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8983603119850159, + "num_tokens": 675509423.0, + "step": 17402 + }, + { + "epoch": 2.2138404783106473, + "grad_norm": 2.5681777000427246, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.9103014469146729, + "num_tokens": 675551961.0, + "step": 17403 + }, + { + "epoch": 2.213967688589238, + "grad_norm": 2.8050477504730225, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9099904298782349, + "num_tokens": 675589989.0, + "step": 17404 + }, + { + "epoch": 2.2140948988678284, + "grad_norm": 2.761780023574829, + "learning_rate": 1e-06, + "loss": 0.2249, + "mean_token_accuracy": 0.9163774251937866, + "num_tokens": 675628876.0, + "step": 17405 + }, + { + "epoch": 2.214222109146419, + "grad_norm": 2.864856243133545, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8958793878555298, + "num_tokens": 675668811.0, + "step": 17406 + }, + { + "epoch": 2.2143493194250095, + "grad_norm": 2.7123539447784424, + "learning_rate": 1e-06, + "loss": 0.2349, + "mean_token_accuracy": 0.9148956537246704, + "num_tokens": 675710329.0, + "step": 17407 + }, + { + "epoch": 2.2144765297036, + "grad_norm": 3.201601982116699, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8946760892868042, + "num_tokens": 675744012.0, + "step": 17408 + }, + { + "epoch": 2.2146037399821905, + "grad_norm": 2.8493733406066895, + "learning_rate": 1e-06, + "loss": 0.2288, + "mean_token_accuracy": 0.914756178855896, + "num_tokens": 675777787.0, + "step": 17409 + }, + { + "epoch": 2.214730950260781, + "grad_norm": 2.8639321327209473, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8989393711090088, + "num_tokens": 675819760.0, + "step": 17410 + }, + { + "epoch": 2.2148581605393716, + "grad_norm": 2.9318692684173584, + "learning_rate": 1e-06, + "loss": 0.2226, + "mean_token_accuracy": 0.9167567491531372, + "num_tokens": 675854496.0, + "step": 17411 + }, + { + "epoch": 2.214985370817962, + "grad_norm": 2.930516004562378, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9122436046600342, + "num_tokens": 675890020.0, + "step": 17412 + }, + { + "epoch": 2.2151125810965526, + "grad_norm": 2.857107400894165, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9047384262084961, + "num_tokens": 675932711.0, + "step": 17413 + }, + { + "epoch": 2.215239791375143, + "grad_norm": 3.1948442459106445, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8873999118804932, + "num_tokens": 675976332.0, + "step": 17414 + }, + { + "epoch": 2.2153670016537337, + "grad_norm": 2.6532394886016846, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9128634929656982, + "num_tokens": 676017195.0, + "step": 17415 + }, + { + "epoch": 2.215494211932324, + "grad_norm": 2.8711421489715576, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9005287289619446, + "num_tokens": 676057302.0, + "step": 17416 + }, + { + "epoch": 2.2156214222109147, + "grad_norm": 2.7766120433807373, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9087911248207092, + "num_tokens": 676096317.0, + "step": 17417 + }, + { + "epoch": 2.2157486324895053, + "grad_norm": 2.7570698261260986, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9119947552680969, + "num_tokens": 676138341.0, + "step": 17418 + }, + { + "epoch": 2.215875842768096, + "grad_norm": 2.7379438877105713, + "learning_rate": 1e-06, + "loss": 0.2363, + "mean_token_accuracy": 0.91249680519104, + "num_tokens": 676177640.0, + "step": 17419 + }, + { + "epoch": 2.2160030530466863, + "grad_norm": 2.7961955070495605, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9044884443283081, + "num_tokens": 676220148.0, + "step": 17420 + }, + { + "epoch": 2.216130263325277, + "grad_norm": 3.0497653484344482, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9091343283653259, + "num_tokens": 676254780.0, + "step": 17421 + }, + { + "epoch": 2.2162574736038674, + "grad_norm": 3.0312511920928955, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9042308330535889, + "num_tokens": 676292013.0, + "step": 17422 + }, + { + "epoch": 2.216384683882458, + "grad_norm": 2.6278905868530273, + "learning_rate": 1e-06, + "loss": 0.2242, + "mean_token_accuracy": 0.916130542755127, + "num_tokens": 676333422.0, + "step": 17423 + }, + { + "epoch": 2.216511894161048, + "grad_norm": 2.7560291290283203, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9033548831939697, + "num_tokens": 676373528.0, + "step": 17424 + }, + { + "epoch": 2.2166391044396385, + "grad_norm": 3.0333101749420166, + "learning_rate": 1e-06, + "loss": 0.3281, + "mean_token_accuracy": 0.881251871585846, + "num_tokens": 676416279.0, + "step": 17425 + }, + { + "epoch": 2.216766314718229, + "grad_norm": 2.9627273082733154, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9019143581390381, + "num_tokens": 676450787.0, + "step": 17426 + }, + { + "epoch": 2.2168935249968196, + "grad_norm": 2.716102123260498, + "learning_rate": 1e-06, + "loss": 0.225, + "mean_token_accuracy": 0.9157279133796692, + "num_tokens": 676492618.0, + "step": 17427 + }, + { + "epoch": 2.21702073527541, + "grad_norm": 2.899395227432251, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8999846577644348, + "num_tokens": 676532709.0, + "step": 17428 + }, + { + "epoch": 2.2171479455540006, + "grad_norm": 3.203174352645874, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8967071771621704, + "num_tokens": 676570175.0, + "step": 17429 + }, + { + "epoch": 2.217275155832591, + "grad_norm": 3.231003522872925, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.8987498879432678, + "num_tokens": 676601846.0, + "step": 17430 + }, + { + "epoch": 2.2174023661111817, + "grad_norm": 3.1380345821380615, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9061553478240967, + "num_tokens": 676637590.0, + "step": 17431 + }, + { + "epoch": 2.217529576389772, + "grad_norm": 3.0803232192993164, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8986177444458008, + "num_tokens": 676671760.0, + "step": 17432 + }, + { + "epoch": 2.2176567866683627, + "grad_norm": 2.793323516845703, + "learning_rate": 1e-06, + "loss": 0.2364, + "mean_token_accuracy": 0.9107456207275391, + "num_tokens": 676713419.0, + "step": 17433 + }, + { + "epoch": 2.2177839969469533, + "grad_norm": 2.733861207962036, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.902911901473999, + "num_tokens": 676757577.0, + "step": 17434 + }, + { + "epoch": 2.217911207225544, + "grad_norm": 2.9791131019592285, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9136396646499634, + "num_tokens": 676790396.0, + "step": 17435 + }, + { + "epoch": 2.2180384175041343, + "grad_norm": 3.284353733062744, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.8992522358894348, + "num_tokens": 676822361.0, + "step": 17436 + }, + { + "epoch": 2.218165627782725, + "grad_norm": 2.8001794815063477, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9059224128723145, + "num_tokens": 676863046.0, + "step": 17437 + }, + { + "epoch": 2.2182928380613154, + "grad_norm": 2.760740041732788, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9058942198753357, + "num_tokens": 676901658.0, + "step": 17438 + }, + { + "epoch": 2.218420048339906, + "grad_norm": 2.7156784534454346, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.904391884803772, + "num_tokens": 676941034.0, + "step": 17439 + }, + { + "epoch": 2.2185472586184964, + "grad_norm": 2.9844746589660645, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.905214786529541, + "num_tokens": 676980412.0, + "step": 17440 + }, + { + "epoch": 2.218674468897087, + "grad_norm": 2.9287517070770264, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9081759452819824, + "num_tokens": 677015378.0, + "step": 17441 + }, + { + "epoch": 2.2188016791756775, + "grad_norm": 2.872727394104004, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9068379402160645, + "num_tokens": 677056178.0, + "step": 17442 + }, + { + "epoch": 2.218928889454268, + "grad_norm": 2.9530797004699707, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.8967419266700745, + "num_tokens": 677095008.0, + "step": 17443 + }, + { + "epoch": 2.2190560997328586, + "grad_norm": 3.13195538520813, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.897663950920105, + "num_tokens": 677137630.0, + "step": 17444 + }, + { + "epoch": 2.219183310011449, + "grad_norm": 2.7408459186553955, + "learning_rate": 1e-06, + "loss": 0.2267, + "mean_token_accuracy": 0.9131531715393066, + "num_tokens": 677175763.0, + "step": 17445 + }, + { + "epoch": 2.2193105202900396, + "grad_norm": 3.218909978866577, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9034827947616577, + "num_tokens": 677208148.0, + "step": 17446 + }, + { + "epoch": 2.21943773056863, + "grad_norm": 3.1342854499816895, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9047245979309082, + "num_tokens": 677241273.0, + "step": 17447 + }, + { + "epoch": 2.2195649408472207, + "grad_norm": 3.1098127365112305, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9012267589569092, + "num_tokens": 677277008.0, + "step": 17448 + }, + { + "epoch": 2.2196921511258108, + "grad_norm": 3.2266011238098145, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9069123864173889, + "num_tokens": 677310103.0, + "step": 17449 + }, + { + "epoch": 2.2198193614044013, + "grad_norm": 2.6489853858947754, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9086645841598511, + "num_tokens": 677357086.0, + "step": 17450 + }, + { + "epoch": 2.219946571682992, + "grad_norm": 2.646174430847168, + "learning_rate": 1e-06, + "loss": 0.2226, + "mean_token_accuracy": 0.9169905185699463, + "num_tokens": 677396493.0, + "step": 17451 + }, + { + "epoch": 2.2200737819615823, + "grad_norm": 2.606905460357666, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9101252555847168, + "num_tokens": 677439241.0, + "step": 17452 + }, + { + "epoch": 2.220200992240173, + "grad_norm": 2.899369239807129, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9002978801727295, + "num_tokens": 677478096.0, + "step": 17453 + }, + { + "epoch": 2.2203282025187634, + "grad_norm": 2.8637640476226807, + "learning_rate": 1e-06, + "loss": 0.2276, + "mean_token_accuracy": 0.912606954574585, + "num_tokens": 677511066.0, + "step": 17454 + }, + { + "epoch": 2.220455412797354, + "grad_norm": 2.624403476715088, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9108864068984985, + "num_tokens": 677554580.0, + "step": 17455 + }, + { + "epoch": 2.2205826230759445, + "grad_norm": 2.8202695846557617, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9092615842819214, + "num_tokens": 677592826.0, + "step": 17456 + }, + { + "epoch": 2.220709833354535, + "grad_norm": 2.644010305404663, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9078559279441833, + "num_tokens": 677636773.0, + "step": 17457 + }, + { + "epoch": 2.2208370436331255, + "grad_norm": 2.920332193374634, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9060630798339844, + "num_tokens": 677675367.0, + "step": 17458 + }, + { + "epoch": 2.220964253911716, + "grad_norm": 3.0088694095611572, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9061469435691833, + "num_tokens": 677711071.0, + "step": 17459 + }, + { + "epoch": 2.2210914641903066, + "grad_norm": 2.9552974700927734, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8955186605453491, + "num_tokens": 677749447.0, + "step": 17460 + }, + { + "epoch": 2.221218674468897, + "grad_norm": 2.7781364917755127, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9048221111297607, + "num_tokens": 677793008.0, + "step": 17461 + }, + { + "epoch": 2.2213458847474876, + "grad_norm": 2.7308261394500732, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9050432443618774, + "num_tokens": 677840168.0, + "step": 17462 + }, + { + "epoch": 2.221473095026078, + "grad_norm": 3.1905460357666016, + "learning_rate": 1e-06, + "loss": 0.3113, + "mean_token_accuracy": 0.8869219422340393, + "num_tokens": 677875925.0, + "step": 17463 + }, + { + "epoch": 2.2216003053046687, + "grad_norm": 3.1209967136383057, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9008394479751587, + "num_tokens": 677910190.0, + "step": 17464 + }, + { + "epoch": 2.221727515583259, + "grad_norm": 2.6296191215515137, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9087265729904175, + "num_tokens": 677953399.0, + "step": 17465 + }, + { + "epoch": 2.2218547258618497, + "grad_norm": 2.7742044925689697, + "learning_rate": 1e-06, + "loss": 0.2288, + "mean_token_accuracy": 0.9146477580070496, + "num_tokens": 677991379.0, + "step": 17466 + }, + { + "epoch": 2.2219819361404403, + "grad_norm": 2.5275826454162598, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9140701293945312, + "num_tokens": 678034028.0, + "step": 17467 + }, + { + "epoch": 2.222109146419031, + "grad_norm": 2.651218891143799, + "learning_rate": 1e-06, + "loss": 0.2103, + "mean_token_accuracy": 0.9199018478393555, + "num_tokens": 678074855.0, + "step": 17468 + }, + { + "epoch": 2.2222363566976213, + "grad_norm": 2.942671537399292, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9068771600723267, + "num_tokens": 678113432.0, + "step": 17469 + }, + { + "epoch": 2.222363566976212, + "grad_norm": 2.7744228839874268, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.902591347694397, + "num_tokens": 678155299.0, + "step": 17470 + }, + { + "epoch": 2.2224907772548024, + "grad_norm": 2.6860294342041016, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9056038856506348, + "num_tokens": 678199432.0, + "step": 17471 + }, + { + "epoch": 2.222617987533393, + "grad_norm": 2.788611888885498, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9091944694519043, + "num_tokens": 678237580.0, + "step": 17472 + }, + { + "epoch": 2.2227451978119834, + "grad_norm": 3.1257479190826416, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.893303632736206, + "num_tokens": 678272990.0, + "step": 17473 + }, + { + "epoch": 2.2228724080905735, + "grad_norm": 3.1037867069244385, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9023607969284058, + "num_tokens": 678311228.0, + "step": 17474 + }, + { + "epoch": 2.222999618369164, + "grad_norm": 3.241189956665039, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9007340669631958, + "num_tokens": 678346210.0, + "step": 17475 + }, + { + "epoch": 2.2231268286477546, + "grad_norm": 2.883842706680298, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9020320177078247, + "num_tokens": 678386945.0, + "step": 17476 + }, + { + "epoch": 2.223254038926345, + "grad_norm": 2.6559345722198486, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9116252660751343, + "num_tokens": 678427432.0, + "step": 17477 + }, + { + "epoch": 2.2233812492049356, + "grad_norm": 2.880113363265991, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9116926193237305, + "num_tokens": 678460160.0, + "step": 17478 + }, + { + "epoch": 2.223508459483526, + "grad_norm": 2.7177298069000244, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.902273416519165, + "num_tokens": 678503210.0, + "step": 17479 + }, + { + "epoch": 2.2236356697621167, + "grad_norm": 2.934556245803833, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9050382375717163, + "num_tokens": 678544084.0, + "step": 17480 + }, + { + "epoch": 2.223762880040707, + "grad_norm": 2.944357395172119, + "learning_rate": 1e-06, + "loss": 0.233, + "mean_token_accuracy": 0.911353349685669, + "num_tokens": 678579984.0, + "step": 17481 + }, + { + "epoch": 2.2238900903192977, + "grad_norm": 2.881382703781128, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9103671312332153, + "num_tokens": 678616674.0, + "step": 17482 + }, + { + "epoch": 2.2240173005978883, + "grad_norm": 2.7224481105804443, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9127925634384155, + "num_tokens": 678656201.0, + "step": 17483 + }, + { + "epoch": 2.224144510876479, + "grad_norm": 2.9822440147399902, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9006400108337402, + "num_tokens": 678692770.0, + "step": 17484 + }, + { + "epoch": 2.2242717211550693, + "grad_norm": 2.7833023071289062, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.9173349738121033, + "num_tokens": 678726184.0, + "step": 17485 + }, + { + "epoch": 2.22439893143366, + "grad_norm": 2.790262222290039, + "learning_rate": 1e-06, + "loss": 0.2264, + "mean_token_accuracy": 0.9161505699157715, + "num_tokens": 678761612.0, + "step": 17486 + }, + { + "epoch": 2.2245261417122504, + "grad_norm": 2.9483907222747803, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8994816541671753, + "num_tokens": 678803147.0, + "step": 17487 + }, + { + "epoch": 2.224653351990841, + "grad_norm": 2.7833313941955566, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9135968089103699, + "num_tokens": 678841546.0, + "step": 17488 + }, + { + "epoch": 2.2247805622694314, + "grad_norm": 2.7833282947540283, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.913345992565155, + "num_tokens": 678880926.0, + "step": 17489 + }, + { + "epoch": 2.224907772548022, + "grad_norm": 3.02569580078125, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8954250812530518, + "num_tokens": 678921396.0, + "step": 17490 + }, + { + "epoch": 2.2250349828266125, + "grad_norm": 2.6614949703216553, + "learning_rate": 1e-06, + "loss": 0.2345, + "mean_token_accuracy": 0.9140231609344482, + "num_tokens": 678960195.0, + "step": 17491 + }, + { + "epoch": 2.225162193105203, + "grad_norm": 2.7423722743988037, + "learning_rate": 1e-06, + "loss": 0.2295, + "mean_token_accuracy": 0.911105751991272, + "num_tokens": 678997099.0, + "step": 17492 + }, + { + "epoch": 2.2252894033837936, + "grad_norm": 3.0765957832336426, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9074149131774902, + "num_tokens": 679034549.0, + "step": 17493 + }, + { + "epoch": 2.225416613662384, + "grad_norm": 2.852039337158203, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9080973267555237, + "num_tokens": 679074106.0, + "step": 17494 + }, + { + "epoch": 2.2255438239409746, + "grad_norm": 2.490035057067871, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9066555500030518, + "num_tokens": 679120973.0, + "step": 17495 + }, + { + "epoch": 2.225671034219565, + "grad_norm": 2.9467477798461914, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9083517789840698, + "num_tokens": 679154297.0, + "step": 17496 + }, + { + "epoch": 2.225798244498155, + "grad_norm": 3.0225982666015625, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.8983404636383057, + "num_tokens": 679193410.0, + "step": 17497 + }, + { + "epoch": 2.225925454776746, + "grad_norm": 2.700512409210205, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9038399457931519, + "num_tokens": 679234202.0, + "step": 17498 + }, + { + "epoch": 2.2260526650553363, + "grad_norm": 3.066871404647827, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.8993152379989624, + "num_tokens": 679272120.0, + "step": 17499 + }, + { + "epoch": 2.226179875333927, + "grad_norm": 2.7267377376556396, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9005054235458374, + "num_tokens": 679317650.0, + "step": 17500 + }, + { + "epoch": 2.2263070856125173, + "grad_norm": 2.8727359771728516, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9094542264938354, + "num_tokens": 679354998.0, + "step": 17501 + }, + { + "epoch": 2.226434295891108, + "grad_norm": 2.824733018875122, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9005612134933472, + "num_tokens": 679397393.0, + "step": 17502 + }, + { + "epoch": 2.2265615061696984, + "grad_norm": 3.094755172729492, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8989505767822266, + "num_tokens": 679433785.0, + "step": 17503 + }, + { + "epoch": 2.226688716448289, + "grad_norm": 2.8536205291748047, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9104089140892029, + "num_tokens": 679475167.0, + "step": 17504 + }, + { + "epoch": 2.2268159267268794, + "grad_norm": 2.8522629737854004, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9038047194480896, + "num_tokens": 679513169.0, + "step": 17505 + }, + { + "epoch": 2.22694313700547, + "grad_norm": 2.89847993850708, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9057093858718872, + "num_tokens": 679551665.0, + "step": 17506 + }, + { + "epoch": 2.2270703472840605, + "grad_norm": 2.766852617263794, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9057127237319946, + "num_tokens": 679592160.0, + "step": 17507 + }, + { + "epoch": 2.227197557562651, + "grad_norm": 2.58194899559021, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9110016822814941, + "num_tokens": 679633132.0, + "step": 17508 + }, + { + "epoch": 2.2273247678412416, + "grad_norm": 2.8268094062805176, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.8990941047668457, + "num_tokens": 679675238.0, + "step": 17509 + }, + { + "epoch": 2.227451978119832, + "grad_norm": 3.296529769897461, + "learning_rate": 1e-06, + "loss": 0.2165, + "mean_token_accuracy": 0.9190211296081543, + "num_tokens": 679703601.0, + "step": 17510 + }, + { + "epoch": 2.2275791883984226, + "grad_norm": 2.9775471687316895, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9070963263511658, + "num_tokens": 679740818.0, + "step": 17511 + }, + { + "epoch": 2.227706398677013, + "grad_norm": 2.894778251647949, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9005613923072815, + "num_tokens": 679783379.0, + "step": 17512 + }, + { + "epoch": 2.2278336089556037, + "grad_norm": 2.861008882522583, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8984373807907104, + "num_tokens": 679824104.0, + "step": 17513 + }, + { + "epoch": 2.227960819234194, + "grad_norm": 2.9130632877349854, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9077567458152771, + "num_tokens": 679861399.0, + "step": 17514 + }, + { + "epoch": 2.2280880295127847, + "grad_norm": 2.8034567832946777, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9092922210693359, + "num_tokens": 679899780.0, + "step": 17515 + }, + { + "epoch": 2.2282152397913753, + "grad_norm": 2.9009485244750977, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9074937701225281, + "num_tokens": 679937186.0, + "step": 17516 + }, + { + "epoch": 2.228342450069966, + "grad_norm": 3.095130443572998, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9059211611747742, + "num_tokens": 679973149.0, + "step": 17517 + }, + { + "epoch": 2.2284696603485563, + "grad_norm": 3.154209613800049, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9033404588699341, + "num_tokens": 680010778.0, + "step": 17518 + }, + { + "epoch": 2.228596870627147, + "grad_norm": 2.9031665325164795, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9026058912277222, + "num_tokens": 680050444.0, + "step": 17519 + }, + { + "epoch": 2.2287240809057374, + "grad_norm": 2.7068045139312744, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.913826584815979, + "num_tokens": 680086522.0, + "step": 17520 + }, + { + "epoch": 2.228851291184328, + "grad_norm": 2.6181256771087646, + "learning_rate": 1e-06, + "loss": 0.2262, + "mean_token_accuracy": 0.9159400463104248, + "num_tokens": 680127149.0, + "step": 17521 + }, + { + "epoch": 2.228978501462918, + "grad_norm": 2.860307216644287, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8978656530380249, + "num_tokens": 680167433.0, + "step": 17522 + }, + { + "epoch": 2.2291057117415085, + "grad_norm": 2.966360569000244, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8953978419303894, + "num_tokens": 680210725.0, + "step": 17523 + }, + { + "epoch": 2.229232922020099, + "grad_norm": 2.6942250728607178, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8983612060546875, + "num_tokens": 680256241.0, + "step": 17524 + }, + { + "epoch": 2.2293601322986896, + "grad_norm": 2.4955973625183105, + "learning_rate": 1e-06, + "loss": 0.2218, + "mean_token_accuracy": 0.9184873104095459, + "num_tokens": 680303726.0, + "step": 17525 + }, + { + "epoch": 2.22948734257728, + "grad_norm": 3.019920587539673, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9050256013870239, + "num_tokens": 680338382.0, + "step": 17526 + }, + { + "epoch": 2.2296145528558706, + "grad_norm": 2.7338263988494873, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9098215699195862, + "num_tokens": 680378238.0, + "step": 17527 + }, + { + "epoch": 2.229741763134461, + "grad_norm": 2.772143840789795, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9006788730621338, + "num_tokens": 680419842.0, + "step": 17528 + }, + { + "epoch": 2.2298689734130517, + "grad_norm": 2.8450958728790283, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9092895984649658, + "num_tokens": 680459357.0, + "step": 17529 + }, + { + "epoch": 2.229996183691642, + "grad_norm": 2.887491464614868, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8948357105255127, + "num_tokens": 680502865.0, + "step": 17530 + }, + { + "epoch": 2.2301233939702327, + "grad_norm": 3.0562260150909424, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.902026355266571, + "num_tokens": 680538868.0, + "step": 17531 + }, + { + "epoch": 2.2302506042488233, + "grad_norm": 2.7678933143615723, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9059222936630249, + "num_tokens": 680578418.0, + "step": 17532 + }, + { + "epoch": 2.230377814527414, + "grad_norm": 3.25636625289917, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8870786428451538, + "num_tokens": 680619284.0, + "step": 17533 + }, + { + "epoch": 2.2305050248060043, + "grad_norm": 2.874277353286743, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9057557582855225, + "num_tokens": 680659790.0, + "step": 17534 + }, + { + "epoch": 2.230632235084595, + "grad_norm": 2.6663553714752197, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.9135702848434448, + "num_tokens": 680699913.0, + "step": 17535 + }, + { + "epoch": 2.2307594453631854, + "grad_norm": 2.850703477859497, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9098364114761353, + "num_tokens": 680737382.0, + "step": 17536 + }, + { + "epoch": 2.230886655641776, + "grad_norm": 3.079470634460449, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9069693088531494, + "num_tokens": 680774040.0, + "step": 17537 + }, + { + "epoch": 2.2310138659203664, + "grad_norm": 2.847473621368408, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.897936224937439, + "num_tokens": 680816373.0, + "step": 17538 + }, + { + "epoch": 2.231141076198957, + "grad_norm": 2.7611632347106934, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9058279991149902, + "num_tokens": 680860368.0, + "step": 17539 + }, + { + "epoch": 2.2312682864775475, + "grad_norm": 2.6803829669952393, + "learning_rate": 1e-06, + "loss": 0.237, + "mean_token_accuracy": 0.9119528532028198, + "num_tokens": 680902735.0, + "step": 17540 + }, + { + "epoch": 2.231395496756138, + "grad_norm": 2.7456722259521484, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9032296538352966, + "num_tokens": 680949153.0, + "step": 17541 + }, + { + "epoch": 2.2315227070347285, + "grad_norm": 2.9140849113464355, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9085147976875305, + "num_tokens": 680982358.0, + "step": 17542 + }, + { + "epoch": 2.231649917313319, + "grad_norm": 2.7862818241119385, + "learning_rate": 1e-06, + "loss": 0.2301, + "mean_token_accuracy": 0.9152883887290955, + "num_tokens": 681018229.0, + "step": 17543 + }, + { + "epoch": 2.2317771275919096, + "grad_norm": 2.9397480487823486, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9114454984664917, + "num_tokens": 681056483.0, + "step": 17544 + }, + { + "epoch": 2.2319043378705, + "grad_norm": 2.9743950366973877, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9049333333969116, + "num_tokens": 681093431.0, + "step": 17545 + }, + { + "epoch": 2.2320315481490907, + "grad_norm": 2.7062418460845947, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9053475856781006, + "num_tokens": 681136906.0, + "step": 17546 + }, + { + "epoch": 2.2321587584276807, + "grad_norm": 2.8956997394561768, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.904052734375, + "num_tokens": 681175529.0, + "step": 17547 + }, + { + "epoch": 2.2322859687062713, + "grad_norm": 2.995525360107422, + "learning_rate": 1e-06, + "loss": 0.2144, + "mean_token_accuracy": 0.9199209809303284, + "num_tokens": 681204249.0, + "step": 17548 + }, + { + "epoch": 2.232413178984862, + "grad_norm": 2.76952862739563, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9038618803024292, + "num_tokens": 681248735.0, + "step": 17549 + }, + { + "epoch": 2.2325403892634523, + "grad_norm": 2.981229305267334, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9007032513618469, + "num_tokens": 681285328.0, + "step": 17550 + }, + { + "epoch": 2.232667599542043, + "grad_norm": 2.7538001537323, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9077035784721375, + "num_tokens": 681323788.0, + "step": 17551 + }, + { + "epoch": 2.2327948098206334, + "grad_norm": 2.7497990131378174, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9107385277748108, + "num_tokens": 681367982.0, + "step": 17552 + }, + { + "epoch": 2.232922020099224, + "grad_norm": 2.806483268737793, + "learning_rate": 1e-06, + "loss": 0.2339, + "mean_token_accuracy": 0.9104228019714355, + "num_tokens": 681405975.0, + "step": 17553 + }, + { + "epoch": 2.2330492303778144, + "grad_norm": 2.914970874786377, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.906840443611145, + "num_tokens": 681445233.0, + "step": 17554 + }, + { + "epoch": 2.233176440656405, + "grad_norm": 3.0434927940368652, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.904484748840332, + "num_tokens": 681479604.0, + "step": 17555 + }, + { + "epoch": 2.2333036509349955, + "grad_norm": 3.025873899459839, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9009757041931152, + "num_tokens": 681517079.0, + "step": 17556 + }, + { + "epoch": 2.233430861213586, + "grad_norm": 2.935608148574829, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.8986181020736694, + "num_tokens": 681555855.0, + "step": 17557 + }, + { + "epoch": 2.2335580714921766, + "grad_norm": 2.908644914627075, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9074724912643433, + "num_tokens": 681592506.0, + "step": 17558 + }, + { + "epoch": 2.233685281770767, + "grad_norm": 2.872530937194824, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9021919965744019, + "num_tokens": 681635088.0, + "step": 17559 + }, + { + "epoch": 2.2338124920493576, + "grad_norm": 2.765298843383789, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9040746092796326, + "num_tokens": 681676963.0, + "step": 17560 + }, + { + "epoch": 2.233939702327948, + "grad_norm": 3.0140655040740967, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9020477533340454, + "num_tokens": 681711879.0, + "step": 17561 + }, + { + "epoch": 2.2340669126065387, + "grad_norm": 3.158897876739502, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8997470736503601, + "num_tokens": 681747872.0, + "step": 17562 + }, + { + "epoch": 2.234194122885129, + "grad_norm": 3.1556930541992188, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9033889174461365, + "num_tokens": 681786709.0, + "step": 17563 + }, + { + "epoch": 2.2343213331637197, + "grad_norm": 2.999213695526123, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8949878215789795, + "num_tokens": 681822787.0, + "step": 17564 + }, + { + "epoch": 2.2344485434423103, + "grad_norm": 2.939298152923584, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9132915735244751, + "num_tokens": 681856416.0, + "step": 17565 + }, + { + "epoch": 2.234575753720901, + "grad_norm": 2.707324743270874, + "learning_rate": 1e-06, + "loss": 0.2292, + "mean_token_accuracy": 0.9155466556549072, + "num_tokens": 681897634.0, + "step": 17566 + }, + { + "epoch": 2.2347029639994913, + "grad_norm": 2.861572265625, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9060844779014587, + "num_tokens": 681938196.0, + "step": 17567 + }, + { + "epoch": 2.234830174278082, + "grad_norm": 3.0287060737609863, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9067623019218445, + "num_tokens": 681973287.0, + "step": 17568 + }, + { + "epoch": 2.2349573845566724, + "grad_norm": 3.2873475551605225, + "learning_rate": 1e-06, + "loss": 0.2299, + "mean_token_accuracy": 0.9119417667388916, + "num_tokens": 682003304.0, + "step": 17569 + }, + { + "epoch": 2.235084594835263, + "grad_norm": 2.522135019302368, + "learning_rate": 1e-06, + "loss": 0.2264, + "mean_token_accuracy": 0.9151283502578735, + "num_tokens": 682046618.0, + "step": 17570 + }, + { + "epoch": 2.2352118051138534, + "grad_norm": 2.659020185470581, + "learning_rate": 1e-06, + "loss": 0.2413, + "mean_token_accuracy": 0.9090494513511658, + "num_tokens": 682089476.0, + "step": 17571 + }, + { + "epoch": 2.2353390153924435, + "grad_norm": 2.84639310836792, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9097486138343811, + "num_tokens": 682127453.0, + "step": 17572 + }, + { + "epoch": 2.235466225671034, + "grad_norm": 3.156157970428467, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.8991414308547974, + "num_tokens": 682162552.0, + "step": 17573 + }, + { + "epoch": 2.2355934359496246, + "grad_norm": 2.972132921218872, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9027262926101685, + "num_tokens": 682198590.0, + "step": 17574 + }, + { + "epoch": 2.235720646228215, + "grad_norm": 2.874518394470215, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8936786651611328, + "num_tokens": 682238407.0, + "step": 17575 + }, + { + "epoch": 2.2358478565068056, + "grad_norm": 2.89857816696167, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9089306592941284, + "num_tokens": 682274796.0, + "step": 17576 + }, + { + "epoch": 2.235975066785396, + "grad_norm": 2.914959669113159, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9015705585479736, + "num_tokens": 682313115.0, + "step": 17577 + }, + { + "epoch": 2.2361022770639867, + "grad_norm": 2.6994404792785645, + "learning_rate": 1e-06, + "loss": 0.2413, + "mean_token_accuracy": 0.9080410599708557, + "num_tokens": 682355065.0, + "step": 17578 + }, + { + "epoch": 2.236229487342577, + "grad_norm": 2.787034511566162, + "learning_rate": 1e-06, + "loss": 0.2337, + "mean_token_accuracy": 0.9141546487808228, + "num_tokens": 682395040.0, + "step": 17579 + }, + { + "epoch": 2.2363566976211677, + "grad_norm": 2.8982105255126953, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9096981287002563, + "num_tokens": 682432915.0, + "step": 17580 + }, + { + "epoch": 2.2364839078997583, + "grad_norm": 3.047928810119629, + "learning_rate": 1e-06, + "loss": 0.2321, + "mean_token_accuracy": 0.9146993160247803, + "num_tokens": 682466148.0, + "step": 17581 + }, + { + "epoch": 2.236611118178349, + "grad_norm": 2.783707618713379, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.908988356590271, + "num_tokens": 682508050.0, + "step": 17582 + }, + { + "epoch": 2.2367383284569393, + "grad_norm": 2.8068408966064453, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9098380208015442, + "num_tokens": 682548824.0, + "step": 17583 + }, + { + "epoch": 2.23686553873553, + "grad_norm": 2.445047616958618, + "learning_rate": 1e-06, + "loss": 0.2204, + "mean_token_accuracy": 0.9159964323043823, + "num_tokens": 682593642.0, + "step": 17584 + }, + { + "epoch": 2.2369927490141204, + "grad_norm": 2.976679563522339, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8960287570953369, + "num_tokens": 682632530.0, + "step": 17585 + }, + { + "epoch": 2.237119959292711, + "grad_norm": 3.091254234313965, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.8998844623565674, + "num_tokens": 682670600.0, + "step": 17586 + }, + { + "epoch": 2.2372471695713014, + "grad_norm": 2.821895122528076, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.903296947479248, + "num_tokens": 682711254.0, + "step": 17587 + }, + { + "epoch": 2.237374379849892, + "grad_norm": 2.913256883621216, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8928501605987549, + "num_tokens": 682750802.0, + "step": 17588 + }, + { + "epoch": 2.2375015901284825, + "grad_norm": 2.8683948516845703, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.896776020526886, + "num_tokens": 682790709.0, + "step": 17589 + }, + { + "epoch": 2.237628800407073, + "grad_norm": 2.6209237575531006, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9100073575973511, + "num_tokens": 682836842.0, + "step": 17590 + }, + { + "epoch": 2.2377560106856635, + "grad_norm": 3.032466411590576, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9093867540359497, + "num_tokens": 682870441.0, + "step": 17591 + }, + { + "epoch": 2.237883220964254, + "grad_norm": 2.706489086151123, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9109916687011719, + "num_tokens": 682909386.0, + "step": 17592 + }, + { + "epoch": 2.2380104312428446, + "grad_norm": 2.582263708114624, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9111562967300415, + "num_tokens": 682951200.0, + "step": 17593 + }, + { + "epoch": 2.238137641521435, + "grad_norm": 2.6518356800079346, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9056499004364014, + "num_tokens": 682994270.0, + "step": 17594 + }, + { + "epoch": 2.238264851800025, + "grad_norm": 2.7642264366149902, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9027448892593384, + "num_tokens": 683035669.0, + "step": 17595 + }, + { + "epoch": 2.238392062078616, + "grad_norm": 3.2294809818267822, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9057989120483398, + "num_tokens": 683068208.0, + "step": 17596 + }, + { + "epoch": 2.2385192723572063, + "grad_norm": 2.936326503753662, + "learning_rate": 1e-06, + "loss": 0.2302, + "mean_token_accuracy": 0.9157764911651611, + "num_tokens": 683107983.0, + "step": 17597 + }, + { + "epoch": 2.238646482635797, + "grad_norm": 2.686131000518799, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9051982760429382, + "num_tokens": 683150458.0, + "step": 17598 + }, + { + "epoch": 2.2387736929143873, + "grad_norm": 2.7455179691314697, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9066805839538574, + "num_tokens": 683194508.0, + "step": 17599 + }, + { + "epoch": 2.238900903192978, + "grad_norm": 2.904383897781372, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9049133062362671, + "num_tokens": 683234737.0, + "step": 17600 + }, + { + "epoch": 2.2390281134715684, + "grad_norm": 2.668402910232544, + "learning_rate": 1e-06, + "loss": 0.2217, + "mean_token_accuracy": 0.9177396297454834, + "num_tokens": 683276917.0, + "step": 17601 + }, + { + "epoch": 2.239155323750159, + "grad_norm": 2.9592928886413574, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9020545482635498, + "num_tokens": 683313731.0, + "step": 17602 + }, + { + "epoch": 2.2392825340287494, + "grad_norm": 2.8065693378448486, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9103388786315918, + "num_tokens": 683351971.0, + "step": 17603 + }, + { + "epoch": 2.23940974430734, + "grad_norm": 3.17728590965271, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8884881734848022, + "num_tokens": 683388487.0, + "step": 17604 + }, + { + "epoch": 2.2395369545859305, + "grad_norm": 2.7611541748046875, + "learning_rate": 1e-06, + "loss": 0.2435, + "mean_token_accuracy": 0.9117975831031799, + "num_tokens": 683427612.0, + "step": 17605 + }, + { + "epoch": 2.239664164864521, + "grad_norm": 2.936129093170166, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9067543745040894, + "num_tokens": 683466710.0, + "step": 17606 + }, + { + "epoch": 2.2397913751431116, + "grad_norm": 3.1175572872161865, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.9095778465270996, + "num_tokens": 683497995.0, + "step": 17607 + }, + { + "epoch": 2.239918585421702, + "grad_norm": 2.8541057109832764, + "learning_rate": 1e-06, + "loss": 0.2405, + "mean_token_accuracy": 0.9095900058746338, + "num_tokens": 683534282.0, + "step": 17608 + }, + { + "epoch": 2.2400457957002926, + "grad_norm": 2.721682071685791, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9041481018066406, + "num_tokens": 683575032.0, + "step": 17609 + }, + { + "epoch": 2.240173005978883, + "grad_norm": 3.0653769969940186, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8887979984283447, + "num_tokens": 683615178.0, + "step": 17610 + }, + { + "epoch": 2.2403002162574737, + "grad_norm": 2.723050832748413, + "learning_rate": 1e-06, + "loss": 0.2362, + "mean_token_accuracy": 0.9092718362808228, + "num_tokens": 683652970.0, + "step": 17611 + }, + { + "epoch": 2.240427426536064, + "grad_norm": 2.550367593765259, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9174529314041138, + "num_tokens": 683695770.0, + "step": 17612 + }, + { + "epoch": 2.2405546368146547, + "grad_norm": 2.542926788330078, + "learning_rate": 1e-06, + "loss": 0.2351, + "mean_token_accuracy": 0.9117772579193115, + "num_tokens": 683740449.0, + "step": 17613 + }, + { + "epoch": 2.2406818470932452, + "grad_norm": 2.9832215309143066, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8913009166717529, + "num_tokens": 683778257.0, + "step": 17614 + }, + { + "epoch": 2.2408090573718358, + "grad_norm": 2.7242937088012695, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.901685357093811, + "num_tokens": 683821149.0, + "step": 17615 + }, + { + "epoch": 2.2409362676504263, + "grad_norm": 2.956799268722534, + "learning_rate": 1e-06, + "loss": 0.3091, + "mean_token_accuracy": 0.8900259733200073, + "num_tokens": 683858840.0, + "step": 17616 + }, + { + "epoch": 2.241063477929017, + "grad_norm": 2.786839485168457, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8959441781044006, + "num_tokens": 683904481.0, + "step": 17617 + }, + { + "epoch": 2.2411906882076074, + "grad_norm": 2.772766351699829, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9035748243331909, + "num_tokens": 683944981.0, + "step": 17618 + }, + { + "epoch": 2.241317898486198, + "grad_norm": 2.779679775238037, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9002081751823425, + "num_tokens": 683984594.0, + "step": 17619 + }, + { + "epoch": 2.241445108764788, + "grad_norm": 2.854174852371216, + "learning_rate": 1e-06, + "loss": 0.213, + "mean_token_accuracy": 0.9204763770103455, + "num_tokens": 684020786.0, + "step": 17620 + }, + { + "epoch": 2.2415723190433785, + "grad_norm": 2.9110286235809326, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9102320671081543, + "num_tokens": 684056766.0, + "step": 17621 + }, + { + "epoch": 2.241699529321969, + "grad_norm": 2.7288095951080322, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9046053886413574, + "num_tokens": 684098031.0, + "step": 17622 + }, + { + "epoch": 2.2418267396005596, + "grad_norm": 3.0449047088623047, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8942087888717651, + "num_tokens": 684135726.0, + "step": 17623 + }, + { + "epoch": 2.24195394987915, + "grad_norm": 2.882148504257202, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9040790796279907, + "num_tokens": 684171884.0, + "step": 17624 + }, + { + "epoch": 2.2420811601577406, + "grad_norm": 3.212103843688965, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9086143970489502, + "num_tokens": 684200476.0, + "step": 17625 + }, + { + "epoch": 2.242208370436331, + "grad_norm": 2.785566806793213, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9011044502258301, + "num_tokens": 684241677.0, + "step": 17626 + }, + { + "epoch": 2.2423355807149217, + "grad_norm": 2.5659584999084473, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9099175930023193, + "num_tokens": 684287745.0, + "step": 17627 + }, + { + "epoch": 2.242462790993512, + "grad_norm": 2.810343027114868, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9158139824867249, + "num_tokens": 684322612.0, + "step": 17628 + }, + { + "epoch": 2.2425900012721027, + "grad_norm": 2.832691192626953, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9022693634033203, + "num_tokens": 684362389.0, + "step": 17629 + }, + { + "epoch": 2.2427172115506933, + "grad_norm": 2.955475091934204, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9089432954788208, + "num_tokens": 684397460.0, + "step": 17630 + }, + { + "epoch": 2.242844421829284, + "grad_norm": 2.7911739349365234, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9078167676925659, + "num_tokens": 684435839.0, + "step": 17631 + }, + { + "epoch": 2.2429716321078743, + "grad_norm": 3.1236109733581543, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8945955038070679, + "num_tokens": 684474570.0, + "step": 17632 + }, + { + "epoch": 2.243098842386465, + "grad_norm": 2.712458610534668, + "learning_rate": 1e-06, + "loss": 0.2274, + "mean_token_accuracy": 0.913619875907898, + "num_tokens": 684513757.0, + "step": 17633 + }, + { + "epoch": 2.2432260526650554, + "grad_norm": 2.9984374046325684, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.920685887336731, + "num_tokens": 684546723.0, + "step": 17634 + }, + { + "epoch": 2.243353262943646, + "grad_norm": 2.5979349613189697, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9083731174468994, + "num_tokens": 684590290.0, + "step": 17635 + }, + { + "epoch": 2.2434804732222364, + "grad_norm": 2.919301748275757, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9087396860122681, + "num_tokens": 684631123.0, + "step": 17636 + }, + { + "epoch": 2.243607683500827, + "grad_norm": 3.3240787982940674, + "learning_rate": 1e-06, + "loss": 0.3238, + "mean_token_accuracy": 0.8829355239868164, + "num_tokens": 684670742.0, + "step": 17637 + }, + { + "epoch": 2.2437348937794175, + "grad_norm": 2.8497934341430664, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.8995879292488098, + "num_tokens": 684712624.0, + "step": 17638 + }, + { + "epoch": 2.243862104058008, + "grad_norm": 3.1101036071777344, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.8964825868606567, + "num_tokens": 684746365.0, + "step": 17639 + }, + { + "epoch": 2.2439893143365985, + "grad_norm": 2.983389377593994, + "learning_rate": 1e-06, + "loss": 0.231, + "mean_token_accuracy": 0.9128111600875854, + "num_tokens": 684781376.0, + "step": 17640 + }, + { + "epoch": 2.244116524615189, + "grad_norm": 2.6639161109924316, + "learning_rate": 1e-06, + "loss": 0.2318, + "mean_token_accuracy": 0.9113901257514954, + "num_tokens": 684822790.0, + "step": 17641 + }, + { + "epoch": 2.2442437348937796, + "grad_norm": 3.1277849674224854, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.8985202312469482, + "num_tokens": 684858682.0, + "step": 17642 + }, + { + "epoch": 2.24437094517237, + "grad_norm": 3.1423709392547607, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8992626070976257, + "num_tokens": 684897113.0, + "step": 17643 + }, + { + "epoch": 2.2444981554509607, + "grad_norm": 2.8553128242492676, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9024945497512817, + "num_tokens": 684934741.0, + "step": 17644 + }, + { + "epoch": 2.2446253657295507, + "grad_norm": 3.1904733180999756, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8898404240608215, + "num_tokens": 684969900.0, + "step": 17645 + }, + { + "epoch": 2.2447525760081413, + "grad_norm": 2.6733181476593018, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9008834362030029, + "num_tokens": 685014350.0, + "step": 17646 + }, + { + "epoch": 2.244879786286732, + "grad_norm": 2.4965174198150635, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9101526737213135, + "num_tokens": 685060957.0, + "step": 17647 + }, + { + "epoch": 2.2450069965653223, + "grad_norm": 3.079712390899658, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8941501379013062, + "num_tokens": 685095543.0, + "step": 17648 + }, + { + "epoch": 2.245134206843913, + "grad_norm": 2.7254583835601807, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9111474752426147, + "num_tokens": 685135404.0, + "step": 17649 + }, + { + "epoch": 2.2452614171225034, + "grad_norm": 3.0296220779418945, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9081916213035583, + "num_tokens": 685167711.0, + "step": 17650 + }, + { + "epoch": 2.245388627401094, + "grad_norm": 3.0551469326019287, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8969175815582275, + "num_tokens": 685206281.0, + "step": 17651 + }, + { + "epoch": 2.2455158376796844, + "grad_norm": 3.246326446533203, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8915826678276062, + "num_tokens": 685240708.0, + "step": 17652 + }, + { + "epoch": 2.245643047958275, + "grad_norm": 2.621551275253296, + "learning_rate": 1e-06, + "loss": 0.2197, + "mean_token_accuracy": 0.9194977879524231, + "num_tokens": 685279802.0, + "step": 17653 + }, + { + "epoch": 2.2457702582368655, + "grad_norm": 3.1172914505004883, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8934216499328613, + "num_tokens": 685316821.0, + "step": 17654 + }, + { + "epoch": 2.245897468515456, + "grad_norm": 2.9359896183013916, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9032936096191406, + "num_tokens": 685354756.0, + "step": 17655 + }, + { + "epoch": 2.2460246787940465, + "grad_norm": 3.1686389446258545, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9068862199783325, + "num_tokens": 685391782.0, + "step": 17656 + }, + { + "epoch": 2.246151889072637, + "grad_norm": 2.8958852291107178, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9083850383758545, + "num_tokens": 685430682.0, + "step": 17657 + }, + { + "epoch": 2.2462790993512276, + "grad_norm": 2.706932783126831, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9096236228942871, + "num_tokens": 685473077.0, + "step": 17658 + }, + { + "epoch": 2.246406309629818, + "grad_norm": 9.08894157409668, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9081416130065918, + "num_tokens": 685517161.0, + "step": 17659 + }, + { + "epoch": 2.2465335199084087, + "grad_norm": 2.9269654750823975, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9127256870269775, + "num_tokens": 685553157.0, + "step": 17660 + }, + { + "epoch": 2.246660730186999, + "grad_norm": 2.707397222518921, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9106047749519348, + "num_tokens": 685595595.0, + "step": 17661 + }, + { + "epoch": 2.2467879404655897, + "grad_norm": 2.8303661346435547, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9031189680099487, + "num_tokens": 685639743.0, + "step": 17662 + }, + { + "epoch": 2.2469151507441802, + "grad_norm": 2.965071439743042, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9106547832489014, + "num_tokens": 685679499.0, + "step": 17663 + }, + { + "epoch": 2.2470423610227708, + "grad_norm": 2.7865021228790283, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9089870452880859, + "num_tokens": 685722038.0, + "step": 17664 + }, + { + "epoch": 2.2471695713013613, + "grad_norm": 3.1809206008911133, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8888572454452515, + "num_tokens": 685760936.0, + "step": 17665 + }, + { + "epoch": 2.247296781579952, + "grad_norm": 3.333984136581421, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.889998197555542, + "num_tokens": 685795196.0, + "step": 17666 + }, + { + "epoch": 2.2474239918585424, + "grad_norm": 2.6401052474975586, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9101362824440002, + "num_tokens": 685841727.0, + "step": 17667 + }, + { + "epoch": 2.247551202137133, + "grad_norm": 3.356006622314453, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8993891477584839, + "num_tokens": 685878138.0, + "step": 17668 + }, + { + "epoch": 2.2476784124157234, + "grad_norm": 2.8031013011932373, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9073063135147095, + "num_tokens": 685917343.0, + "step": 17669 + }, + { + "epoch": 2.2478056226943135, + "grad_norm": 2.836737871170044, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9084494113922119, + "num_tokens": 685954076.0, + "step": 17670 + }, + { + "epoch": 2.247932832972904, + "grad_norm": 2.7967848777770996, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9136337041854858, + "num_tokens": 685992367.0, + "step": 17671 + }, + { + "epoch": 2.2480600432514946, + "grad_norm": 2.961198329925537, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9018076658248901, + "num_tokens": 686027864.0, + "step": 17672 + }, + { + "epoch": 2.248187253530085, + "grad_norm": 3.1834475994110107, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.8973392248153687, + "num_tokens": 686059584.0, + "step": 17673 + }, + { + "epoch": 2.2483144638086756, + "grad_norm": 2.7025394439697266, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9059157371520996, + "num_tokens": 686101892.0, + "step": 17674 + }, + { + "epoch": 2.248441674087266, + "grad_norm": 2.8939507007598877, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.9124182462692261, + "num_tokens": 686138271.0, + "step": 17675 + }, + { + "epoch": 2.2485688843658567, + "grad_norm": 2.822723627090454, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.9093561172485352, + "num_tokens": 686174207.0, + "step": 17676 + }, + { + "epoch": 2.248696094644447, + "grad_norm": 2.7823410034179688, + "learning_rate": 1e-06, + "loss": 0.2445, + "mean_token_accuracy": 0.9102895259857178, + "num_tokens": 686213627.0, + "step": 17677 + }, + { + "epoch": 2.2488233049230377, + "grad_norm": 2.8469300270080566, + "learning_rate": 1e-06, + "loss": 0.2232, + "mean_token_accuracy": 0.9156315326690674, + "num_tokens": 686253197.0, + "step": 17678 + }, + { + "epoch": 2.2489505152016283, + "grad_norm": 3.23286509513855, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8940531015396118, + "num_tokens": 686289431.0, + "step": 17679 + }, + { + "epoch": 2.249077725480219, + "grad_norm": 2.7158045768737793, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9070646166801453, + "num_tokens": 686334612.0, + "step": 17680 + }, + { + "epoch": 2.2492049357588093, + "grad_norm": 2.887040376663208, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9107869863510132, + "num_tokens": 686370268.0, + "step": 17681 + }, + { + "epoch": 2.2493321460374, + "grad_norm": 2.9610185623168945, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9108161926269531, + "num_tokens": 686409947.0, + "step": 17682 + }, + { + "epoch": 2.2494593563159904, + "grad_norm": 2.978214740753174, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9134361147880554, + "num_tokens": 686445385.0, + "step": 17683 + }, + { + "epoch": 2.249586566594581, + "grad_norm": 2.848397731781006, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9003306031227112, + "num_tokens": 686486308.0, + "step": 17684 + }, + { + "epoch": 2.2497137768731714, + "grad_norm": 2.876768112182617, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9031919836997986, + "num_tokens": 686528648.0, + "step": 17685 + }, + { + "epoch": 2.249840987151762, + "grad_norm": 3.009003162384033, + "learning_rate": 1e-06, + "loss": 0.2351, + "mean_token_accuracy": 0.91327965259552, + "num_tokens": 686561168.0, + "step": 17686 + }, + { + "epoch": 2.2499681974303525, + "grad_norm": 2.4365057945251465, + "learning_rate": 1e-06, + "loss": 0.2185, + "mean_token_accuracy": 0.9194901585578918, + "num_tokens": 686607444.0, + "step": 17687 + }, + { + "epoch": 2.250095407708943, + "grad_norm": 2.7642836570739746, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.911091148853302, + "num_tokens": 686645960.0, + "step": 17688 + }, + { + "epoch": 2.2502226179875335, + "grad_norm": 2.8076188564300537, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.8991946578025818, + "num_tokens": 686690924.0, + "step": 17689 + }, + { + "epoch": 2.250349828266124, + "grad_norm": 2.8583967685699463, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9103107452392578, + "num_tokens": 686731120.0, + "step": 17690 + }, + { + "epoch": 2.2504770385447146, + "grad_norm": 2.6495351791381836, + "learning_rate": 1e-06, + "loss": 0.2009, + "mean_token_accuracy": 0.924567699432373, + "num_tokens": 686767714.0, + "step": 17691 + }, + { + "epoch": 2.250604248823305, + "grad_norm": 2.839012861251831, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9058919548988342, + "num_tokens": 686808590.0, + "step": 17692 + }, + { + "epoch": 2.250731459101895, + "grad_norm": 2.7238247394561768, + "learning_rate": 1e-06, + "loss": 0.2289, + "mean_token_accuracy": 0.913063645362854, + "num_tokens": 686848067.0, + "step": 17693 + }, + { + "epoch": 2.250858669380486, + "grad_norm": 3.153174877166748, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8964793682098389, + "num_tokens": 686885054.0, + "step": 17694 + }, + { + "epoch": 2.2509858796590763, + "grad_norm": 2.977717399597168, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9050371050834656, + "num_tokens": 686922187.0, + "step": 17695 + }, + { + "epoch": 2.251113089937667, + "grad_norm": 2.9521453380584717, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9069832563400269, + "num_tokens": 686962426.0, + "step": 17696 + }, + { + "epoch": 2.2512403002162573, + "grad_norm": 2.9564356803894043, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.8986420631408691, + "num_tokens": 686996842.0, + "step": 17697 + }, + { + "epoch": 2.251367510494848, + "grad_norm": 3.250814914703369, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8958125710487366, + "num_tokens": 687032901.0, + "step": 17698 + }, + { + "epoch": 2.2514947207734384, + "grad_norm": 2.730163812637329, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.906965970993042, + "num_tokens": 687072892.0, + "step": 17699 + }, + { + "epoch": 2.251621931052029, + "grad_norm": 2.9563982486724854, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9063879251480103, + "num_tokens": 687109243.0, + "step": 17700 + }, + { + "epoch": 2.2517491413306194, + "grad_norm": 2.8116214275360107, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9052064418792725, + "num_tokens": 687148554.0, + "step": 17701 + }, + { + "epoch": 2.25187635160921, + "grad_norm": 2.961916446685791, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9006937146186829, + "num_tokens": 687185922.0, + "step": 17702 + }, + { + "epoch": 2.2520035618878005, + "grad_norm": 2.6731762886047363, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9079390168190002, + "num_tokens": 687229368.0, + "step": 17703 + }, + { + "epoch": 2.252130772166391, + "grad_norm": 2.709088087081909, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9179363250732422, + "num_tokens": 687268019.0, + "step": 17704 + }, + { + "epoch": 2.2522579824449815, + "grad_norm": 3.0371193885803223, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9037810564041138, + "num_tokens": 687304930.0, + "step": 17705 + }, + { + "epoch": 2.252385192723572, + "grad_norm": 2.7982048988342285, + "learning_rate": 1e-06, + "loss": 0.2108, + "mean_token_accuracy": 0.9210143089294434, + "num_tokens": 687339330.0, + "step": 17706 + }, + { + "epoch": 2.2525124030021626, + "grad_norm": 3.045607805252075, + "learning_rate": 1e-06, + "loss": 0.3132, + "mean_token_accuracy": 0.8865469694137573, + "num_tokens": 687375731.0, + "step": 17707 + }, + { + "epoch": 2.252639613280753, + "grad_norm": 2.685561418533325, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.9141273498535156, + "num_tokens": 687412943.0, + "step": 17708 + }, + { + "epoch": 2.2527668235593437, + "grad_norm": 3.222337484359741, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8918644785881042, + "num_tokens": 687446200.0, + "step": 17709 + }, + { + "epoch": 2.252894033837934, + "grad_norm": 2.9096245765686035, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.8995024561882019, + "num_tokens": 687485602.0, + "step": 17710 + }, + { + "epoch": 2.2530212441165247, + "grad_norm": 3.144568920135498, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9017340540885925, + "num_tokens": 687518294.0, + "step": 17711 + }, + { + "epoch": 2.2531484543951152, + "grad_norm": 2.8311002254486084, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9096227288246155, + "num_tokens": 687556519.0, + "step": 17712 + }, + { + "epoch": 2.2532756646737058, + "grad_norm": 3.0175421237945557, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9017194509506226, + "num_tokens": 687597082.0, + "step": 17713 + }, + { + "epoch": 2.2534028749522963, + "grad_norm": 2.984323501586914, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8960959911346436, + "num_tokens": 687637595.0, + "step": 17714 + }, + { + "epoch": 2.253530085230887, + "grad_norm": 2.7232723236083984, + "learning_rate": 1e-06, + "loss": 0.2019, + "mean_token_accuracy": 0.9228041172027588, + "num_tokens": 687668647.0, + "step": 17715 + }, + { + "epoch": 2.2536572955094774, + "grad_norm": 2.4834229946136475, + "learning_rate": 1e-06, + "loss": 0.2305, + "mean_token_accuracy": 0.9154401421546936, + "num_tokens": 687714697.0, + "step": 17716 + }, + { + "epoch": 2.253784505788068, + "grad_norm": 3.084772825241089, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.9006520509719849, + "num_tokens": 687752496.0, + "step": 17717 + }, + { + "epoch": 2.253911716066658, + "grad_norm": 3.0408244132995605, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9055052995681763, + "num_tokens": 687791012.0, + "step": 17718 + }, + { + "epoch": 2.254038926345249, + "grad_norm": 2.7653114795684814, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.9122298359870911, + "num_tokens": 687831470.0, + "step": 17719 + }, + { + "epoch": 2.254166136623839, + "grad_norm": 3.002349376678467, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.904149055480957, + "num_tokens": 687866679.0, + "step": 17720 + }, + { + "epoch": 2.2542933469024296, + "grad_norm": 2.618178367614746, + "learning_rate": 1e-06, + "loss": 0.2325, + "mean_token_accuracy": 0.9172855615615845, + "num_tokens": 687905067.0, + "step": 17721 + }, + { + "epoch": 2.25442055718102, + "grad_norm": 2.644036293029785, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9076961278915405, + "num_tokens": 687946430.0, + "step": 17722 + }, + { + "epoch": 2.2545477674596106, + "grad_norm": 2.9572393894195557, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.901350200176239, + "num_tokens": 687985395.0, + "step": 17723 + }, + { + "epoch": 2.254674977738201, + "grad_norm": 3.309098482131958, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8972569704055786, + "num_tokens": 688022534.0, + "step": 17724 + }, + { + "epoch": 2.2548021880167917, + "grad_norm": 2.987504005432129, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9070571660995483, + "num_tokens": 688057679.0, + "step": 17725 + }, + { + "epoch": 2.254929398295382, + "grad_norm": 3.2911338806152344, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9030688405036926, + "num_tokens": 688088202.0, + "step": 17726 + }, + { + "epoch": 2.2550566085739727, + "grad_norm": 2.9789774417877197, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9045690298080444, + "num_tokens": 688125392.0, + "step": 17727 + }, + { + "epoch": 2.2551838188525632, + "grad_norm": 2.7122554779052734, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.903446614742279, + "num_tokens": 688166589.0, + "step": 17728 + }, + { + "epoch": 2.2553110291311538, + "grad_norm": 3.2579216957092285, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8973572850227356, + "num_tokens": 688198847.0, + "step": 17729 + }, + { + "epoch": 2.2554382394097443, + "grad_norm": 2.9337968826293945, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9107879400253296, + "num_tokens": 688233360.0, + "step": 17730 + }, + { + "epoch": 2.255565449688335, + "grad_norm": 2.6412668228149414, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9026444554328918, + "num_tokens": 688281447.0, + "step": 17731 + }, + { + "epoch": 2.2556926599669254, + "grad_norm": 2.963383674621582, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.896865725517273, + "num_tokens": 688323093.0, + "step": 17732 + }, + { + "epoch": 2.255819870245516, + "grad_norm": 2.6989424228668213, + "learning_rate": 1e-06, + "loss": 0.2233, + "mean_token_accuracy": 0.9174853563308716, + "num_tokens": 688360327.0, + "step": 17733 + }, + { + "epoch": 2.2559470805241064, + "grad_norm": 2.76428484916687, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8996922373771667, + "num_tokens": 688407429.0, + "step": 17734 + }, + { + "epoch": 2.256074290802697, + "grad_norm": 2.7827775478363037, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9079809188842773, + "num_tokens": 688445578.0, + "step": 17735 + }, + { + "epoch": 2.2562015010812875, + "grad_norm": 3.087369918823242, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.902153491973877, + "num_tokens": 688482490.0, + "step": 17736 + }, + { + "epoch": 2.256328711359878, + "grad_norm": 2.9492738246917725, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9018195271492004, + "num_tokens": 688520326.0, + "step": 17737 + }, + { + "epoch": 2.2564559216384685, + "grad_norm": 3.0414862632751465, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8865450620651245, + "num_tokens": 688561167.0, + "step": 17738 + }, + { + "epoch": 2.256583131917059, + "grad_norm": 2.831516742706299, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9062919020652771, + "num_tokens": 688599617.0, + "step": 17739 + }, + { + "epoch": 2.2567103421956496, + "grad_norm": 2.650103807449341, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9082210063934326, + "num_tokens": 688642940.0, + "step": 17740 + }, + { + "epoch": 2.2568375524742397, + "grad_norm": 2.599151134490967, + "learning_rate": 1e-06, + "loss": 0.2294, + "mean_token_accuracy": 0.9162846803665161, + "num_tokens": 688685903.0, + "step": 17741 + }, + { + "epoch": 2.2569647627528306, + "grad_norm": 2.9059247970581055, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.8997116684913635, + "num_tokens": 688722999.0, + "step": 17742 + }, + { + "epoch": 2.2570919730314207, + "grad_norm": 2.8777740001678467, + "learning_rate": 1e-06, + "loss": 0.2324, + "mean_token_accuracy": 0.9142879843711853, + "num_tokens": 688759783.0, + "step": 17743 + }, + { + "epoch": 2.2572191833100113, + "grad_norm": 2.935741662979126, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8951118588447571, + "num_tokens": 688798065.0, + "step": 17744 + }, + { + "epoch": 2.257346393588602, + "grad_norm": 3.0471837520599365, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9096139669418335, + "num_tokens": 688830716.0, + "step": 17745 + }, + { + "epoch": 2.2574736038671923, + "grad_norm": 3.0202505588531494, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8932932615280151, + "num_tokens": 688870364.0, + "step": 17746 + }, + { + "epoch": 2.257600814145783, + "grad_norm": 2.838203191757202, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9083728790283203, + "num_tokens": 688907694.0, + "step": 17747 + }, + { + "epoch": 2.2577280244243734, + "grad_norm": 2.8624281883239746, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9000442028045654, + "num_tokens": 688947266.0, + "step": 17748 + }, + { + "epoch": 2.257855234702964, + "grad_norm": 2.9951303005218506, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.900708794593811, + "num_tokens": 688983921.0, + "step": 17749 + }, + { + "epoch": 2.2579824449815544, + "grad_norm": 3.026427984237671, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9072245359420776, + "num_tokens": 689017617.0, + "step": 17750 + }, + { + "epoch": 2.258109655260145, + "grad_norm": 2.6884021759033203, + "learning_rate": 1e-06, + "loss": 0.2422, + "mean_token_accuracy": 0.9098411798477173, + "num_tokens": 689057054.0, + "step": 17751 + }, + { + "epoch": 2.2582368655387355, + "grad_norm": 2.8521482944488525, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9159111380577087, + "num_tokens": 689094544.0, + "step": 17752 + }, + { + "epoch": 2.258364075817326, + "grad_norm": 2.498180866241455, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.9137930274009705, + "num_tokens": 689143360.0, + "step": 17753 + }, + { + "epoch": 2.2584912860959165, + "grad_norm": 2.9434478282928467, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8967012166976929, + "num_tokens": 689183755.0, + "step": 17754 + }, + { + "epoch": 2.258618496374507, + "grad_norm": 2.937915325164795, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9100110530853271, + "num_tokens": 689220998.0, + "step": 17755 + }, + { + "epoch": 2.2587457066530976, + "grad_norm": 2.802708625793457, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9079887866973877, + "num_tokens": 689259486.0, + "step": 17756 + }, + { + "epoch": 2.258872916931688, + "grad_norm": 2.9294309616088867, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9023099541664124, + "num_tokens": 689299911.0, + "step": 17757 + }, + { + "epoch": 2.2590001272102787, + "grad_norm": 3.055474042892456, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.895017147064209, + "num_tokens": 689337026.0, + "step": 17758 + }, + { + "epoch": 2.259127337488869, + "grad_norm": 2.7789297103881836, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9056569933891296, + "num_tokens": 689377903.0, + "step": 17759 + }, + { + "epoch": 2.2592545477674597, + "grad_norm": 2.893589735031128, + "learning_rate": 1e-06, + "loss": 0.2164, + "mean_token_accuracy": 0.9169124364852905, + "num_tokens": 689412381.0, + "step": 17760 + }, + { + "epoch": 2.2593817580460502, + "grad_norm": 2.7685906887054443, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9037742018699646, + "num_tokens": 689454773.0, + "step": 17761 + }, + { + "epoch": 2.2595089683246408, + "grad_norm": 2.784153938293457, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9041455984115601, + "num_tokens": 689495601.0, + "step": 17762 + }, + { + "epoch": 2.2596361786032313, + "grad_norm": 2.639021158218384, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9079829454421997, + "num_tokens": 689540656.0, + "step": 17763 + }, + { + "epoch": 2.259763388881822, + "grad_norm": 2.7024433612823486, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.91191565990448, + "num_tokens": 689582726.0, + "step": 17764 + }, + { + "epoch": 2.2598905991604123, + "grad_norm": 2.7096989154815674, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9089643955230713, + "num_tokens": 689623212.0, + "step": 17765 + }, + { + "epoch": 2.2600178094390024, + "grad_norm": 3.0120954513549805, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9076201319694519, + "num_tokens": 689658117.0, + "step": 17766 + }, + { + "epoch": 2.2601450197175934, + "grad_norm": 2.7856993675231934, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9096150398254395, + "num_tokens": 689696065.0, + "step": 17767 + }, + { + "epoch": 2.2602722299961835, + "grad_norm": 2.8332157135009766, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.904949426651001, + "num_tokens": 689737004.0, + "step": 17768 + }, + { + "epoch": 2.260399440274774, + "grad_norm": 2.8426947593688965, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9053264856338501, + "num_tokens": 689774267.0, + "step": 17769 + }, + { + "epoch": 2.2605266505533645, + "grad_norm": 3.0360755920410156, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9083012342453003, + "num_tokens": 689810620.0, + "step": 17770 + }, + { + "epoch": 2.260653860831955, + "grad_norm": 2.8488826751708984, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9087581038475037, + "num_tokens": 689848867.0, + "step": 17771 + }, + { + "epoch": 2.2607810711105456, + "grad_norm": 3.064147710800171, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8991044759750366, + "num_tokens": 689885105.0, + "step": 17772 + }, + { + "epoch": 2.260908281389136, + "grad_norm": 2.9753172397613525, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8975552320480347, + "num_tokens": 689925289.0, + "step": 17773 + }, + { + "epoch": 2.2610354916677267, + "grad_norm": 2.828582763671875, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9091286063194275, + "num_tokens": 689963953.0, + "step": 17774 + }, + { + "epoch": 2.261162701946317, + "grad_norm": 2.541062355041504, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9067971110343933, + "num_tokens": 690009525.0, + "step": 17775 + }, + { + "epoch": 2.2612899122249077, + "grad_norm": 2.7477426528930664, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9039554595947266, + "num_tokens": 690049978.0, + "step": 17776 + }, + { + "epoch": 2.2614171225034982, + "grad_norm": 2.7785418033599854, + "learning_rate": 1e-06, + "loss": 0.2215, + "mean_token_accuracy": 0.916534423828125, + "num_tokens": 690087100.0, + "step": 17777 + }, + { + "epoch": 2.2615443327820888, + "grad_norm": 2.937505006790161, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9052793383598328, + "num_tokens": 690128644.0, + "step": 17778 + }, + { + "epoch": 2.2616715430606793, + "grad_norm": 2.797538995742798, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9046280980110168, + "num_tokens": 690171658.0, + "step": 17779 + }, + { + "epoch": 2.26179875333927, + "grad_norm": 2.9341790676116943, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9113070964813232, + "num_tokens": 690211542.0, + "step": 17780 + }, + { + "epoch": 2.2619259636178604, + "grad_norm": 2.798776388168335, + "learning_rate": 1e-06, + "loss": 0.2284, + "mean_token_accuracy": 0.9146172404289246, + "num_tokens": 690246300.0, + "step": 17781 + }, + { + "epoch": 2.262053173896451, + "grad_norm": 2.784097909927368, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.89156174659729, + "num_tokens": 690292641.0, + "step": 17782 + }, + { + "epoch": 2.2621803841750414, + "grad_norm": 2.708569049835205, + "learning_rate": 1e-06, + "loss": 0.2303, + "mean_token_accuracy": 0.9152141213417053, + "num_tokens": 690331437.0, + "step": 17783 + }, + { + "epoch": 2.262307594453632, + "grad_norm": 2.791395425796509, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9127398133277893, + "num_tokens": 690373249.0, + "step": 17784 + }, + { + "epoch": 2.2624348047322225, + "grad_norm": 2.6044974327087402, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.8986666798591614, + "num_tokens": 690420243.0, + "step": 17785 + }, + { + "epoch": 2.262562015010813, + "grad_norm": 2.8868446350097656, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9111497402191162, + "num_tokens": 690459534.0, + "step": 17786 + }, + { + "epoch": 2.2626892252894035, + "grad_norm": 2.7547011375427246, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9000426530838013, + "num_tokens": 690505785.0, + "step": 17787 + }, + { + "epoch": 2.262816435567994, + "grad_norm": 3.402141571044922, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.900266706943512, + "num_tokens": 690538734.0, + "step": 17788 + }, + { + "epoch": 2.2629436458465846, + "grad_norm": 3.0090982913970947, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8978881239891052, + "num_tokens": 690577397.0, + "step": 17789 + }, + { + "epoch": 2.263070856125175, + "grad_norm": 2.9645237922668457, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8955950736999512, + "num_tokens": 690614857.0, + "step": 17790 + }, + { + "epoch": 2.263198066403765, + "grad_norm": 2.8320045471191406, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9040738344192505, + "num_tokens": 690656647.0, + "step": 17791 + }, + { + "epoch": 2.263325276682356, + "grad_norm": 2.5907833576202393, + "learning_rate": 1e-06, + "loss": 0.2224, + "mean_token_accuracy": 0.9152934551239014, + "num_tokens": 690696922.0, + "step": 17792 + }, + { + "epoch": 2.2634524869609463, + "grad_norm": 2.8962414264678955, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9063113927841187, + "num_tokens": 690734875.0, + "step": 17793 + }, + { + "epoch": 2.263579697239537, + "grad_norm": 3.0473666191101074, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9068308472633362, + "num_tokens": 690771116.0, + "step": 17794 + }, + { + "epoch": 2.2637069075181273, + "grad_norm": 2.8171658515930176, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9070969820022583, + "num_tokens": 690809950.0, + "step": 17795 + }, + { + "epoch": 2.263834117796718, + "grad_norm": 2.746913194656372, + "learning_rate": 1e-06, + "loss": 0.2282, + "mean_token_accuracy": 0.914318323135376, + "num_tokens": 690849198.0, + "step": 17796 + }, + { + "epoch": 2.2639613280753084, + "grad_norm": 3.0263516902923584, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9035900831222534, + "num_tokens": 690886775.0, + "step": 17797 + }, + { + "epoch": 2.264088538353899, + "grad_norm": 3.114112377166748, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8951514363288879, + "num_tokens": 690926328.0, + "step": 17798 + }, + { + "epoch": 2.2642157486324894, + "grad_norm": 2.808516025543213, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9073393940925598, + "num_tokens": 690965990.0, + "step": 17799 + }, + { + "epoch": 2.26434295891108, + "grad_norm": 2.826235055923462, + "learning_rate": 1e-06, + "loss": 0.2116, + "mean_token_accuracy": 0.9199011325836182, + "num_tokens": 691000325.0, + "step": 17800 + }, + { + "epoch": 2.2644701691896705, + "grad_norm": 3.035057783126831, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9035831689834595, + "num_tokens": 691040040.0, + "step": 17801 + }, + { + "epoch": 2.264597379468261, + "grad_norm": 2.995729684829712, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9037772417068481, + "num_tokens": 691074741.0, + "step": 17802 + }, + { + "epoch": 2.2647245897468515, + "grad_norm": 2.6952288150787354, + "learning_rate": 1e-06, + "loss": 0.2218, + "mean_token_accuracy": 0.9163181781768799, + "num_tokens": 691113691.0, + "step": 17803 + }, + { + "epoch": 2.264851800025442, + "grad_norm": 2.9062018394470215, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9096953868865967, + "num_tokens": 691153483.0, + "step": 17804 + }, + { + "epoch": 2.2649790103040326, + "grad_norm": 3.0248351097106934, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8990450501441956, + "num_tokens": 691192678.0, + "step": 17805 + }, + { + "epoch": 2.265106220582623, + "grad_norm": 2.5249764919281006, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9084174633026123, + "num_tokens": 691239715.0, + "step": 17806 + }, + { + "epoch": 2.2652334308612136, + "grad_norm": 2.686267375946045, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9021695256233215, + "num_tokens": 691280354.0, + "step": 17807 + }, + { + "epoch": 2.265360641139804, + "grad_norm": 2.9395828247070312, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8924013376235962, + "num_tokens": 691320516.0, + "step": 17808 + }, + { + "epoch": 2.2654878514183947, + "grad_norm": 2.800532817840576, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9096450805664062, + "num_tokens": 691358943.0, + "step": 17809 + }, + { + "epoch": 2.2656150616969852, + "grad_norm": 2.7944767475128174, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9089756011962891, + "num_tokens": 691397427.0, + "step": 17810 + }, + { + "epoch": 2.2657422719755758, + "grad_norm": 2.7900454998016357, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8896927833557129, + "num_tokens": 691439977.0, + "step": 17811 + }, + { + "epoch": 2.2658694822541663, + "grad_norm": 2.901385545730591, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.900384247303009, + "num_tokens": 691479315.0, + "step": 17812 + }, + { + "epoch": 2.265996692532757, + "grad_norm": 2.9185495376586914, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9065836668014526, + "num_tokens": 691517729.0, + "step": 17813 + }, + { + "epoch": 2.2661239028113473, + "grad_norm": 2.790135383605957, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9147113561630249, + "num_tokens": 691555882.0, + "step": 17814 + }, + { + "epoch": 2.266251113089938, + "grad_norm": 2.753293514251709, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9038169384002686, + "num_tokens": 691594919.0, + "step": 17815 + }, + { + "epoch": 2.266378323368528, + "grad_norm": 3.2683277130126953, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9120404720306396, + "num_tokens": 691623426.0, + "step": 17816 + }, + { + "epoch": 2.266505533647119, + "grad_norm": 2.851857900619507, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9032626152038574, + "num_tokens": 691666008.0, + "step": 17817 + }, + { + "epoch": 2.266632743925709, + "grad_norm": 3.2584433555603027, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8925150036811829, + "num_tokens": 691702853.0, + "step": 17818 + }, + { + "epoch": 2.2667599542042995, + "grad_norm": 2.9337546825408936, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9054523706436157, + "num_tokens": 691740880.0, + "step": 17819 + }, + { + "epoch": 2.26688716448289, + "grad_norm": 3.0052363872528076, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9032235741615295, + "num_tokens": 691780387.0, + "step": 17820 + }, + { + "epoch": 2.2670143747614806, + "grad_norm": 2.7532496452331543, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9022427201271057, + "num_tokens": 691826153.0, + "step": 17821 + }, + { + "epoch": 2.267141585040071, + "grad_norm": 2.5552542209625244, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.909101128578186, + "num_tokens": 691872861.0, + "step": 17822 + }, + { + "epoch": 2.2672687953186617, + "grad_norm": 2.9480693340301514, + "learning_rate": 1e-06, + "loss": 0.2198, + "mean_token_accuracy": 0.916187047958374, + "num_tokens": 691904471.0, + "step": 17823 + }, + { + "epoch": 2.267396005597252, + "grad_norm": 2.7710442543029785, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9133592247962952, + "num_tokens": 691941122.0, + "step": 17824 + }, + { + "epoch": 2.2675232158758427, + "grad_norm": 3.0349948406219482, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9030038118362427, + "num_tokens": 691979146.0, + "step": 17825 + }, + { + "epoch": 2.2676504261544332, + "grad_norm": 2.7148141860961914, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9082351922988892, + "num_tokens": 692020843.0, + "step": 17826 + }, + { + "epoch": 2.2677776364330238, + "grad_norm": 2.8109207153320312, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.900819718837738, + "num_tokens": 692066294.0, + "step": 17827 + }, + { + "epoch": 2.2679048467116143, + "grad_norm": 2.718513011932373, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8983206748962402, + "num_tokens": 692112947.0, + "step": 17828 + }, + { + "epoch": 2.268032056990205, + "grad_norm": 2.8788328170776367, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9093724489212036, + "num_tokens": 692147130.0, + "step": 17829 + }, + { + "epoch": 2.2681592672687954, + "grad_norm": 2.9529688358306885, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.901713490486145, + "num_tokens": 692183132.0, + "step": 17830 + }, + { + "epoch": 2.268286477547386, + "grad_norm": 3.0715553760528564, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8989747166633606, + "num_tokens": 692223399.0, + "step": 17831 + }, + { + "epoch": 2.2684136878259764, + "grad_norm": 2.9934184551239014, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9105257987976074, + "num_tokens": 692257544.0, + "step": 17832 + }, + { + "epoch": 2.268540898104567, + "grad_norm": 3.057981252670288, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8859080076217651, + "num_tokens": 692295192.0, + "step": 17833 + }, + { + "epoch": 2.2686681083831575, + "grad_norm": 3.00968074798584, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8890169858932495, + "num_tokens": 692335766.0, + "step": 17834 + }, + { + "epoch": 2.268795318661748, + "grad_norm": 2.685744047164917, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9043979048728943, + "num_tokens": 692379838.0, + "step": 17835 + }, + { + "epoch": 2.2689225289403385, + "grad_norm": 2.887051820755005, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9003526568412781, + "num_tokens": 692417655.0, + "step": 17836 + }, + { + "epoch": 2.269049739218929, + "grad_norm": 3.1958889961242676, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.911261796951294, + "num_tokens": 692447737.0, + "step": 17837 + }, + { + "epoch": 2.2691769494975196, + "grad_norm": 3.9853224754333496, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9050767421722412, + "num_tokens": 692486344.0, + "step": 17838 + }, + { + "epoch": 2.2693041597761097, + "grad_norm": 2.691729784011841, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9096144437789917, + "num_tokens": 692526835.0, + "step": 17839 + }, + { + "epoch": 2.2694313700547006, + "grad_norm": 2.922013521194458, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9000920057296753, + "num_tokens": 692567875.0, + "step": 17840 + }, + { + "epoch": 2.2695585803332907, + "grad_norm": 2.8556299209594727, + "learning_rate": 1e-06, + "loss": 0.2303, + "mean_token_accuracy": 0.9162355661392212, + "num_tokens": 692603465.0, + "step": 17841 + }, + { + "epoch": 2.2696857906118812, + "grad_norm": 2.6233644485473633, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9132980108261108, + "num_tokens": 692646407.0, + "step": 17842 + }, + { + "epoch": 2.2698130008904718, + "grad_norm": 2.7271900177001953, + "learning_rate": 1e-06, + "loss": 0.2216, + "mean_token_accuracy": 0.9165734052658081, + "num_tokens": 692683726.0, + "step": 17843 + }, + { + "epoch": 2.2699402111690623, + "grad_norm": 2.6928162574768066, + "learning_rate": 1e-06, + "loss": 0.2161, + "mean_token_accuracy": 0.9204421043395996, + "num_tokens": 692722380.0, + "step": 17844 + }, + { + "epoch": 2.270067421447653, + "grad_norm": 2.808077812194824, + "learning_rate": 1e-06, + "loss": 0.2099, + "mean_token_accuracy": 0.9197667837142944, + "num_tokens": 692756514.0, + "step": 17845 + }, + { + "epoch": 2.2701946317262434, + "grad_norm": 2.933115243911743, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9082911610603333, + "num_tokens": 692794111.0, + "step": 17846 + }, + { + "epoch": 2.270321842004834, + "grad_norm": 2.7516350746154785, + "learning_rate": 1e-06, + "loss": 0.2296, + "mean_token_accuracy": 0.9128307700157166, + "num_tokens": 692830717.0, + "step": 17847 + }, + { + "epoch": 2.2704490522834244, + "grad_norm": 2.928349256515503, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8991725444793701, + "num_tokens": 692870873.0, + "step": 17848 + }, + { + "epoch": 2.270576262562015, + "grad_norm": 3.1144275665283203, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9086562991142273, + "num_tokens": 692904342.0, + "step": 17849 + }, + { + "epoch": 2.2707034728406055, + "grad_norm": 2.809499979019165, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.905828058719635, + "num_tokens": 692945321.0, + "step": 17850 + }, + { + "epoch": 2.270830683119196, + "grad_norm": 2.9578545093536377, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9057891368865967, + "num_tokens": 692982109.0, + "step": 17851 + }, + { + "epoch": 2.2709578933977865, + "grad_norm": 3.009748697280884, + "learning_rate": 1e-06, + "loss": 0.2289, + "mean_token_accuracy": 0.916685938835144, + "num_tokens": 693014027.0, + "step": 17852 + }, + { + "epoch": 2.271085103676377, + "grad_norm": 3.26863169670105, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8991423845291138, + "num_tokens": 693048861.0, + "step": 17853 + }, + { + "epoch": 2.2712123139549676, + "grad_norm": 2.9311296939849854, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9009379744529724, + "num_tokens": 693086306.0, + "step": 17854 + }, + { + "epoch": 2.271339524233558, + "grad_norm": 2.6653265953063965, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9068499803543091, + "num_tokens": 693130129.0, + "step": 17855 + }, + { + "epoch": 2.2714667345121486, + "grad_norm": 2.8381059169769287, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9023352265357971, + "num_tokens": 693169512.0, + "step": 17856 + }, + { + "epoch": 2.271593944790739, + "grad_norm": 2.83659291267395, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9078144431114197, + "num_tokens": 693209435.0, + "step": 17857 + }, + { + "epoch": 2.2717211550693297, + "grad_norm": 2.8925232887268066, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9028674960136414, + "num_tokens": 693251538.0, + "step": 17858 + }, + { + "epoch": 2.2718483653479202, + "grad_norm": 3.026979446411133, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9039994478225708, + "num_tokens": 693287932.0, + "step": 17859 + }, + { + "epoch": 2.2719755756265108, + "grad_norm": 2.9648752212524414, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9108741283416748, + "num_tokens": 693328371.0, + "step": 17860 + }, + { + "epoch": 2.2721027859051013, + "grad_norm": 3.0221188068389893, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.8979954123497009, + "num_tokens": 693363076.0, + "step": 17861 + }, + { + "epoch": 2.272229996183692, + "grad_norm": 2.6270363330841064, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9062919616699219, + "num_tokens": 693406760.0, + "step": 17862 + }, + { + "epoch": 2.2723572064622823, + "grad_norm": 2.6350414752960205, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9068968892097473, + "num_tokens": 693450912.0, + "step": 17863 + }, + { + "epoch": 2.2724844167408724, + "grad_norm": 2.944049835205078, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9092375636100769, + "num_tokens": 693486130.0, + "step": 17864 + }, + { + "epoch": 2.2726116270194634, + "grad_norm": 3.120232343673706, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9108865261077881, + "num_tokens": 693519100.0, + "step": 17865 + }, + { + "epoch": 2.2727388372980535, + "grad_norm": 2.6623945236206055, + "learning_rate": 1e-06, + "loss": 0.2225, + "mean_token_accuracy": 0.9161752462387085, + "num_tokens": 693561046.0, + "step": 17866 + }, + { + "epoch": 2.272866047576644, + "grad_norm": 2.762747049331665, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9076026678085327, + "num_tokens": 693599415.0, + "step": 17867 + }, + { + "epoch": 2.2729932578552345, + "grad_norm": 3.0411107540130615, + "learning_rate": 1e-06, + "loss": 0.3082, + "mean_token_accuracy": 0.8879080414772034, + "num_tokens": 693643495.0, + "step": 17868 + }, + { + "epoch": 2.273120468133825, + "grad_norm": 2.8972580432891846, + "learning_rate": 1e-06, + "loss": 0.2195, + "mean_token_accuracy": 0.9167382717132568, + "num_tokens": 693678765.0, + "step": 17869 + }, + { + "epoch": 2.2732476784124156, + "grad_norm": 3.0786049365997314, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9098834991455078, + "num_tokens": 693711900.0, + "step": 17870 + }, + { + "epoch": 2.273374888691006, + "grad_norm": 3.4113097190856934, + "learning_rate": 1e-06, + "loss": 0.3092, + "mean_token_accuracy": 0.8896288275718689, + "num_tokens": 693743871.0, + "step": 17871 + }, + { + "epoch": 2.2735020989695967, + "grad_norm": 2.9169023036956787, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9051299691200256, + "num_tokens": 693780229.0, + "step": 17872 + }, + { + "epoch": 2.273629309248187, + "grad_norm": 2.5440549850463867, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9088438749313354, + "num_tokens": 693827539.0, + "step": 17873 + }, + { + "epoch": 2.2737565195267777, + "grad_norm": 2.8337340354919434, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9063683748245239, + "num_tokens": 693868617.0, + "step": 17874 + }, + { + "epoch": 2.2738837298053682, + "grad_norm": 2.908560276031494, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9032220840454102, + "num_tokens": 693906246.0, + "step": 17875 + }, + { + "epoch": 2.2740109400839588, + "grad_norm": 3.0785934925079346, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.893807053565979, + "num_tokens": 693946505.0, + "step": 17876 + }, + { + "epoch": 2.2741381503625493, + "grad_norm": 2.6976163387298584, + "learning_rate": 1e-06, + "loss": 0.232, + "mean_token_accuracy": 0.9121938943862915, + "num_tokens": 693984853.0, + "step": 17877 + }, + { + "epoch": 2.27426536064114, + "grad_norm": 2.566143751144409, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9085851907730103, + "num_tokens": 694031324.0, + "step": 17878 + }, + { + "epoch": 2.2743925709197303, + "grad_norm": 2.7688815593719482, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9062299132347107, + "num_tokens": 694073171.0, + "step": 17879 + }, + { + "epoch": 2.274519781198321, + "grad_norm": 2.7703754901885986, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9106378555297852, + "num_tokens": 694113423.0, + "step": 17880 + }, + { + "epoch": 2.2746469914769114, + "grad_norm": 2.7388317584991455, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9092767238616943, + "num_tokens": 694152860.0, + "step": 17881 + }, + { + "epoch": 2.274774201755502, + "grad_norm": 3.120925188064575, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9015616774559021, + "num_tokens": 694188068.0, + "step": 17882 + }, + { + "epoch": 2.2749014120340925, + "grad_norm": 2.8303427696228027, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9088902473449707, + "num_tokens": 694225827.0, + "step": 17883 + }, + { + "epoch": 2.275028622312683, + "grad_norm": 2.758734941482544, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.8991446495056152, + "num_tokens": 694269623.0, + "step": 17884 + }, + { + "epoch": 2.2751558325912735, + "grad_norm": 2.9809470176696777, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.8976289629936218, + "num_tokens": 694309359.0, + "step": 17885 + }, + { + "epoch": 2.275283042869864, + "grad_norm": 3.0698163509368896, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9051820039749146, + "num_tokens": 694344267.0, + "step": 17886 + }, + { + "epoch": 2.2754102531484546, + "grad_norm": 2.685237169265747, + "learning_rate": 1e-06, + "loss": 0.2167, + "mean_token_accuracy": 0.9189781546592712, + "num_tokens": 694381156.0, + "step": 17887 + }, + { + "epoch": 2.275537463427045, + "grad_norm": 2.8987958431243896, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.8984653949737549, + "num_tokens": 694424803.0, + "step": 17888 + }, + { + "epoch": 2.275664673705635, + "grad_norm": 2.7940354347229004, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9025623798370361, + "num_tokens": 694469431.0, + "step": 17889 + }, + { + "epoch": 2.275791883984226, + "grad_norm": 2.7451863288879395, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9165036678314209, + "num_tokens": 694511676.0, + "step": 17890 + }, + { + "epoch": 2.2759190942628162, + "grad_norm": 2.8464479446411133, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9050517678260803, + "num_tokens": 694549793.0, + "step": 17891 + }, + { + "epoch": 2.2760463045414068, + "grad_norm": 2.373713970184326, + "learning_rate": 1e-06, + "loss": 0.2001, + "mean_token_accuracy": 0.9223344326019287, + "num_tokens": 694595206.0, + "step": 17892 + }, + { + "epoch": 2.2761735148199973, + "grad_norm": 2.743199110031128, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9088929295539856, + "num_tokens": 694635436.0, + "step": 17893 + }, + { + "epoch": 2.276300725098588, + "grad_norm": 2.9510326385498047, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9074556231498718, + "num_tokens": 694673652.0, + "step": 17894 + }, + { + "epoch": 2.2764279353771784, + "grad_norm": 3.0145628452301025, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9075683355331421, + "num_tokens": 694710014.0, + "step": 17895 + }, + { + "epoch": 2.276555145655769, + "grad_norm": 3.1152682304382324, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.8998610377311707, + "num_tokens": 694747915.0, + "step": 17896 + }, + { + "epoch": 2.2766823559343594, + "grad_norm": 2.9486324787139893, + "learning_rate": 1e-06, + "loss": 0.2405, + "mean_token_accuracy": 0.9097384214401245, + "num_tokens": 694785365.0, + "step": 17897 + }, + { + "epoch": 2.27680956621295, + "grad_norm": 2.7383008003234863, + "learning_rate": 1e-06, + "loss": 0.2338, + "mean_token_accuracy": 0.9141026139259338, + "num_tokens": 694823198.0, + "step": 17898 + }, + { + "epoch": 2.2769367764915405, + "grad_norm": 2.5344934463500977, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9111793041229248, + "num_tokens": 694866820.0, + "step": 17899 + }, + { + "epoch": 2.277063986770131, + "grad_norm": 2.9652435779571533, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9039133787155151, + "num_tokens": 694902541.0, + "step": 17900 + }, + { + "epoch": 2.2771911970487215, + "grad_norm": 3.019289255142212, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9036424160003662, + "num_tokens": 694938162.0, + "step": 17901 + }, + { + "epoch": 2.277318407327312, + "grad_norm": 2.930842161178589, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8975816965103149, + "num_tokens": 694976034.0, + "step": 17902 + }, + { + "epoch": 2.2774456176059026, + "grad_norm": 3.2872893810272217, + "learning_rate": 1e-06, + "loss": 0.3237, + "mean_token_accuracy": 0.8844344615936279, + "num_tokens": 695013416.0, + "step": 17903 + }, + { + "epoch": 2.277572827884493, + "grad_norm": 2.898138999938965, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8940597772598267, + "num_tokens": 695053715.0, + "step": 17904 + }, + { + "epoch": 2.2777000381630836, + "grad_norm": 2.8004205226898193, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9131033420562744, + "num_tokens": 695089650.0, + "step": 17905 + }, + { + "epoch": 2.277827248441674, + "grad_norm": 2.8219423294067383, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9111140966415405, + "num_tokens": 695129137.0, + "step": 17906 + }, + { + "epoch": 2.2779544587202647, + "grad_norm": 2.8533735275268555, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8952472805976868, + "num_tokens": 695172118.0, + "step": 17907 + }, + { + "epoch": 2.2780816689988552, + "grad_norm": 2.943697214126587, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.9131624698638916, + "num_tokens": 695207697.0, + "step": 17908 + }, + { + "epoch": 2.2782088792774458, + "grad_norm": 2.8253979682922363, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9030461311340332, + "num_tokens": 695247308.0, + "step": 17909 + }, + { + "epoch": 2.2783360895560363, + "grad_norm": 2.7714383602142334, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9096500873565674, + "num_tokens": 695286043.0, + "step": 17910 + }, + { + "epoch": 2.278463299834627, + "grad_norm": 2.7877495288848877, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9023613333702087, + "num_tokens": 695327076.0, + "step": 17911 + }, + { + "epoch": 2.2785905101132173, + "grad_norm": 2.8550262451171875, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9066190719604492, + "num_tokens": 695366426.0, + "step": 17912 + }, + { + "epoch": 2.278717720391808, + "grad_norm": 2.8346974849700928, + "learning_rate": 1e-06, + "loss": 0.2148, + "mean_token_accuracy": 0.9184596538543701, + "num_tokens": 695399361.0, + "step": 17913 + }, + { + "epoch": 2.278844930670398, + "grad_norm": 2.8630993366241455, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9002222418785095, + "num_tokens": 695440513.0, + "step": 17914 + }, + { + "epoch": 2.278972140948989, + "grad_norm": 2.840026378631592, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9061191082000732, + "num_tokens": 695482204.0, + "step": 17915 + }, + { + "epoch": 2.279099351227579, + "grad_norm": 3.090911626815796, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9029582738876343, + "num_tokens": 695515759.0, + "step": 17916 + }, + { + "epoch": 2.2792265615061695, + "grad_norm": 2.5235953330993652, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.9131790995597839, + "num_tokens": 695560606.0, + "step": 17917 + }, + { + "epoch": 2.27935377178476, + "grad_norm": 3.1108224391937256, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8946913480758667, + "num_tokens": 695599258.0, + "step": 17918 + }, + { + "epoch": 2.2794809820633506, + "grad_norm": 2.9481334686279297, + "learning_rate": 1e-06, + "loss": 0.2239, + "mean_token_accuracy": 0.9162276983261108, + "num_tokens": 695629414.0, + "step": 17919 + }, + { + "epoch": 2.279608192341941, + "grad_norm": 3.095365285873413, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8942986726760864, + "num_tokens": 695665084.0, + "step": 17920 + }, + { + "epoch": 2.2797354026205316, + "grad_norm": 2.7582287788391113, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9108321666717529, + "num_tokens": 695704473.0, + "step": 17921 + }, + { + "epoch": 2.279862612899122, + "grad_norm": 3.035128355026245, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8946017026901245, + "num_tokens": 695745455.0, + "step": 17922 + }, + { + "epoch": 2.2799898231777127, + "grad_norm": 2.770704507827759, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9093484878540039, + "num_tokens": 695788235.0, + "step": 17923 + }, + { + "epoch": 2.2801170334563032, + "grad_norm": 2.6573047637939453, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9093600511550903, + "num_tokens": 695832565.0, + "step": 17924 + }, + { + "epoch": 2.2802442437348938, + "grad_norm": 2.693371057510376, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9031684398651123, + "num_tokens": 695875383.0, + "step": 17925 + }, + { + "epoch": 2.2803714540134843, + "grad_norm": 2.9456801414489746, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9081962704658508, + "num_tokens": 695913455.0, + "step": 17926 + }, + { + "epoch": 2.280498664292075, + "grad_norm": 2.914844036102295, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8984529376029968, + "num_tokens": 695952697.0, + "step": 17927 + }, + { + "epoch": 2.2806258745706653, + "grad_norm": 2.9265027046203613, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9035991430282593, + "num_tokens": 695991255.0, + "step": 17928 + }, + { + "epoch": 2.280753084849256, + "grad_norm": 2.7123701572418213, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9096660614013672, + "num_tokens": 696032506.0, + "step": 17929 + }, + { + "epoch": 2.2808802951278464, + "grad_norm": 2.981614112854004, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9016604423522949, + "num_tokens": 696071227.0, + "step": 17930 + }, + { + "epoch": 2.281007505406437, + "grad_norm": 2.6127915382385254, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9051792621612549, + "num_tokens": 696119537.0, + "step": 17931 + }, + { + "epoch": 2.2811347156850275, + "grad_norm": 3.196206569671631, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9074146747589111, + "num_tokens": 696150689.0, + "step": 17932 + }, + { + "epoch": 2.281261925963618, + "grad_norm": 2.7362985610961914, + "learning_rate": 1e-06, + "loss": 0.2168, + "mean_token_accuracy": 0.9213551878929138, + "num_tokens": 696188845.0, + "step": 17933 + }, + { + "epoch": 2.2813891362422085, + "grad_norm": 3.0257270336151123, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9006746411323547, + "num_tokens": 696229225.0, + "step": 17934 + }, + { + "epoch": 2.281516346520799, + "grad_norm": 2.7781522274017334, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9025942087173462, + "num_tokens": 696272584.0, + "step": 17935 + }, + { + "epoch": 2.2816435567993896, + "grad_norm": 3.243441343307495, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9046000838279724, + "num_tokens": 696303567.0, + "step": 17936 + }, + { + "epoch": 2.2817707670779797, + "grad_norm": 3.0019404888153076, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9075349569320679, + "num_tokens": 696339009.0, + "step": 17937 + }, + { + "epoch": 2.2818979773565706, + "grad_norm": 2.6754512786865234, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9094535708427429, + "num_tokens": 696382236.0, + "step": 17938 + }, + { + "epoch": 2.2820251876351607, + "grad_norm": 3.196181535720825, + "learning_rate": 1e-06, + "loss": 0.2958, + "mean_token_accuracy": 0.8912765979766846, + "num_tokens": 696416919.0, + "step": 17939 + }, + { + "epoch": 2.2821523979137512, + "grad_norm": 2.808051109313965, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9019244313240051, + "num_tokens": 696457003.0, + "step": 17940 + }, + { + "epoch": 2.2822796081923418, + "grad_norm": 2.9985544681549072, + "learning_rate": 1e-06, + "loss": 0.242, + "mean_token_accuracy": 0.9114205241203308, + "num_tokens": 696490832.0, + "step": 17941 + }, + { + "epoch": 2.2824068184709323, + "grad_norm": 2.7887606620788574, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9071754217147827, + "num_tokens": 696531019.0, + "step": 17942 + }, + { + "epoch": 2.282534028749523, + "grad_norm": 2.945697784423828, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9128538370132446, + "num_tokens": 696567670.0, + "step": 17943 + }, + { + "epoch": 2.2826612390281134, + "grad_norm": 3.036938190460205, + "learning_rate": 1e-06, + "loss": 0.2215, + "mean_token_accuracy": 0.9173673987388611, + "num_tokens": 696596192.0, + "step": 17944 + }, + { + "epoch": 2.282788449306704, + "grad_norm": 2.823946237564087, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9039876461029053, + "num_tokens": 696637669.0, + "step": 17945 + }, + { + "epoch": 2.2829156595852944, + "grad_norm": 3.1058857440948486, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9145650267601013, + "num_tokens": 696672929.0, + "step": 17946 + }, + { + "epoch": 2.283042869863885, + "grad_norm": 2.9436535835266113, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8953129053115845, + "num_tokens": 696711299.0, + "step": 17947 + }, + { + "epoch": 2.2831700801424755, + "grad_norm": 2.624854564666748, + "learning_rate": 1e-06, + "loss": 0.2078, + "mean_token_accuracy": 0.921101987361908, + "num_tokens": 696748481.0, + "step": 17948 + }, + { + "epoch": 2.283297290421066, + "grad_norm": 3.1515097618103027, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8954738974571228, + "num_tokens": 696784227.0, + "step": 17949 + }, + { + "epoch": 2.2834245006996565, + "grad_norm": 3.01220440864563, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.9033984541893005, + "num_tokens": 696822056.0, + "step": 17950 + }, + { + "epoch": 2.283551710978247, + "grad_norm": 2.6991329193115234, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9006941318511963, + "num_tokens": 696866987.0, + "step": 17951 + }, + { + "epoch": 2.2836789212568376, + "grad_norm": 2.739434003829956, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9127362966537476, + "num_tokens": 696905919.0, + "step": 17952 + }, + { + "epoch": 2.283806131535428, + "grad_norm": 3.071363687515259, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9023966789245605, + "num_tokens": 696940051.0, + "step": 17953 + }, + { + "epoch": 2.2839333418140186, + "grad_norm": 3.215595245361328, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8926767110824585, + "num_tokens": 696972691.0, + "step": 17954 + }, + { + "epoch": 2.284060552092609, + "grad_norm": 3.02787446975708, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8980883955955505, + "num_tokens": 697012547.0, + "step": 17955 + }, + { + "epoch": 2.2841877623711997, + "grad_norm": 3.0569546222686768, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.8973565101623535, + "num_tokens": 697049785.0, + "step": 17956 + }, + { + "epoch": 2.28431497264979, + "grad_norm": 2.4489710330963135, + "learning_rate": 1e-06, + "loss": 0.2196, + "mean_token_accuracy": 0.9186678528785706, + "num_tokens": 697092320.0, + "step": 17957 + }, + { + "epoch": 2.2844421829283807, + "grad_norm": 2.847943067550659, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9102481007575989, + "num_tokens": 697133629.0, + "step": 17958 + }, + { + "epoch": 2.2845693932069713, + "grad_norm": 2.8345117568969727, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.913262128829956, + "num_tokens": 697169055.0, + "step": 17959 + }, + { + "epoch": 2.284696603485562, + "grad_norm": 2.8841676712036133, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9021396040916443, + "num_tokens": 697210353.0, + "step": 17960 + }, + { + "epoch": 2.2848238137641523, + "grad_norm": 2.7478792667388916, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8958916664123535, + "num_tokens": 697252441.0, + "step": 17961 + }, + { + "epoch": 2.2849510240427424, + "grad_norm": 2.529313802719116, + "learning_rate": 1e-06, + "loss": 0.2288, + "mean_token_accuracy": 0.9139545559883118, + "num_tokens": 697295029.0, + "step": 17962 + }, + { + "epoch": 2.2850782343213334, + "grad_norm": 2.9082884788513184, + "learning_rate": 1e-06, + "loss": 0.2425, + "mean_token_accuracy": 0.909432590007782, + "num_tokens": 697334179.0, + "step": 17963 + }, + { + "epoch": 2.2852054445999235, + "grad_norm": 3.1706254482269287, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8924416303634644, + "num_tokens": 697373503.0, + "step": 17964 + }, + { + "epoch": 2.285332654878514, + "grad_norm": 2.965015411376953, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.909054160118103, + "num_tokens": 697412264.0, + "step": 17965 + }, + { + "epoch": 2.2854598651571045, + "grad_norm": 2.8266096115112305, + "learning_rate": 1e-06, + "loss": 0.2203, + "mean_token_accuracy": 0.9164419174194336, + "num_tokens": 697447371.0, + "step": 17966 + }, + { + "epoch": 2.285587075435695, + "grad_norm": 3.0792458057403564, + "learning_rate": 1e-06, + "loss": 0.2311, + "mean_token_accuracy": 0.9132962822914124, + "num_tokens": 697476094.0, + "step": 17967 + }, + { + "epoch": 2.2857142857142856, + "grad_norm": 3.1349475383758545, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9059232473373413, + "num_tokens": 697508226.0, + "step": 17968 + }, + { + "epoch": 2.285841495992876, + "grad_norm": 2.8285934925079346, + "learning_rate": 1e-06, + "loss": 0.2333, + "mean_token_accuracy": 0.9129494428634644, + "num_tokens": 697549467.0, + "step": 17969 + }, + { + "epoch": 2.2859687062714666, + "grad_norm": 2.6695339679718018, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9085102081298828, + "num_tokens": 697596565.0, + "step": 17970 + }, + { + "epoch": 2.286095916550057, + "grad_norm": 2.9632389545440674, + "learning_rate": 1e-06, + "loss": 0.2362, + "mean_token_accuracy": 0.9117873907089233, + "num_tokens": 697631496.0, + "step": 17971 + }, + { + "epoch": 2.2862231268286477, + "grad_norm": 2.9669551849365234, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9030932784080505, + "num_tokens": 697670697.0, + "step": 17972 + }, + { + "epoch": 2.2863503371072382, + "grad_norm": 2.8968865871429443, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9017797708511353, + "num_tokens": 697709294.0, + "step": 17973 + }, + { + "epoch": 2.2864775473858288, + "grad_norm": 3.128514528274536, + "learning_rate": 1e-06, + "loss": 0.3204, + "mean_token_accuracy": 0.8824320435523987, + "num_tokens": 697749635.0, + "step": 17974 + }, + { + "epoch": 2.2866047576644193, + "grad_norm": 2.950725793838501, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9050605297088623, + "num_tokens": 697787942.0, + "step": 17975 + }, + { + "epoch": 2.28673196794301, + "grad_norm": 3.1459710597991943, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.905274510383606, + "num_tokens": 697820535.0, + "step": 17976 + }, + { + "epoch": 2.2868591782216003, + "grad_norm": 3.20192289352417, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9013537168502808, + "num_tokens": 697851615.0, + "step": 17977 + }, + { + "epoch": 2.286986388500191, + "grad_norm": 3.3227176666259766, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.9004757404327393, + "num_tokens": 697885127.0, + "step": 17978 + }, + { + "epoch": 2.2871135987787814, + "grad_norm": 2.9452552795410156, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9048852324485779, + "num_tokens": 697926224.0, + "step": 17979 + }, + { + "epoch": 2.287240809057372, + "grad_norm": 2.895387887954712, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.8996874690055847, + "num_tokens": 697965725.0, + "step": 17980 + }, + { + "epoch": 2.2873680193359625, + "grad_norm": 2.6497013568878174, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8998638391494751, + "num_tokens": 698009847.0, + "step": 17981 + }, + { + "epoch": 2.287495229614553, + "grad_norm": 3.3401739597320557, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9049674272537231, + "num_tokens": 698038770.0, + "step": 17982 + }, + { + "epoch": 2.2876224398931435, + "grad_norm": 3.00301456451416, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9076130390167236, + "num_tokens": 698071679.0, + "step": 17983 + }, + { + "epoch": 2.287749650171734, + "grad_norm": 3.3960273265838623, + "learning_rate": 1e-06, + "loss": 0.3167, + "mean_token_accuracy": 0.8840448260307312, + "num_tokens": 698105732.0, + "step": 17984 + }, + { + "epoch": 2.2878768604503246, + "grad_norm": 2.929517984390259, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9069355726242065, + "num_tokens": 698142267.0, + "step": 17985 + }, + { + "epoch": 2.288004070728915, + "grad_norm": 2.6953799724578857, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9105682373046875, + "num_tokens": 698182181.0, + "step": 17986 + }, + { + "epoch": 2.288131281007505, + "grad_norm": 2.667539596557617, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8960943818092346, + "num_tokens": 698228845.0, + "step": 17987 + }, + { + "epoch": 2.288258491286096, + "grad_norm": 2.696716547012329, + "learning_rate": 1e-06, + "loss": 0.2392, + "mean_token_accuracy": 0.9090181589126587, + "num_tokens": 698267080.0, + "step": 17988 + }, + { + "epoch": 2.2883857015646862, + "grad_norm": 3.0997567176818848, + "learning_rate": 1e-06, + "loss": 0.2994, + "mean_token_accuracy": 0.889873206615448, + "num_tokens": 698303373.0, + "step": 17989 + }, + { + "epoch": 2.2885129118432768, + "grad_norm": 2.7415080070495605, + "learning_rate": 1e-06, + "loss": 0.2364, + "mean_token_accuracy": 0.9136566519737244, + "num_tokens": 698343784.0, + "step": 17990 + }, + { + "epoch": 2.2886401221218673, + "grad_norm": 2.8044888973236084, + "learning_rate": 1e-06, + "loss": 0.2338, + "mean_token_accuracy": 0.9122937321662903, + "num_tokens": 698379270.0, + "step": 17991 + }, + { + "epoch": 2.288767332400458, + "grad_norm": 2.7975664138793945, + "learning_rate": 1e-06, + "loss": 0.2411, + "mean_token_accuracy": 0.9129887223243713, + "num_tokens": 698416068.0, + "step": 17992 + }, + { + "epoch": 2.2888945426790483, + "grad_norm": 2.9717516899108887, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8968662023544312, + "num_tokens": 698455935.0, + "step": 17993 + }, + { + "epoch": 2.289021752957639, + "grad_norm": 2.9059560298919678, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9053240418434143, + "num_tokens": 698496201.0, + "step": 17994 + }, + { + "epoch": 2.2891489632362294, + "grad_norm": 3.1038858890533447, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8962360620498657, + "num_tokens": 698532702.0, + "step": 17995 + }, + { + "epoch": 2.28927617351482, + "grad_norm": 2.9945926666259766, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9013634324073792, + "num_tokens": 698571603.0, + "step": 17996 + }, + { + "epoch": 2.2894033837934105, + "grad_norm": 2.8784539699554443, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8940016031265259, + "num_tokens": 698613192.0, + "step": 17997 + }, + { + "epoch": 2.289530594072001, + "grad_norm": 2.7352919578552246, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9099820852279663, + "num_tokens": 698654620.0, + "step": 17998 + }, + { + "epoch": 2.2896578043505915, + "grad_norm": 3.011585235595703, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8928715586662292, + "num_tokens": 698697197.0, + "step": 17999 + }, + { + "epoch": 2.289785014629182, + "grad_norm": 2.8799383640289307, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9058085680007935, + "num_tokens": 698735258.0, + "step": 18000 + }, + { + "epoch": 2.2899122249077726, + "grad_norm": 3.1477959156036377, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.9025490283966064, + "num_tokens": 698769507.0, + "step": 18001 + }, + { + "epoch": 2.290039435186363, + "grad_norm": 2.5332653522491455, + "learning_rate": 1e-06, + "loss": 0.2322, + "mean_token_accuracy": 0.9125022888183594, + "num_tokens": 698812723.0, + "step": 18002 + }, + { + "epoch": 2.2901666454649536, + "grad_norm": 2.907802104949951, + "learning_rate": 1e-06, + "loss": 0.2245, + "mean_token_accuracy": 0.9157857894897461, + "num_tokens": 698846681.0, + "step": 18003 + }, + { + "epoch": 2.290293855743544, + "grad_norm": 2.794548749923706, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9076476097106934, + "num_tokens": 698884461.0, + "step": 18004 + }, + { + "epoch": 2.2904210660221347, + "grad_norm": 3.118936538696289, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8983744978904724, + "num_tokens": 698920579.0, + "step": 18005 + }, + { + "epoch": 2.290548276300725, + "grad_norm": 2.728912353515625, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9006030559539795, + "num_tokens": 698963864.0, + "step": 18006 + }, + { + "epoch": 2.2906754865793157, + "grad_norm": 2.6149582862854004, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9087428450584412, + "num_tokens": 699005485.0, + "step": 18007 + }, + { + "epoch": 2.2908026968579063, + "grad_norm": 2.847275495529175, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9038854241371155, + "num_tokens": 699046372.0, + "step": 18008 + }, + { + "epoch": 2.290929907136497, + "grad_norm": 2.929389476776123, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.905384361743927, + "num_tokens": 699082849.0, + "step": 18009 + }, + { + "epoch": 2.2910571174150873, + "grad_norm": 2.814495325088501, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9035921096801758, + "num_tokens": 699121791.0, + "step": 18010 + }, + { + "epoch": 2.291184327693678, + "grad_norm": 2.8265020847320557, + "learning_rate": 1e-06, + "loss": 0.234, + "mean_token_accuracy": 0.9118244051933289, + "num_tokens": 699157919.0, + "step": 18011 + }, + { + "epoch": 2.291311537972268, + "grad_norm": 3.0793004035949707, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8948999643325806, + "num_tokens": 699195557.0, + "step": 18012 + }, + { + "epoch": 2.291438748250859, + "grad_norm": 2.808941602706909, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8945332765579224, + "num_tokens": 699236813.0, + "step": 18013 + }, + { + "epoch": 2.291565958529449, + "grad_norm": 3.024855136871338, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9015650749206543, + "num_tokens": 699275269.0, + "step": 18014 + }, + { + "epoch": 2.2916931688080395, + "grad_norm": 2.8883683681488037, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9029800891876221, + "num_tokens": 699317257.0, + "step": 18015 + }, + { + "epoch": 2.29182037908663, + "grad_norm": 3.2025644779205322, + "learning_rate": 1e-06, + "loss": 0.2324, + "mean_token_accuracy": 0.9157921075820923, + "num_tokens": 699347344.0, + "step": 18016 + }, + { + "epoch": 2.2919475893652206, + "grad_norm": 2.8655004501342773, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.901470422744751, + "num_tokens": 699385795.0, + "step": 18017 + }, + { + "epoch": 2.292074799643811, + "grad_norm": 2.8728156089782715, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9136240482330322, + "num_tokens": 699422229.0, + "step": 18018 + }, + { + "epoch": 2.2922020099224016, + "grad_norm": 2.8427534103393555, + "learning_rate": 1e-06, + "loss": 0.2073, + "mean_token_accuracy": 0.9209873080253601, + "num_tokens": 699456795.0, + "step": 18019 + }, + { + "epoch": 2.292329220200992, + "grad_norm": 2.9887545108795166, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.8948407173156738, + "num_tokens": 699501533.0, + "step": 18020 + }, + { + "epoch": 2.2924564304795827, + "grad_norm": 3.048398971557617, + "learning_rate": 1e-06, + "loss": 0.234, + "mean_token_accuracy": 0.9132815003395081, + "num_tokens": 699534451.0, + "step": 18021 + }, + { + "epoch": 2.2925836407581732, + "grad_norm": 2.82049298286438, + "learning_rate": 1e-06, + "loss": 0.2284, + "mean_token_accuracy": 0.9141525626182556, + "num_tokens": 699571100.0, + "step": 18022 + }, + { + "epoch": 2.2927108510367638, + "grad_norm": 2.6522512435913086, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9117188453674316, + "num_tokens": 699613979.0, + "step": 18023 + }, + { + "epoch": 2.2928380613153543, + "grad_norm": 2.635266065597534, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9116371870040894, + "num_tokens": 699657200.0, + "step": 18024 + }, + { + "epoch": 2.292965271593945, + "grad_norm": 3.0489988327026367, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9058339595794678, + "num_tokens": 699693399.0, + "step": 18025 + }, + { + "epoch": 2.2930924818725353, + "grad_norm": 2.636824607849121, + "learning_rate": 1e-06, + "loss": 0.216, + "mean_token_accuracy": 0.9191142916679382, + "num_tokens": 699733969.0, + "step": 18026 + }, + { + "epoch": 2.293219692151126, + "grad_norm": 3.159045934677124, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.8984637260437012, + "num_tokens": 699768965.0, + "step": 18027 + }, + { + "epoch": 2.2933469024297164, + "grad_norm": 2.8632514476776123, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9115256667137146, + "num_tokens": 699808101.0, + "step": 18028 + }, + { + "epoch": 2.293474112708307, + "grad_norm": 2.977055788040161, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9108827710151672, + "num_tokens": 699845281.0, + "step": 18029 + }, + { + "epoch": 2.2936013229868975, + "grad_norm": 2.8162214756011963, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9091665744781494, + "num_tokens": 699883327.0, + "step": 18030 + }, + { + "epoch": 2.293728533265488, + "grad_norm": 2.6815531253814697, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9059809446334839, + "num_tokens": 699930213.0, + "step": 18031 + }, + { + "epoch": 2.2938557435440785, + "grad_norm": 2.855288505554199, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9075311422348022, + "num_tokens": 699966718.0, + "step": 18032 + }, + { + "epoch": 2.293982953822669, + "grad_norm": 2.923583984375, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9091551303863525, + "num_tokens": 700005679.0, + "step": 18033 + }, + { + "epoch": 2.2941101641012596, + "grad_norm": 2.886183738708496, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9012662172317505, + "num_tokens": 700047251.0, + "step": 18034 + }, + { + "epoch": 2.2942373743798496, + "grad_norm": 2.7507174015045166, + "learning_rate": 1e-06, + "loss": 0.2349, + "mean_token_accuracy": 0.9136359095573425, + "num_tokens": 700085460.0, + "step": 18035 + }, + { + "epoch": 2.2943645846584406, + "grad_norm": 2.562741994857788, + "learning_rate": 1e-06, + "loss": 0.197, + "mean_token_accuracy": 0.9262931942939758, + "num_tokens": 700125041.0, + "step": 18036 + }, + { + "epoch": 2.2944917949370307, + "grad_norm": 2.8155758380889893, + "learning_rate": 1e-06, + "loss": 0.2151, + "mean_token_accuracy": 0.9216184020042419, + "num_tokens": 700162523.0, + "step": 18037 + }, + { + "epoch": 2.2946190052156212, + "grad_norm": 2.796004295349121, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8970452547073364, + "num_tokens": 700209289.0, + "step": 18038 + }, + { + "epoch": 2.2947462154942118, + "grad_norm": 2.7965071201324463, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9015337228775024, + "num_tokens": 700253430.0, + "step": 18039 + }, + { + "epoch": 2.2948734257728023, + "grad_norm": 3.2662742137908936, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9024672508239746, + "num_tokens": 700285476.0, + "step": 18040 + }, + { + "epoch": 2.295000636051393, + "grad_norm": 2.7255043983459473, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9064851999282837, + "num_tokens": 700329671.0, + "step": 18041 + }, + { + "epoch": 2.2951278463299833, + "grad_norm": 3.084632396697998, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9080357551574707, + "num_tokens": 700365703.0, + "step": 18042 + }, + { + "epoch": 2.295255056608574, + "grad_norm": 2.807206869125366, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.8989003896713257, + "num_tokens": 700408669.0, + "step": 18043 + }, + { + "epoch": 2.2953822668871644, + "grad_norm": 2.912499189376831, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9063220024108887, + "num_tokens": 700446434.0, + "step": 18044 + }, + { + "epoch": 2.295509477165755, + "grad_norm": 2.754089117050171, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9048887491226196, + "num_tokens": 700488303.0, + "step": 18045 + }, + { + "epoch": 2.2956366874443455, + "grad_norm": 2.7491583824157715, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.897574245929718, + "num_tokens": 700530344.0, + "step": 18046 + }, + { + "epoch": 2.295763897722936, + "grad_norm": 2.7856686115264893, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9063853025436401, + "num_tokens": 700572956.0, + "step": 18047 + }, + { + "epoch": 2.2958911080015265, + "grad_norm": 2.9281442165374756, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9030545949935913, + "num_tokens": 700610186.0, + "step": 18048 + }, + { + "epoch": 2.296018318280117, + "grad_norm": 2.905958414077759, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9040005803108215, + "num_tokens": 700649942.0, + "step": 18049 + }, + { + "epoch": 2.2961455285587076, + "grad_norm": 3.208407402038574, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.8933209776878357, + "num_tokens": 700689695.0, + "step": 18050 + }, + { + "epoch": 2.296272738837298, + "grad_norm": 2.7932474613189697, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9104400873184204, + "num_tokens": 700727448.0, + "step": 18051 + }, + { + "epoch": 2.2963999491158886, + "grad_norm": 3.09031081199646, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8980322480201721, + "num_tokens": 700764806.0, + "step": 18052 + }, + { + "epoch": 2.296527159394479, + "grad_norm": 2.732042074203491, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9094377160072327, + "num_tokens": 700807869.0, + "step": 18053 + }, + { + "epoch": 2.2966543696730697, + "grad_norm": 2.6861345767974854, + "learning_rate": 1e-06, + "loss": 0.229, + "mean_token_accuracy": 0.9142053127288818, + "num_tokens": 700847484.0, + "step": 18054 + }, + { + "epoch": 2.29678157995166, + "grad_norm": 2.6180264949798584, + "learning_rate": 1e-06, + "loss": 0.2435, + "mean_token_accuracy": 0.9109009504318237, + "num_tokens": 700889846.0, + "step": 18055 + }, + { + "epoch": 2.2969087902302507, + "grad_norm": 2.7171876430511475, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9096881151199341, + "num_tokens": 700930905.0, + "step": 18056 + }, + { + "epoch": 2.2970360005088413, + "grad_norm": 3.0201079845428467, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9033100605010986, + "num_tokens": 700968302.0, + "step": 18057 + }, + { + "epoch": 2.297163210787432, + "grad_norm": 3.122018814086914, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9060431718826294, + "num_tokens": 701001387.0, + "step": 18058 + }, + { + "epoch": 2.2972904210660223, + "grad_norm": 2.7213923931121826, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9061647653579712, + "num_tokens": 701046828.0, + "step": 18059 + }, + { + "epoch": 2.2974176313446124, + "grad_norm": 3.015155553817749, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8980560302734375, + "num_tokens": 701087511.0, + "step": 18060 + }, + { + "epoch": 2.2975448416232034, + "grad_norm": 3.0312695503234863, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9036964178085327, + "num_tokens": 701123136.0, + "step": 18061 + }, + { + "epoch": 2.2976720519017935, + "grad_norm": 3.078463315963745, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9125922322273254, + "num_tokens": 701158462.0, + "step": 18062 + }, + { + "epoch": 2.297799262180384, + "grad_norm": 2.97811222076416, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9101572632789612, + "num_tokens": 701192534.0, + "step": 18063 + }, + { + "epoch": 2.2979264724589745, + "grad_norm": 3.018889904022217, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9045798182487488, + "num_tokens": 701231101.0, + "step": 18064 + }, + { + "epoch": 2.298053682737565, + "grad_norm": 2.8930177688598633, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.905472457408905, + "num_tokens": 701274810.0, + "step": 18065 + }, + { + "epoch": 2.2981808930161556, + "grad_norm": 2.8575828075408936, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9066382646560669, + "num_tokens": 701312915.0, + "step": 18066 + }, + { + "epoch": 2.298308103294746, + "grad_norm": 3.495063066482544, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8987330198287964, + "num_tokens": 701354527.0, + "step": 18067 + }, + { + "epoch": 2.2984353135733366, + "grad_norm": 2.7824454307556152, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9006364941596985, + "num_tokens": 701395030.0, + "step": 18068 + }, + { + "epoch": 2.298562523851927, + "grad_norm": 2.876901149749756, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9129575490951538, + "num_tokens": 701432129.0, + "step": 18069 + }, + { + "epoch": 2.2986897341305177, + "grad_norm": 2.8907134532928467, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.8990482091903687, + "num_tokens": 701470894.0, + "step": 18070 + }, + { + "epoch": 2.298816944409108, + "grad_norm": 2.8579764366149902, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.908388078212738, + "num_tokens": 701510306.0, + "step": 18071 + }, + { + "epoch": 2.2989441546876987, + "grad_norm": 2.871123790740967, + "learning_rate": 1e-06, + "loss": 0.2278, + "mean_token_accuracy": 0.915203869342804, + "num_tokens": 701543726.0, + "step": 18072 + }, + { + "epoch": 2.2990713649662893, + "grad_norm": 2.9172420501708984, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8926834464073181, + "num_tokens": 701585207.0, + "step": 18073 + }, + { + "epoch": 2.29919857524488, + "grad_norm": 2.776582956314087, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8947569131851196, + "num_tokens": 701630309.0, + "step": 18074 + }, + { + "epoch": 2.2993257855234703, + "grad_norm": 2.787595272064209, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9037289619445801, + "num_tokens": 701673588.0, + "step": 18075 + }, + { + "epoch": 2.299452995802061, + "grad_norm": 2.956292152404785, + "learning_rate": 1e-06, + "loss": 0.2263, + "mean_token_accuracy": 0.9155769348144531, + "num_tokens": 701712171.0, + "step": 18076 + }, + { + "epoch": 2.2995802060806514, + "grad_norm": 2.8618791103363037, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9064756631851196, + "num_tokens": 701754699.0, + "step": 18077 + }, + { + "epoch": 2.299707416359242, + "grad_norm": 2.921050548553467, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9111766815185547, + "num_tokens": 701788418.0, + "step": 18078 + }, + { + "epoch": 2.2998346266378324, + "grad_norm": 2.752833604812622, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9086803793907166, + "num_tokens": 701829308.0, + "step": 18079 + }, + { + "epoch": 2.299961836916423, + "grad_norm": 3.023561477661133, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.8980104923248291, + "num_tokens": 701870046.0, + "step": 18080 + }, + { + "epoch": 2.3000890471950135, + "grad_norm": 2.8786799907684326, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8939794301986694, + "num_tokens": 701913760.0, + "step": 18081 + }, + { + "epoch": 2.300216257473604, + "grad_norm": 2.97450852394104, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9093639850616455, + "num_tokens": 701951530.0, + "step": 18082 + }, + { + "epoch": 2.3003434677521946, + "grad_norm": 2.6420364379882812, + "learning_rate": 1e-06, + "loss": 0.2096, + "mean_token_accuracy": 0.9221382737159729, + "num_tokens": 701990881.0, + "step": 18083 + }, + { + "epoch": 2.300470678030785, + "grad_norm": 2.97102689743042, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9091834425926208, + "num_tokens": 702025135.0, + "step": 18084 + }, + { + "epoch": 2.300597888309375, + "grad_norm": 2.693199634552002, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9068933725357056, + "num_tokens": 702074814.0, + "step": 18085 + }, + { + "epoch": 2.300725098587966, + "grad_norm": 2.9092750549316406, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9053707122802734, + "num_tokens": 702114281.0, + "step": 18086 + }, + { + "epoch": 2.3008523088665562, + "grad_norm": 2.8919756412506104, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9074986577033997, + "num_tokens": 702150728.0, + "step": 18087 + }, + { + "epoch": 2.3009795191451468, + "grad_norm": 3.1180717945098877, + "learning_rate": 1e-06, + "loss": 0.318, + "mean_token_accuracy": 0.8864124417304993, + "num_tokens": 702191231.0, + "step": 18088 + }, + { + "epoch": 2.3011067294237373, + "grad_norm": 2.888715982437134, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8966218829154968, + "num_tokens": 702233993.0, + "step": 18089 + }, + { + "epoch": 2.301233939702328, + "grad_norm": 2.9557125568389893, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9041184782981873, + "num_tokens": 702268817.0, + "step": 18090 + }, + { + "epoch": 2.3013611499809183, + "grad_norm": 2.870673894882202, + "learning_rate": 1e-06, + "loss": 0.2284, + "mean_token_accuracy": 0.9158910512924194, + "num_tokens": 702306631.0, + "step": 18091 + }, + { + "epoch": 2.301488360259509, + "grad_norm": 2.8859777450561523, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9047919511795044, + "num_tokens": 702343481.0, + "step": 18092 + }, + { + "epoch": 2.3016155705380994, + "grad_norm": 2.731111526489258, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9129419326782227, + "num_tokens": 702385329.0, + "step": 18093 + }, + { + "epoch": 2.30174278081669, + "grad_norm": 2.8365278244018555, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9080354571342468, + "num_tokens": 702422320.0, + "step": 18094 + }, + { + "epoch": 2.3018699910952805, + "grad_norm": 3.138261079788208, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9059255123138428, + "num_tokens": 702454329.0, + "step": 18095 + }, + { + "epoch": 2.301997201373871, + "grad_norm": 2.955629825592041, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9111244678497314, + "num_tokens": 702489310.0, + "step": 18096 + }, + { + "epoch": 2.3021244116524615, + "grad_norm": 3.124206304550171, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8982399106025696, + "num_tokens": 702528033.0, + "step": 18097 + }, + { + "epoch": 2.302251621931052, + "grad_norm": 2.740537643432617, + "learning_rate": 1e-06, + "loss": 0.2296, + "mean_token_accuracy": 0.9159844517707825, + "num_tokens": 702566829.0, + "step": 18098 + }, + { + "epoch": 2.3023788322096426, + "grad_norm": 2.862783193588257, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9000838398933411, + "num_tokens": 702605409.0, + "step": 18099 + }, + { + "epoch": 2.302506042488233, + "grad_norm": 2.8278744220733643, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8963017463684082, + "num_tokens": 702645937.0, + "step": 18100 + }, + { + "epoch": 2.3026332527668236, + "grad_norm": 3.092216730117798, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8974958658218384, + "num_tokens": 702682923.0, + "step": 18101 + }, + { + "epoch": 2.302760463045414, + "grad_norm": 2.8259401321411133, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9080801010131836, + "num_tokens": 702722674.0, + "step": 18102 + }, + { + "epoch": 2.3028876733240047, + "grad_norm": 2.85823392868042, + "learning_rate": 1e-06, + "loss": 0.2339, + "mean_token_accuracy": 0.9142114520072937, + "num_tokens": 702760799.0, + "step": 18103 + }, + { + "epoch": 2.303014883602595, + "grad_norm": 3.0614349842071533, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9103896617889404, + "num_tokens": 702795338.0, + "step": 18104 + }, + { + "epoch": 2.3031420938811857, + "grad_norm": 2.7607421875, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9048714637756348, + "num_tokens": 702838356.0, + "step": 18105 + }, + { + "epoch": 2.3032693041597763, + "grad_norm": 2.8795859813690186, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9073967933654785, + "num_tokens": 702875732.0, + "step": 18106 + }, + { + "epoch": 2.303396514438367, + "grad_norm": 2.735964298248291, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9063615798950195, + "num_tokens": 702919129.0, + "step": 18107 + }, + { + "epoch": 2.3035237247169573, + "grad_norm": 3.0601704120635986, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9100096225738525, + "num_tokens": 702953676.0, + "step": 18108 + }, + { + "epoch": 2.303650934995548, + "grad_norm": 2.842402458190918, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9011743664741516, + "num_tokens": 702995447.0, + "step": 18109 + }, + { + "epoch": 2.303778145274138, + "grad_norm": 3.008977174758911, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9078089594841003, + "num_tokens": 703029118.0, + "step": 18110 + }, + { + "epoch": 2.303905355552729, + "grad_norm": 3.2800846099853516, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8977997303009033, + "num_tokens": 703063722.0, + "step": 18111 + }, + { + "epoch": 2.304032565831319, + "grad_norm": 2.6507010459899902, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.903027355670929, + "num_tokens": 703107310.0, + "step": 18112 + }, + { + "epoch": 2.3041597761099095, + "grad_norm": 2.891810417175293, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9109609723091125, + "num_tokens": 703144463.0, + "step": 18113 + }, + { + "epoch": 2.3042869863885, + "grad_norm": 2.708144426345825, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9076740741729736, + "num_tokens": 703188722.0, + "step": 18114 + }, + { + "epoch": 2.3044141966670906, + "grad_norm": 2.7149713039398193, + "learning_rate": 1e-06, + "loss": 0.2156, + "mean_token_accuracy": 0.9187586307525635, + "num_tokens": 703229119.0, + "step": 18115 + }, + { + "epoch": 2.304541406945681, + "grad_norm": 2.768695831298828, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.9146524667739868, + "num_tokens": 703265325.0, + "step": 18116 + }, + { + "epoch": 2.3046686172242716, + "grad_norm": 2.6242995262145996, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9046567678451538, + "num_tokens": 703310828.0, + "step": 18117 + }, + { + "epoch": 2.304795827502862, + "grad_norm": 2.848334789276123, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9078535437583923, + "num_tokens": 703352070.0, + "step": 18118 + }, + { + "epoch": 2.3049230377814527, + "grad_norm": 2.7134461402893066, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.9070448875427246, + "num_tokens": 703393870.0, + "step": 18119 + }, + { + "epoch": 2.305050248060043, + "grad_norm": 2.801598072052002, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9039314985275269, + "num_tokens": 703434030.0, + "step": 18120 + }, + { + "epoch": 2.3051774583386337, + "grad_norm": 2.9516446590423584, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9057203531265259, + "num_tokens": 703467243.0, + "step": 18121 + }, + { + "epoch": 2.3053046686172243, + "grad_norm": 2.7864341735839844, + "learning_rate": 1e-06, + "loss": 0.2413, + "mean_token_accuracy": 0.9104365110397339, + "num_tokens": 703506109.0, + "step": 18122 + }, + { + "epoch": 2.305431878895815, + "grad_norm": 2.6913604736328125, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9053713083267212, + "num_tokens": 703549724.0, + "step": 18123 + }, + { + "epoch": 2.3055590891744053, + "grad_norm": 2.538517951965332, + "learning_rate": 1e-06, + "loss": 0.22, + "mean_token_accuracy": 0.9173949956893921, + "num_tokens": 703591898.0, + "step": 18124 + }, + { + "epoch": 2.305686299452996, + "grad_norm": 2.531707525253296, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.913542628288269, + "num_tokens": 703637265.0, + "step": 18125 + }, + { + "epoch": 2.3058135097315864, + "grad_norm": 2.68585205078125, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9095531702041626, + "num_tokens": 703681222.0, + "step": 18126 + }, + { + "epoch": 2.305940720010177, + "grad_norm": 3.0501718521118164, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9006128311157227, + "num_tokens": 703717214.0, + "step": 18127 + }, + { + "epoch": 2.3060679302887674, + "grad_norm": 2.8019165992736816, + "learning_rate": 1e-06, + "loss": 0.2276, + "mean_token_accuracy": 0.9149878621101379, + "num_tokens": 703754159.0, + "step": 18128 + }, + { + "epoch": 2.306195140567358, + "grad_norm": 2.650468111038208, + "learning_rate": 1e-06, + "loss": 0.2114, + "mean_token_accuracy": 0.9218514561653137, + "num_tokens": 703794199.0, + "step": 18129 + }, + { + "epoch": 2.3063223508459485, + "grad_norm": 3.1245176792144775, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8908953070640564, + "num_tokens": 703836501.0, + "step": 18130 + }, + { + "epoch": 2.306449561124539, + "grad_norm": 2.6962223052978516, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.9088956117630005, + "num_tokens": 703876436.0, + "step": 18131 + }, + { + "epoch": 2.3065767714031296, + "grad_norm": 2.6370720863342285, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9058809280395508, + "num_tokens": 703922019.0, + "step": 18132 + }, + { + "epoch": 2.3067039816817196, + "grad_norm": 2.81095027923584, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.914996862411499, + "num_tokens": 703959405.0, + "step": 18133 + }, + { + "epoch": 2.3068311919603106, + "grad_norm": 2.916635751724243, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8931522369384766, + "num_tokens": 704001961.0, + "step": 18134 + }, + { + "epoch": 2.3069584022389007, + "grad_norm": 2.8990395069122314, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9044668674468994, + "num_tokens": 704041287.0, + "step": 18135 + }, + { + "epoch": 2.3070856125174912, + "grad_norm": 3.0041463375091553, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.900621771812439, + "num_tokens": 704080104.0, + "step": 18136 + }, + { + "epoch": 2.3072128227960818, + "grad_norm": 2.8200666904449463, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9113441705703735, + "num_tokens": 704119395.0, + "step": 18137 + }, + { + "epoch": 2.3073400330746723, + "grad_norm": 2.917205572128296, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9033690690994263, + "num_tokens": 704160074.0, + "step": 18138 + }, + { + "epoch": 2.307467243353263, + "grad_norm": 2.928222179412842, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9032604098320007, + "num_tokens": 704198961.0, + "step": 18139 + }, + { + "epoch": 2.3075944536318533, + "grad_norm": 2.9930875301361084, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9056150913238525, + "num_tokens": 704236056.0, + "step": 18140 + }, + { + "epoch": 2.307721663910444, + "grad_norm": 2.948765277862549, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9080520868301392, + "num_tokens": 704271254.0, + "step": 18141 + }, + { + "epoch": 2.3078488741890344, + "grad_norm": 2.7210187911987305, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9041616320610046, + "num_tokens": 704316679.0, + "step": 18142 + }, + { + "epoch": 2.307976084467625, + "grad_norm": 3.0694682598114014, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9098472595214844, + "num_tokens": 704350647.0, + "step": 18143 + }, + { + "epoch": 2.3081032947462155, + "grad_norm": 2.8302478790283203, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9107147455215454, + "num_tokens": 704391551.0, + "step": 18144 + }, + { + "epoch": 2.308230505024806, + "grad_norm": 2.668243885040283, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.908474862575531, + "num_tokens": 704435597.0, + "step": 18145 + }, + { + "epoch": 2.3083577153033965, + "grad_norm": 2.607052803039551, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8967179656028748, + "num_tokens": 704480910.0, + "step": 18146 + }, + { + "epoch": 2.308484925581987, + "grad_norm": 3.096118688583374, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8920974135398865, + "num_tokens": 704522402.0, + "step": 18147 + }, + { + "epoch": 2.3086121358605776, + "grad_norm": 2.865844964981079, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9093698859214783, + "num_tokens": 704561607.0, + "step": 18148 + }, + { + "epoch": 2.308739346139168, + "grad_norm": 2.734070062637329, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9105234146118164, + "num_tokens": 704602637.0, + "step": 18149 + }, + { + "epoch": 2.3088665564177586, + "grad_norm": 3.2000954151153564, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8965625166893005, + "num_tokens": 704636968.0, + "step": 18150 + }, + { + "epoch": 2.308993766696349, + "grad_norm": 2.814664602279663, + "learning_rate": 1e-06, + "loss": 0.2187, + "mean_token_accuracy": 0.9183313250541687, + "num_tokens": 704673161.0, + "step": 18151 + }, + { + "epoch": 2.3091209769749397, + "grad_norm": 2.6432275772094727, + "learning_rate": 1e-06, + "loss": 0.2442, + "mean_token_accuracy": 0.9101013541221619, + "num_tokens": 704719389.0, + "step": 18152 + }, + { + "epoch": 2.30924818725353, + "grad_norm": 3.1424386501312256, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8946117162704468, + "num_tokens": 704756143.0, + "step": 18153 + }, + { + "epoch": 2.3093753975321207, + "grad_norm": 2.8466038703918457, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9059484004974365, + "num_tokens": 704793361.0, + "step": 18154 + }, + { + "epoch": 2.3095026078107113, + "grad_norm": 2.78487491607666, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.902891993522644, + "num_tokens": 704838661.0, + "step": 18155 + }, + { + "epoch": 2.309629818089302, + "grad_norm": 2.85958194732666, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9066430330276489, + "num_tokens": 704877673.0, + "step": 18156 + }, + { + "epoch": 2.3097570283678923, + "grad_norm": 2.882840633392334, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9005753993988037, + "num_tokens": 704920007.0, + "step": 18157 + }, + { + "epoch": 2.3098842386464824, + "grad_norm": 2.886016607284546, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9081613421440125, + "num_tokens": 704956868.0, + "step": 18158 + }, + { + "epoch": 2.3100114489250734, + "grad_norm": 2.6229140758514404, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9043598175048828, + "num_tokens": 705001906.0, + "step": 18159 + }, + { + "epoch": 2.3101386592036635, + "grad_norm": 3.096461534500122, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9020050764083862, + "num_tokens": 705038697.0, + "step": 18160 + }, + { + "epoch": 2.310265869482254, + "grad_norm": 2.9635488986968994, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9037763476371765, + "num_tokens": 705077373.0, + "step": 18161 + }, + { + "epoch": 2.3103930797608445, + "grad_norm": 3.1888177394866943, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.9021599292755127, + "num_tokens": 705112475.0, + "step": 18162 + }, + { + "epoch": 2.310520290039435, + "grad_norm": 2.9541702270507812, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9019155502319336, + "num_tokens": 705152367.0, + "step": 18163 + }, + { + "epoch": 2.3106475003180256, + "grad_norm": 2.74981951713562, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9008743762969971, + "num_tokens": 705193933.0, + "step": 18164 + }, + { + "epoch": 2.310774710596616, + "grad_norm": 3.2271578311920166, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9020260572433472, + "num_tokens": 705224244.0, + "step": 18165 + }, + { + "epoch": 2.3109019208752066, + "grad_norm": 2.4860341548919678, + "learning_rate": 1e-06, + "loss": 0.2084, + "mean_token_accuracy": 0.9221453070640564, + "num_tokens": 705265626.0, + "step": 18166 + }, + { + "epoch": 2.311029131153797, + "grad_norm": 2.7882094383239746, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9038689136505127, + "num_tokens": 705308124.0, + "step": 18167 + }, + { + "epoch": 2.3111563414323877, + "grad_norm": 2.7514681816101074, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9085131883621216, + "num_tokens": 705345472.0, + "step": 18168 + }, + { + "epoch": 2.311283551710978, + "grad_norm": 2.82983136177063, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9083547592163086, + "num_tokens": 705383017.0, + "step": 18169 + }, + { + "epoch": 2.3114107619895687, + "grad_norm": 2.605811834335327, + "learning_rate": 1e-06, + "loss": 0.2104, + "mean_token_accuracy": 0.9156059622764587, + "num_tokens": 705421631.0, + "step": 18170 + }, + { + "epoch": 2.3115379722681593, + "grad_norm": 3.03831148147583, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8976083993911743, + "num_tokens": 705460031.0, + "step": 18171 + }, + { + "epoch": 2.31166518254675, + "grad_norm": 2.930743455886841, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9073441028594971, + "num_tokens": 705499359.0, + "step": 18172 + }, + { + "epoch": 2.3117923928253403, + "grad_norm": 2.7456181049346924, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9056161046028137, + "num_tokens": 705541429.0, + "step": 18173 + }, + { + "epoch": 2.311919603103931, + "grad_norm": 2.776906967163086, + "learning_rate": 1e-06, + "loss": 0.2154, + "mean_token_accuracy": 0.9189249277114868, + "num_tokens": 705578741.0, + "step": 18174 + }, + { + "epoch": 2.3120468133825214, + "grad_norm": 2.70813250541687, + "learning_rate": 1e-06, + "loss": 0.2416, + "mean_token_accuracy": 0.9115975499153137, + "num_tokens": 705620967.0, + "step": 18175 + }, + { + "epoch": 2.312174023661112, + "grad_norm": 2.7321646213531494, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9103859663009644, + "num_tokens": 705660525.0, + "step": 18176 + }, + { + "epoch": 2.3123012339397024, + "grad_norm": 2.9609909057617188, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9030543565750122, + "num_tokens": 705697162.0, + "step": 18177 + }, + { + "epoch": 2.312428444218293, + "grad_norm": 3.4050867557525635, + "learning_rate": 1e-06, + "loss": 0.3044, + "mean_token_accuracy": 0.8890262246131897, + "num_tokens": 705732409.0, + "step": 18178 + }, + { + "epoch": 2.3125556544968835, + "grad_norm": 2.7184693813323975, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9062168598175049, + "num_tokens": 705775031.0, + "step": 18179 + }, + { + "epoch": 2.312682864775474, + "grad_norm": 3.063649892807007, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9104303121566772, + "num_tokens": 705808554.0, + "step": 18180 + }, + { + "epoch": 2.3128100750540646, + "grad_norm": 2.887406587600708, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9099671840667725, + "num_tokens": 705847286.0, + "step": 18181 + }, + { + "epoch": 2.312937285332655, + "grad_norm": 2.80561900138855, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9064018726348877, + "num_tokens": 705892704.0, + "step": 18182 + }, + { + "epoch": 2.313064495611245, + "grad_norm": 3.009244203567505, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9057952761650085, + "num_tokens": 705932699.0, + "step": 18183 + }, + { + "epoch": 2.313191705889836, + "grad_norm": 2.766233205795288, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9040036201477051, + "num_tokens": 705975878.0, + "step": 18184 + }, + { + "epoch": 2.313318916168426, + "grad_norm": 3.1253273487091064, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9038993120193481, + "num_tokens": 706010231.0, + "step": 18185 + }, + { + "epoch": 2.3134461264470167, + "grad_norm": 2.8241727352142334, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8986353278160095, + "num_tokens": 706051799.0, + "step": 18186 + }, + { + "epoch": 2.3135733367256073, + "grad_norm": 2.8725149631500244, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9129221439361572, + "num_tokens": 706090836.0, + "step": 18187 + }, + { + "epoch": 2.313700547004198, + "grad_norm": 2.8137214183807373, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.9022409319877625, + "num_tokens": 706133559.0, + "step": 18188 + }, + { + "epoch": 2.3138277572827883, + "grad_norm": 2.8948113918304443, + "learning_rate": 1e-06, + "loss": 0.2425, + "mean_token_accuracy": 0.9080861806869507, + "num_tokens": 706171208.0, + "step": 18189 + }, + { + "epoch": 2.313954967561379, + "grad_norm": 2.845585823059082, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9083123207092285, + "num_tokens": 706211295.0, + "step": 18190 + }, + { + "epoch": 2.3140821778399694, + "grad_norm": 2.8886139392852783, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8917242288589478, + "num_tokens": 706255005.0, + "step": 18191 + }, + { + "epoch": 2.31420938811856, + "grad_norm": 2.885613441467285, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9032754302024841, + "num_tokens": 706292728.0, + "step": 18192 + }, + { + "epoch": 2.3143365983971504, + "grad_norm": 2.662804365158081, + "learning_rate": 1e-06, + "loss": 0.2169, + "mean_token_accuracy": 0.9198850393295288, + "num_tokens": 706329989.0, + "step": 18193 + }, + { + "epoch": 2.314463808675741, + "grad_norm": 2.8737034797668457, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9027220606803894, + "num_tokens": 706370227.0, + "step": 18194 + }, + { + "epoch": 2.3145910189543315, + "grad_norm": 2.831275224685669, + "learning_rate": 1e-06, + "loss": 0.2287, + "mean_token_accuracy": 0.9141848087310791, + "num_tokens": 706406061.0, + "step": 18195 + }, + { + "epoch": 2.314718229232922, + "grad_norm": 2.570006847381592, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9066448211669922, + "num_tokens": 706451021.0, + "step": 18196 + }, + { + "epoch": 2.3148454395115126, + "grad_norm": 3.0715811252593994, + "learning_rate": 1e-06, + "loss": 0.2269, + "mean_token_accuracy": 0.914323091506958, + "num_tokens": 706481622.0, + "step": 18197 + }, + { + "epoch": 2.314972649790103, + "grad_norm": 2.8982887268066406, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9035646915435791, + "num_tokens": 706522622.0, + "step": 18198 + }, + { + "epoch": 2.3150998600686936, + "grad_norm": 3.0328478813171387, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8912227749824524, + "num_tokens": 706564174.0, + "step": 18199 + }, + { + "epoch": 2.315227070347284, + "grad_norm": 3.026315689086914, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9028047919273376, + "num_tokens": 706602294.0, + "step": 18200 + }, + { + "epoch": 2.3153542806258747, + "grad_norm": 3.102464199066162, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9083551168441772, + "num_tokens": 706634613.0, + "step": 18201 + }, + { + "epoch": 2.315481490904465, + "grad_norm": 2.7858834266662598, + "learning_rate": 1e-06, + "loss": 0.2081, + "mean_token_accuracy": 0.9202701449394226, + "num_tokens": 706668304.0, + "step": 18202 + }, + { + "epoch": 2.3156087011830557, + "grad_norm": 2.909836769104004, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.8982419371604919, + "num_tokens": 706710071.0, + "step": 18203 + }, + { + "epoch": 2.3157359114616463, + "grad_norm": 2.949028491973877, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9036713242530823, + "num_tokens": 706748539.0, + "step": 18204 + }, + { + "epoch": 2.315863121740237, + "grad_norm": 2.7144675254821777, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9123677015304565, + "num_tokens": 706791907.0, + "step": 18205 + }, + { + "epoch": 2.3159903320188273, + "grad_norm": 2.842057943344116, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9047880172729492, + "num_tokens": 706830057.0, + "step": 18206 + }, + { + "epoch": 2.316117542297418, + "grad_norm": 2.719881296157837, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.909018874168396, + "num_tokens": 706867618.0, + "step": 18207 + }, + { + "epoch": 2.316244752576008, + "grad_norm": 3.0848655700683594, + "learning_rate": 1e-06, + "loss": 0.2969, + "mean_token_accuracy": 0.8960349559783936, + "num_tokens": 706907953.0, + "step": 18208 + }, + { + "epoch": 2.316371962854599, + "grad_norm": 2.9281957149505615, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9035923480987549, + "num_tokens": 706945893.0, + "step": 18209 + }, + { + "epoch": 2.316499173133189, + "grad_norm": 3.5669667720794678, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8978914022445679, + "num_tokens": 706975565.0, + "step": 18210 + }, + { + "epoch": 2.3166263834117795, + "grad_norm": 2.7006418704986572, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8982886075973511, + "num_tokens": 707019481.0, + "step": 18211 + }, + { + "epoch": 2.31675359369037, + "grad_norm": 3.0682930946350098, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9096369743347168, + "num_tokens": 707052637.0, + "step": 18212 + }, + { + "epoch": 2.3168808039689606, + "grad_norm": 2.8772592544555664, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9085516929626465, + "num_tokens": 707090829.0, + "step": 18213 + }, + { + "epoch": 2.317008014247551, + "grad_norm": 2.9299683570861816, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8881229162216187, + "num_tokens": 707133172.0, + "step": 18214 + }, + { + "epoch": 2.3171352245261416, + "grad_norm": 2.7984015941619873, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9125206470489502, + "num_tokens": 707169757.0, + "step": 18215 + }, + { + "epoch": 2.317262434804732, + "grad_norm": 3.0575578212738037, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9081871509552002, + "num_tokens": 707208135.0, + "step": 18216 + }, + { + "epoch": 2.3173896450833227, + "grad_norm": 2.8900389671325684, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.910032331943512, + "num_tokens": 707247145.0, + "step": 18217 + }, + { + "epoch": 2.317516855361913, + "grad_norm": 2.784818172454834, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9097169637680054, + "num_tokens": 707289525.0, + "step": 18218 + }, + { + "epoch": 2.3176440656405037, + "grad_norm": 3.1013619899749756, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9006615877151489, + "num_tokens": 707327108.0, + "step": 18219 + }, + { + "epoch": 2.3177712759190943, + "grad_norm": 3.366795778274536, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8980504274368286, + "num_tokens": 707359885.0, + "step": 18220 + }, + { + "epoch": 2.317898486197685, + "grad_norm": 2.9003777503967285, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9072121381759644, + "num_tokens": 707401506.0, + "step": 18221 + }, + { + "epoch": 2.3180256964762753, + "grad_norm": 3.057596206665039, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9015733003616333, + "num_tokens": 707438796.0, + "step": 18222 + }, + { + "epoch": 2.318152906754866, + "grad_norm": 3.077127456665039, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9025914669036865, + "num_tokens": 707474468.0, + "step": 18223 + }, + { + "epoch": 2.3182801170334564, + "grad_norm": 3.03006911277771, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9037353992462158, + "num_tokens": 707511943.0, + "step": 18224 + }, + { + "epoch": 2.318407327312047, + "grad_norm": 2.8674895763397217, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8884958028793335, + "num_tokens": 707555260.0, + "step": 18225 + }, + { + "epoch": 2.3185345375906374, + "grad_norm": 2.624716281890869, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.903944730758667, + "num_tokens": 707600906.0, + "step": 18226 + }, + { + "epoch": 2.318661747869228, + "grad_norm": 2.9404380321502686, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9064103364944458, + "num_tokens": 707638450.0, + "step": 18227 + }, + { + "epoch": 2.3187889581478185, + "grad_norm": 3.0624616146087646, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9088822603225708, + "num_tokens": 707673135.0, + "step": 18228 + }, + { + "epoch": 2.318916168426409, + "grad_norm": 2.9907708168029785, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8945248126983643, + "num_tokens": 707710266.0, + "step": 18229 + }, + { + "epoch": 2.3190433787049995, + "grad_norm": 3.0340735912323, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9010328650474548, + "num_tokens": 707745436.0, + "step": 18230 + }, + { + "epoch": 2.3191705889835896, + "grad_norm": 2.7622570991516113, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9039172530174255, + "num_tokens": 707781575.0, + "step": 18231 + }, + { + "epoch": 2.3192977992621806, + "grad_norm": 2.7687039375305176, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9067031145095825, + "num_tokens": 707824579.0, + "step": 18232 + }, + { + "epoch": 2.3194250095407707, + "grad_norm": 2.657088279724121, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.8994570374488831, + "num_tokens": 707867637.0, + "step": 18233 + }, + { + "epoch": 2.319552219819361, + "grad_norm": 2.9216103553771973, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9008154273033142, + "num_tokens": 707904766.0, + "step": 18234 + }, + { + "epoch": 2.3196794300979517, + "grad_norm": 2.9864118099212646, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9032497406005859, + "num_tokens": 707941245.0, + "step": 18235 + }, + { + "epoch": 2.3198066403765423, + "grad_norm": 2.5938267707824707, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9036170840263367, + "num_tokens": 707986805.0, + "step": 18236 + }, + { + "epoch": 2.319933850655133, + "grad_norm": 2.8269379138946533, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9052309989929199, + "num_tokens": 708026051.0, + "step": 18237 + }, + { + "epoch": 2.3200610609337233, + "grad_norm": 2.7058513164520264, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.911745548248291, + "num_tokens": 708064172.0, + "step": 18238 + }, + { + "epoch": 2.320188271212314, + "grad_norm": 2.727743625640869, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9035728573799133, + "num_tokens": 708107589.0, + "step": 18239 + }, + { + "epoch": 2.3203154814909044, + "grad_norm": 2.9659459590911865, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9047194719314575, + "num_tokens": 708147084.0, + "step": 18240 + }, + { + "epoch": 2.320442691769495, + "grad_norm": 3.0917346477508545, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.902466893196106, + "num_tokens": 708182464.0, + "step": 18241 + }, + { + "epoch": 2.3205699020480854, + "grad_norm": 3.0653915405273438, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9051485061645508, + "num_tokens": 708216313.0, + "step": 18242 + }, + { + "epoch": 2.320697112326676, + "grad_norm": 3.90887713432312, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9094333052635193, + "num_tokens": 708256724.0, + "step": 18243 + }, + { + "epoch": 2.3208243226052665, + "grad_norm": 2.4393651485443115, + "learning_rate": 1e-06, + "loss": 0.2181, + "mean_token_accuracy": 0.919205904006958, + "num_tokens": 708299590.0, + "step": 18244 + }, + { + "epoch": 2.320951532883857, + "grad_norm": 3.3787269592285156, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8917328119277954, + "num_tokens": 708337110.0, + "step": 18245 + }, + { + "epoch": 2.3210787431624476, + "grad_norm": 3.0210704803466797, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8985002040863037, + "num_tokens": 708373734.0, + "step": 18246 + }, + { + "epoch": 2.321205953441038, + "grad_norm": 3.132788896560669, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.8948369026184082, + "num_tokens": 708412651.0, + "step": 18247 + }, + { + "epoch": 2.3213331637196286, + "grad_norm": 2.7084808349609375, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9015406966209412, + "num_tokens": 708457167.0, + "step": 18248 + }, + { + "epoch": 2.321460373998219, + "grad_norm": 2.842532157897949, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9055867791175842, + "num_tokens": 708495284.0, + "step": 18249 + }, + { + "epoch": 2.3215875842768097, + "grad_norm": 2.817678928375244, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.9137712121009827, + "num_tokens": 708531508.0, + "step": 18250 + }, + { + "epoch": 2.3217147945554, + "grad_norm": 2.7608230113983154, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9053940773010254, + "num_tokens": 708575382.0, + "step": 18251 + }, + { + "epoch": 2.3218420048339907, + "grad_norm": 3.0493555068969727, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9029700756072998, + "num_tokens": 708612048.0, + "step": 18252 + }, + { + "epoch": 2.3219692151125813, + "grad_norm": 3.271556854248047, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.897828221321106, + "num_tokens": 708647639.0, + "step": 18253 + }, + { + "epoch": 2.322096425391172, + "grad_norm": 2.886962652206421, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9086179733276367, + "num_tokens": 708685827.0, + "step": 18254 + }, + { + "epoch": 2.3222236356697623, + "grad_norm": 2.822192907333374, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8949422240257263, + "num_tokens": 708727929.0, + "step": 18255 + }, + { + "epoch": 2.3223508459483524, + "grad_norm": 2.650899887084961, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9079805612564087, + "num_tokens": 708769420.0, + "step": 18256 + }, + { + "epoch": 2.3224780562269434, + "grad_norm": 3.2104907035827637, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8959965705871582, + "num_tokens": 708806024.0, + "step": 18257 + }, + { + "epoch": 2.3226052665055335, + "grad_norm": 2.8260560035705566, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9030134081840515, + "num_tokens": 708851208.0, + "step": 18258 + }, + { + "epoch": 2.322732476784124, + "grad_norm": 3.03987193107605, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8982090353965759, + "num_tokens": 708888233.0, + "step": 18259 + }, + { + "epoch": 2.3228596870627145, + "grad_norm": 2.9448773860931396, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9109674692153931, + "num_tokens": 708927302.0, + "step": 18260 + }, + { + "epoch": 2.322986897341305, + "grad_norm": 2.821495532989502, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8964927196502686, + "num_tokens": 708972008.0, + "step": 18261 + }, + { + "epoch": 2.3231141076198956, + "grad_norm": 3.0165281295776367, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8965204954147339, + "num_tokens": 709007440.0, + "step": 18262 + }, + { + "epoch": 2.323241317898486, + "grad_norm": 3.148590564727783, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9048952460289001, + "num_tokens": 709040865.0, + "step": 18263 + }, + { + "epoch": 2.3233685281770766, + "grad_norm": 3.0639264583587646, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8972107768058777, + "num_tokens": 709077686.0, + "step": 18264 + }, + { + "epoch": 2.323495738455667, + "grad_norm": 3.2030131816864014, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.912813663482666, + "num_tokens": 709108654.0, + "step": 18265 + }, + { + "epoch": 2.3236229487342577, + "grad_norm": 3.1167025566101074, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9046781063079834, + "num_tokens": 709143625.0, + "step": 18266 + }, + { + "epoch": 2.323750159012848, + "grad_norm": 2.964587926864624, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9063448309898376, + "num_tokens": 709185757.0, + "step": 18267 + }, + { + "epoch": 2.3238773692914387, + "grad_norm": 3.0462639331817627, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9009164571762085, + "num_tokens": 709223857.0, + "step": 18268 + }, + { + "epoch": 2.3240045795700293, + "grad_norm": 3.159147024154663, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9046093225479126, + "num_tokens": 709255041.0, + "step": 18269 + }, + { + "epoch": 2.32413178984862, + "grad_norm": 2.68485426902771, + "learning_rate": 1e-06, + "loss": 0.2214, + "mean_token_accuracy": 0.9188328981399536, + "num_tokens": 709291088.0, + "step": 18270 + }, + { + "epoch": 2.3242590001272103, + "grad_norm": 2.7999961376190186, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.9103301763534546, + "num_tokens": 709330413.0, + "step": 18271 + }, + { + "epoch": 2.324386210405801, + "grad_norm": 2.941821813583374, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9029589891433716, + "num_tokens": 709367244.0, + "step": 18272 + }, + { + "epoch": 2.3245134206843914, + "grad_norm": 2.82944655418396, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9038888216018677, + "num_tokens": 709406766.0, + "step": 18273 + }, + { + "epoch": 2.324640630962982, + "grad_norm": 3.111386299133301, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9031261801719666, + "num_tokens": 709440519.0, + "step": 18274 + }, + { + "epoch": 2.3247678412415724, + "grad_norm": 2.8906562328338623, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9029531478881836, + "num_tokens": 709480461.0, + "step": 18275 + }, + { + "epoch": 2.324895051520163, + "grad_norm": 2.8889081478118896, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.907303512096405, + "num_tokens": 709517761.0, + "step": 18276 + }, + { + "epoch": 2.3250222617987535, + "grad_norm": 2.9039835929870605, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9074534177780151, + "num_tokens": 709555119.0, + "step": 18277 + }, + { + "epoch": 2.325149472077344, + "grad_norm": 2.6430907249450684, + "learning_rate": 1e-06, + "loss": 0.233, + "mean_token_accuracy": 0.9129937291145325, + "num_tokens": 709596793.0, + "step": 18278 + }, + { + "epoch": 2.3252766823559345, + "grad_norm": 2.721909284591675, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9044986963272095, + "num_tokens": 709636553.0, + "step": 18279 + }, + { + "epoch": 2.325403892634525, + "grad_norm": 3.141862392425537, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9125822186470032, + "num_tokens": 709668389.0, + "step": 18280 + }, + { + "epoch": 2.325531102913115, + "grad_norm": 2.7076637744903564, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9031802415847778, + "num_tokens": 709709467.0, + "step": 18281 + }, + { + "epoch": 2.325658313191706, + "grad_norm": 2.955138921737671, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9067122936248779, + "num_tokens": 709748548.0, + "step": 18282 + }, + { + "epoch": 2.325785523470296, + "grad_norm": 2.9040541648864746, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.909784197807312, + "num_tokens": 709782897.0, + "step": 18283 + }, + { + "epoch": 2.3259127337488867, + "grad_norm": 2.8718738555908203, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8978599905967712, + "num_tokens": 709824400.0, + "step": 18284 + }, + { + "epoch": 2.3260399440274773, + "grad_norm": 2.942641496658325, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.904167115688324, + "num_tokens": 709862687.0, + "step": 18285 + }, + { + "epoch": 2.326167154306068, + "grad_norm": 2.939054489135742, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9094178080558777, + "num_tokens": 709897806.0, + "step": 18286 + }, + { + "epoch": 2.3262943645846583, + "grad_norm": 3.085554599761963, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8996205925941467, + "num_tokens": 709936031.0, + "step": 18287 + }, + { + "epoch": 2.326421574863249, + "grad_norm": 2.8485701084136963, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9045255184173584, + "num_tokens": 709977749.0, + "step": 18288 + }, + { + "epoch": 2.3265487851418394, + "grad_norm": 3.1482462882995605, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9061157703399658, + "num_tokens": 710008058.0, + "step": 18289 + }, + { + "epoch": 2.32667599542043, + "grad_norm": 2.644153356552124, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9113086462020874, + "num_tokens": 710051630.0, + "step": 18290 + }, + { + "epoch": 2.3268032056990204, + "grad_norm": 2.9688258171081543, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9016176462173462, + "num_tokens": 710087417.0, + "step": 18291 + }, + { + "epoch": 2.326930415977611, + "grad_norm": 3.260979652404785, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8906170129776001, + "num_tokens": 710124614.0, + "step": 18292 + }, + { + "epoch": 2.3270576262562015, + "grad_norm": 2.85583758354187, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9069206714630127, + "num_tokens": 710163003.0, + "step": 18293 + }, + { + "epoch": 2.327184836534792, + "grad_norm": 2.7318289279937744, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9029555320739746, + "num_tokens": 710206910.0, + "step": 18294 + }, + { + "epoch": 2.3273120468133826, + "grad_norm": 2.993701457977295, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9013805985450745, + "num_tokens": 710245289.0, + "step": 18295 + }, + { + "epoch": 2.327439257091973, + "grad_norm": 2.727999210357666, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.904577910900116, + "num_tokens": 710286088.0, + "step": 18296 + }, + { + "epoch": 2.3275664673705636, + "grad_norm": 3.0769870281219482, + "learning_rate": 1e-06, + "loss": 0.2487, + "mean_token_accuracy": 0.9082518815994263, + "num_tokens": 710320198.0, + "step": 18297 + }, + { + "epoch": 2.327693677649154, + "grad_norm": 2.8271589279174805, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9078787565231323, + "num_tokens": 710360451.0, + "step": 18298 + }, + { + "epoch": 2.3278208879277447, + "grad_norm": 3.139883279800415, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8881984949111938, + "num_tokens": 710397257.0, + "step": 18299 + }, + { + "epoch": 2.327948098206335, + "grad_norm": 3.0438616275787354, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8962560892105103, + "num_tokens": 710437576.0, + "step": 18300 + }, + { + "epoch": 2.3280753084849257, + "grad_norm": 2.8230578899383545, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9017312526702881, + "num_tokens": 710479807.0, + "step": 18301 + }, + { + "epoch": 2.3282025187635162, + "grad_norm": 2.7128002643585205, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.904888391494751, + "num_tokens": 710521015.0, + "step": 18302 + }, + { + "epoch": 2.3283297290421068, + "grad_norm": 2.8038952350616455, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8969104886054993, + "num_tokens": 710561733.0, + "step": 18303 + }, + { + "epoch": 2.3284569393206973, + "grad_norm": 3.068486452102661, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9072644710540771, + "num_tokens": 710596260.0, + "step": 18304 + }, + { + "epoch": 2.328584149599288, + "grad_norm": 2.799962282180786, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9074256420135498, + "num_tokens": 710639426.0, + "step": 18305 + }, + { + "epoch": 2.328711359877878, + "grad_norm": 3.0727202892303467, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9000917673110962, + "num_tokens": 710677509.0, + "step": 18306 + }, + { + "epoch": 2.328838570156469, + "grad_norm": 3.0928173065185547, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8927537798881531, + "num_tokens": 710717751.0, + "step": 18307 + }, + { + "epoch": 2.328965780435059, + "grad_norm": 3.4462928771972656, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8900392651557922, + "num_tokens": 710749858.0, + "step": 18308 + }, + { + "epoch": 2.3290929907136495, + "grad_norm": 3.0870261192321777, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.892737627029419, + "num_tokens": 710789527.0, + "step": 18309 + }, + { + "epoch": 2.32922020099224, + "grad_norm": 2.865738868713379, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.8988901376724243, + "num_tokens": 710829674.0, + "step": 18310 + }, + { + "epoch": 2.3293474112708306, + "grad_norm": 2.781168222427368, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.907625675201416, + "num_tokens": 710871562.0, + "step": 18311 + }, + { + "epoch": 2.329474621549421, + "grad_norm": 2.8182311058044434, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9069030284881592, + "num_tokens": 710913578.0, + "step": 18312 + }, + { + "epoch": 2.3296018318280116, + "grad_norm": 2.612743377685547, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9123953580856323, + "num_tokens": 710950470.0, + "step": 18313 + }, + { + "epoch": 2.329729042106602, + "grad_norm": 2.882997512817383, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.8986341953277588, + "num_tokens": 710989047.0, + "step": 18314 + }, + { + "epoch": 2.3298562523851927, + "grad_norm": 2.712047576904297, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9062283635139465, + "num_tokens": 711031464.0, + "step": 18315 + }, + { + "epoch": 2.329983462663783, + "grad_norm": 2.6454918384552, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9021453857421875, + "num_tokens": 711075279.0, + "step": 18316 + }, + { + "epoch": 2.3301106729423737, + "grad_norm": 3.0033698081970215, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9009982347488403, + "num_tokens": 711115246.0, + "step": 18317 + }, + { + "epoch": 2.3302378832209643, + "grad_norm": 2.8040738105773926, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9097887277603149, + "num_tokens": 711155026.0, + "step": 18318 + }, + { + "epoch": 2.330365093499555, + "grad_norm": 2.950133800506592, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9037673473358154, + "num_tokens": 711193512.0, + "step": 18319 + }, + { + "epoch": 2.3304923037781453, + "grad_norm": 2.9651899337768555, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9106945991516113, + "num_tokens": 711225836.0, + "step": 18320 + }, + { + "epoch": 2.330619514056736, + "grad_norm": 2.753617525100708, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9087342023849487, + "num_tokens": 711265658.0, + "step": 18321 + }, + { + "epoch": 2.3307467243353264, + "grad_norm": 2.925449848175049, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.903599202632904, + "num_tokens": 711304705.0, + "step": 18322 + }, + { + "epoch": 2.330873934613917, + "grad_norm": 2.7924468517303467, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9045224785804749, + "num_tokens": 711345482.0, + "step": 18323 + }, + { + "epoch": 2.3310011448925074, + "grad_norm": 3.1666462421417236, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9049558639526367, + "num_tokens": 711382691.0, + "step": 18324 + }, + { + "epoch": 2.331128355171098, + "grad_norm": 2.933224678039551, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.9109561443328857, + "num_tokens": 711422044.0, + "step": 18325 + }, + { + "epoch": 2.3312555654496885, + "grad_norm": 2.963397979736328, + "learning_rate": 1e-06, + "loss": 0.2234, + "mean_token_accuracy": 0.9139664173126221, + "num_tokens": 711454691.0, + "step": 18326 + }, + { + "epoch": 2.331382775728279, + "grad_norm": 2.487691879272461, + "learning_rate": 1e-06, + "loss": 0.2248, + "mean_token_accuracy": 0.9158598780632019, + "num_tokens": 711501255.0, + "step": 18327 + }, + { + "epoch": 2.3315099860068695, + "grad_norm": 2.741858720779419, + "learning_rate": 1e-06, + "loss": 0.2392, + "mean_token_accuracy": 0.9135306477546692, + "num_tokens": 711540621.0, + "step": 18328 + }, + { + "epoch": 2.3316371962854596, + "grad_norm": 2.8472530841827393, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9030086994171143, + "num_tokens": 711583280.0, + "step": 18329 + }, + { + "epoch": 2.3317644065640506, + "grad_norm": 2.7926602363586426, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9099259376525879, + "num_tokens": 711623048.0, + "step": 18330 + }, + { + "epoch": 2.3318916168426407, + "grad_norm": 2.8467109203338623, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.912511944770813, + "num_tokens": 711659463.0, + "step": 18331 + }, + { + "epoch": 2.332018827121231, + "grad_norm": 2.7697930335998535, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8974288702011108, + "num_tokens": 711704003.0, + "step": 18332 + }, + { + "epoch": 2.3321460373998217, + "grad_norm": 3.024662733078003, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.8990871906280518, + "num_tokens": 711740546.0, + "step": 18333 + }, + { + "epoch": 2.3322732476784123, + "grad_norm": 3.1827075481414795, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9062447547912598, + "num_tokens": 711771848.0, + "step": 18334 + }, + { + "epoch": 2.332400457957003, + "grad_norm": 3.0025837421417236, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.8998775482177734, + "num_tokens": 711808226.0, + "step": 18335 + }, + { + "epoch": 2.3325276682355933, + "grad_norm": 2.9054019451141357, + "learning_rate": 1e-06, + "loss": 0.2393, + "mean_token_accuracy": 0.9136371612548828, + "num_tokens": 711847962.0, + "step": 18336 + }, + { + "epoch": 2.332654878514184, + "grad_norm": 2.714789867401123, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9054543972015381, + "num_tokens": 711892179.0, + "step": 18337 + }, + { + "epoch": 2.3327820887927744, + "grad_norm": 2.6860456466674805, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9030765295028687, + "num_tokens": 711932869.0, + "step": 18338 + }, + { + "epoch": 2.332909299071365, + "grad_norm": 2.608452320098877, + "learning_rate": 1e-06, + "loss": 0.2296, + "mean_token_accuracy": 0.9138412475585938, + "num_tokens": 711972202.0, + "step": 18339 + }, + { + "epoch": 2.3330365093499554, + "grad_norm": 2.5472376346588135, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.91207355260849, + "num_tokens": 712013687.0, + "step": 18340 + }, + { + "epoch": 2.333163719628546, + "grad_norm": 3.024216413497925, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.8992522954940796, + "num_tokens": 712050162.0, + "step": 18341 + }, + { + "epoch": 2.3332909299071365, + "grad_norm": 3.006237268447876, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.906174898147583, + "num_tokens": 712086105.0, + "step": 18342 + }, + { + "epoch": 2.333418140185727, + "grad_norm": 3.196385383605957, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8952735066413879, + "num_tokens": 712120208.0, + "step": 18343 + }, + { + "epoch": 2.3335453504643175, + "grad_norm": 2.6583189964294434, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9122485518455505, + "num_tokens": 712158839.0, + "step": 18344 + }, + { + "epoch": 2.333672560742908, + "grad_norm": 2.7897279262542725, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9019694328308105, + "num_tokens": 712200950.0, + "step": 18345 + }, + { + "epoch": 2.3337997710214986, + "grad_norm": 2.6393821239471436, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9047867059707642, + "num_tokens": 712244441.0, + "step": 18346 + }, + { + "epoch": 2.333926981300089, + "grad_norm": 2.935312271118164, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9017534852027893, + "num_tokens": 712279617.0, + "step": 18347 + }, + { + "epoch": 2.3340541915786797, + "grad_norm": 2.7738020420074463, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8995450735092163, + "num_tokens": 712322648.0, + "step": 18348 + }, + { + "epoch": 2.33418140185727, + "grad_norm": 2.7723171710968018, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9126620292663574, + "num_tokens": 712360522.0, + "step": 18349 + }, + { + "epoch": 2.3343086121358607, + "grad_norm": 3.180602788925171, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9105035066604614, + "num_tokens": 712394085.0, + "step": 18350 + }, + { + "epoch": 2.3344358224144512, + "grad_norm": 2.8577775955200195, + "learning_rate": 1e-06, + "loss": 0.2093, + "mean_token_accuracy": 0.9225304126739502, + "num_tokens": 712429731.0, + "step": 18351 + }, + { + "epoch": 2.3345630326930418, + "grad_norm": 2.9209821224212646, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8962743282318115, + "num_tokens": 712469104.0, + "step": 18352 + }, + { + "epoch": 2.3346902429716323, + "grad_norm": 3.013561248779297, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.8985262513160706, + "num_tokens": 712506337.0, + "step": 18353 + }, + { + "epoch": 2.3348174532502224, + "grad_norm": 2.821176290512085, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9110711812973022, + "num_tokens": 712545383.0, + "step": 18354 + }, + { + "epoch": 2.3349446635288134, + "grad_norm": 2.941150188446045, + "learning_rate": 1e-06, + "loss": 0.2345, + "mean_token_accuracy": 0.9134668111801147, + "num_tokens": 712581318.0, + "step": 18355 + }, + { + "epoch": 2.3350718738074034, + "grad_norm": 2.7568600177764893, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9087268114089966, + "num_tokens": 712621790.0, + "step": 18356 + }, + { + "epoch": 2.335199084085994, + "grad_norm": 2.634397506713867, + "learning_rate": 1e-06, + "loss": 0.2195, + "mean_token_accuracy": 0.9157503843307495, + "num_tokens": 712658311.0, + "step": 18357 + }, + { + "epoch": 2.3353262943645845, + "grad_norm": 2.9016411304473877, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9047146439552307, + "num_tokens": 712696625.0, + "step": 18358 + }, + { + "epoch": 2.335453504643175, + "grad_norm": 2.9013285636901855, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9023160338401794, + "num_tokens": 712734876.0, + "step": 18359 + }, + { + "epoch": 2.3355807149217656, + "grad_norm": 2.828592538833618, + "learning_rate": 1e-06, + "loss": 0.2123, + "mean_token_accuracy": 0.9193549752235413, + "num_tokens": 712771321.0, + "step": 18360 + }, + { + "epoch": 2.335707925200356, + "grad_norm": 2.988288640975952, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9001113176345825, + "num_tokens": 712810867.0, + "step": 18361 + }, + { + "epoch": 2.3358351354789466, + "grad_norm": 2.8348567485809326, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9119494557380676, + "num_tokens": 712846471.0, + "step": 18362 + }, + { + "epoch": 2.335962345757537, + "grad_norm": 3.0455172061920166, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.896342396736145, + "num_tokens": 712882482.0, + "step": 18363 + }, + { + "epoch": 2.3360895560361277, + "grad_norm": 2.940244674682617, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8961931467056274, + "num_tokens": 712920014.0, + "step": 18364 + }, + { + "epoch": 2.336216766314718, + "grad_norm": 2.6837685108184814, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9059275984764099, + "num_tokens": 712961137.0, + "step": 18365 + }, + { + "epoch": 2.3363439765933087, + "grad_norm": 2.6803715229034424, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9137054681777954, + "num_tokens": 713003046.0, + "step": 18366 + }, + { + "epoch": 2.3364711868718993, + "grad_norm": 2.7431280612945557, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9035882949829102, + "num_tokens": 713043979.0, + "step": 18367 + }, + { + "epoch": 2.33659839715049, + "grad_norm": 2.693438768386841, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9040384292602539, + "num_tokens": 713089380.0, + "step": 18368 + }, + { + "epoch": 2.3367256074290803, + "grad_norm": 2.965829372406006, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.897392213344574, + "num_tokens": 713132256.0, + "step": 18369 + }, + { + "epoch": 2.336852817707671, + "grad_norm": 2.9984798431396484, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9008216857910156, + "num_tokens": 713169548.0, + "step": 18370 + }, + { + "epoch": 2.3369800279862614, + "grad_norm": 3.1348867416381836, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9040377140045166, + "num_tokens": 713203819.0, + "step": 18371 + }, + { + "epoch": 2.337107238264852, + "grad_norm": 2.8406596183776855, + "learning_rate": 1e-06, + "loss": 0.2264, + "mean_token_accuracy": 0.9158223867416382, + "num_tokens": 713238505.0, + "step": 18372 + }, + { + "epoch": 2.3372344485434424, + "grad_norm": 2.7307732105255127, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9050227403640747, + "num_tokens": 713278968.0, + "step": 18373 + }, + { + "epoch": 2.337361658822033, + "grad_norm": 2.611632823944092, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9113622903823853, + "num_tokens": 713321288.0, + "step": 18374 + }, + { + "epoch": 2.3374888691006235, + "grad_norm": 2.8942394256591797, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8978766202926636, + "num_tokens": 713363307.0, + "step": 18375 + }, + { + "epoch": 2.337616079379214, + "grad_norm": 2.8387680053710938, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9076242446899414, + "num_tokens": 713400803.0, + "step": 18376 + }, + { + "epoch": 2.3377432896578045, + "grad_norm": 2.863447666168213, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9109189510345459, + "num_tokens": 713441177.0, + "step": 18377 + }, + { + "epoch": 2.337870499936395, + "grad_norm": 2.527069330215454, + "learning_rate": 1e-06, + "loss": 0.209, + "mean_token_accuracy": 0.9205597639083862, + "num_tokens": 713479986.0, + "step": 18378 + }, + { + "epoch": 2.337997710214985, + "grad_norm": 3.104724407196045, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9060714244842529, + "num_tokens": 713512266.0, + "step": 18379 + }, + { + "epoch": 2.338124920493576, + "grad_norm": 2.6814427375793457, + "learning_rate": 1e-06, + "loss": 0.231, + "mean_token_accuracy": 0.9145252108573914, + "num_tokens": 713552324.0, + "step": 18380 + }, + { + "epoch": 2.338252130772166, + "grad_norm": 2.8294577598571777, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9061267375946045, + "num_tokens": 713592324.0, + "step": 18381 + }, + { + "epoch": 2.3383793410507567, + "grad_norm": 2.8542470932006836, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9045313000679016, + "num_tokens": 713634157.0, + "step": 18382 + }, + { + "epoch": 2.3385065513293473, + "grad_norm": 3.133556365966797, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8983637094497681, + "num_tokens": 713668844.0, + "step": 18383 + }, + { + "epoch": 2.338633761607938, + "grad_norm": 2.8339030742645264, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8968300819396973, + "num_tokens": 713716382.0, + "step": 18384 + }, + { + "epoch": 2.3387609718865283, + "grad_norm": 2.9423489570617676, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9031509757041931, + "num_tokens": 713753385.0, + "step": 18385 + }, + { + "epoch": 2.338888182165119, + "grad_norm": 2.852241277694702, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8962123394012451, + "num_tokens": 713794722.0, + "step": 18386 + }, + { + "epoch": 2.3390153924437094, + "grad_norm": 3.1212737560272217, + "learning_rate": 1e-06, + "loss": 0.2942, + "mean_token_accuracy": 0.8929897546768188, + "num_tokens": 713833916.0, + "step": 18387 + }, + { + "epoch": 2.3391426027223, + "grad_norm": 2.6217050552368164, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9098798036575317, + "num_tokens": 713876620.0, + "step": 18388 + }, + { + "epoch": 2.3392698130008904, + "grad_norm": 2.927818536758423, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9058322906494141, + "num_tokens": 713917789.0, + "step": 18389 + }, + { + "epoch": 2.339397023279481, + "grad_norm": 2.672264814376831, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9048311710357666, + "num_tokens": 713958320.0, + "step": 18390 + }, + { + "epoch": 2.3395242335580715, + "grad_norm": 2.945655584335327, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9091268181800842, + "num_tokens": 713994639.0, + "step": 18391 + }, + { + "epoch": 2.339651443836662, + "grad_norm": 2.9150338172912598, + "learning_rate": 1e-06, + "loss": 0.2294, + "mean_token_accuracy": 0.9137139916419983, + "num_tokens": 714029976.0, + "step": 18392 + }, + { + "epoch": 2.3397786541152525, + "grad_norm": 2.729259490966797, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9107367396354675, + "num_tokens": 714071117.0, + "step": 18393 + }, + { + "epoch": 2.339905864393843, + "grad_norm": 2.691526174545288, + "learning_rate": 1e-06, + "loss": 0.2244, + "mean_token_accuracy": 0.9154603481292725, + "num_tokens": 714109596.0, + "step": 18394 + }, + { + "epoch": 2.3400330746724336, + "grad_norm": 2.775973320007324, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9100714921951294, + "num_tokens": 714151485.0, + "step": 18395 + }, + { + "epoch": 2.340160284951024, + "grad_norm": 2.761363983154297, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9025197625160217, + "num_tokens": 714194474.0, + "step": 18396 + }, + { + "epoch": 2.3402874952296147, + "grad_norm": 3.056981086730957, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9001866579055786, + "num_tokens": 714231366.0, + "step": 18397 + }, + { + "epoch": 2.340414705508205, + "grad_norm": 2.853499412536621, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9056576490402222, + "num_tokens": 714274036.0, + "step": 18398 + }, + { + "epoch": 2.3405419157867957, + "grad_norm": 2.739307403564453, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9047805070877075, + "num_tokens": 714314623.0, + "step": 18399 + }, + { + "epoch": 2.3406691260653862, + "grad_norm": 2.7678000926971436, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9036993980407715, + "num_tokens": 714356715.0, + "step": 18400 + }, + { + "epoch": 2.3407963363439768, + "grad_norm": 2.9033501148223877, + "learning_rate": 1e-06, + "loss": 0.2245, + "mean_token_accuracy": 0.9158011674880981, + "num_tokens": 714389728.0, + "step": 18401 + }, + { + "epoch": 2.3409235466225673, + "grad_norm": 2.697232484817505, + "learning_rate": 1e-06, + "loss": 0.2306, + "mean_token_accuracy": 0.9131213426589966, + "num_tokens": 714428888.0, + "step": 18402 + }, + { + "epoch": 2.341050756901158, + "grad_norm": 2.8732426166534424, + "learning_rate": 1e-06, + "loss": 0.232, + "mean_token_accuracy": 0.9112933278083801, + "num_tokens": 714464092.0, + "step": 18403 + }, + { + "epoch": 2.341177967179748, + "grad_norm": 2.9243805408477783, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.898404061794281, + "num_tokens": 714504683.0, + "step": 18404 + }, + { + "epoch": 2.341305177458339, + "grad_norm": 2.8357646465301514, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.902666449546814, + "num_tokens": 714545151.0, + "step": 18405 + }, + { + "epoch": 2.341432387736929, + "grad_norm": 2.8006832599639893, + "learning_rate": 1e-06, + "loss": 0.2258, + "mean_token_accuracy": 0.9136992692947388, + "num_tokens": 714581327.0, + "step": 18406 + }, + { + "epoch": 2.3415595980155195, + "grad_norm": 2.7296388149261475, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9010210037231445, + "num_tokens": 714628507.0, + "step": 18407 + }, + { + "epoch": 2.34168680829411, + "grad_norm": 3.1183621883392334, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9041716456413269, + "num_tokens": 714665422.0, + "step": 18408 + }, + { + "epoch": 2.3418140185727006, + "grad_norm": 2.8548803329467773, + "learning_rate": 1e-06, + "loss": 0.3122, + "mean_token_accuracy": 0.8861833214759827, + "num_tokens": 714714446.0, + "step": 18409 + }, + { + "epoch": 2.341941228851291, + "grad_norm": 3.516038417816162, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9088924527168274, + "num_tokens": 714747357.0, + "step": 18410 + }, + { + "epoch": 2.3420684391298816, + "grad_norm": 2.837420701980591, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9040010571479797, + "num_tokens": 714786300.0, + "step": 18411 + }, + { + "epoch": 2.342195649408472, + "grad_norm": 3.3660149574279785, + "learning_rate": 1e-06, + "loss": 0.3077, + "mean_token_accuracy": 0.8872628211975098, + "num_tokens": 714822911.0, + "step": 18412 + }, + { + "epoch": 2.3423228596870627, + "grad_norm": 3.0630407333374023, + "learning_rate": 1e-06, + "loss": 0.2241, + "mean_token_accuracy": 0.9163194894790649, + "num_tokens": 714854797.0, + "step": 18413 + }, + { + "epoch": 2.342450069965653, + "grad_norm": 2.569981813430786, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9116122126579285, + "num_tokens": 714900868.0, + "step": 18414 + }, + { + "epoch": 2.3425772802442437, + "grad_norm": 2.6329476833343506, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.9156432151794434, + "num_tokens": 714940595.0, + "step": 18415 + }, + { + "epoch": 2.3427044905228342, + "grad_norm": 2.970449924468994, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9064012169837952, + "num_tokens": 714975700.0, + "step": 18416 + }, + { + "epoch": 2.3428317008014248, + "grad_norm": 2.966459274291992, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9032373428344727, + "num_tokens": 715012562.0, + "step": 18417 + }, + { + "epoch": 2.3429589110800153, + "grad_norm": 2.7496297359466553, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9018341302871704, + "num_tokens": 715056066.0, + "step": 18418 + }, + { + "epoch": 2.343086121358606, + "grad_norm": 3.004539728164673, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9056249260902405, + "num_tokens": 715090745.0, + "step": 18419 + }, + { + "epoch": 2.3432133316371964, + "grad_norm": 3.0091702938079834, + "learning_rate": 1e-06, + "loss": 0.2316, + "mean_token_accuracy": 0.9137734770774841, + "num_tokens": 715124778.0, + "step": 18420 + }, + { + "epoch": 2.343340541915787, + "grad_norm": 3.0971055030822754, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8896197080612183, + "num_tokens": 715160013.0, + "step": 18421 + }, + { + "epoch": 2.3434677521943774, + "grad_norm": 2.898341655731201, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.8991367816925049, + "num_tokens": 715200096.0, + "step": 18422 + }, + { + "epoch": 2.343594962472968, + "grad_norm": 3.041224241256714, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9046462178230286, + "num_tokens": 715233351.0, + "step": 18423 + }, + { + "epoch": 2.3437221727515585, + "grad_norm": 3.55556321144104, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.9035365581512451, + "num_tokens": 715261674.0, + "step": 18424 + }, + { + "epoch": 2.343849383030149, + "grad_norm": 3.012887716293335, + "learning_rate": 1e-06, + "loss": 0.2201, + "mean_token_accuracy": 0.9175130128860474, + "num_tokens": 715293579.0, + "step": 18425 + }, + { + "epoch": 2.3439765933087395, + "grad_norm": 3.0191195011138916, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.899436354637146, + "num_tokens": 715332584.0, + "step": 18426 + }, + { + "epoch": 2.3441038035873296, + "grad_norm": 2.8954172134399414, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9137846231460571, + "num_tokens": 715369362.0, + "step": 18427 + }, + { + "epoch": 2.3442310138659206, + "grad_norm": 2.7455310821533203, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9053508043289185, + "num_tokens": 715411839.0, + "step": 18428 + }, + { + "epoch": 2.3443582241445107, + "grad_norm": 2.884054183959961, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8991445899009705, + "num_tokens": 715452664.0, + "step": 18429 + }, + { + "epoch": 2.344485434423101, + "grad_norm": 2.834883451461792, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8996737599372864, + "num_tokens": 715495501.0, + "step": 18430 + }, + { + "epoch": 2.3446126447016917, + "grad_norm": 2.9474802017211914, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.898005485534668, + "num_tokens": 715534363.0, + "step": 18431 + }, + { + "epoch": 2.3447398549802823, + "grad_norm": 2.8672239780426025, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9014915823936462, + "num_tokens": 715572401.0, + "step": 18432 + }, + { + "epoch": 2.344867065258873, + "grad_norm": 2.8844966888427734, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.89906907081604, + "num_tokens": 715617340.0, + "step": 18433 + }, + { + "epoch": 2.3449942755374633, + "grad_norm": 2.987588405609131, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9051656126976013, + "num_tokens": 715652630.0, + "step": 18434 + }, + { + "epoch": 2.345121485816054, + "grad_norm": 2.81326961517334, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8991938829421997, + "num_tokens": 715691292.0, + "step": 18435 + }, + { + "epoch": 2.3452486960946444, + "grad_norm": 2.788729190826416, + "learning_rate": 1e-06, + "loss": 0.2369, + "mean_token_accuracy": 0.9103675484657288, + "num_tokens": 715730333.0, + "step": 18436 + }, + { + "epoch": 2.345375906373235, + "grad_norm": 2.624124765396118, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9079107642173767, + "num_tokens": 715776010.0, + "step": 18437 + }, + { + "epoch": 2.3455031166518254, + "grad_norm": 2.950551748275757, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9079698324203491, + "num_tokens": 715812648.0, + "step": 18438 + }, + { + "epoch": 2.345630326930416, + "grad_norm": 2.800856113433838, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9072849750518799, + "num_tokens": 715848915.0, + "step": 18439 + }, + { + "epoch": 2.3457575372090065, + "grad_norm": 2.821571111679077, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9012848138809204, + "num_tokens": 715886439.0, + "step": 18440 + }, + { + "epoch": 2.345884747487597, + "grad_norm": 3.0412561893463135, + "learning_rate": 1e-06, + "loss": 0.2405, + "mean_token_accuracy": 0.9120625853538513, + "num_tokens": 715919706.0, + "step": 18441 + }, + { + "epoch": 2.3460119577661875, + "grad_norm": 2.8985483646392822, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9087846279144287, + "num_tokens": 715954548.0, + "step": 18442 + }, + { + "epoch": 2.346139168044778, + "grad_norm": 2.922584056854248, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9082010984420776, + "num_tokens": 715990706.0, + "step": 18443 + }, + { + "epoch": 2.3462663783233686, + "grad_norm": 3.074380874633789, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9062181711196899, + "num_tokens": 716026344.0, + "step": 18444 + }, + { + "epoch": 2.346393588601959, + "grad_norm": 2.925523519515991, + "learning_rate": 1e-06, + "loss": 0.2254, + "mean_token_accuracy": 0.9148247241973877, + "num_tokens": 716059159.0, + "step": 18445 + }, + { + "epoch": 2.3465207988805497, + "grad_norm": 2.7078258991241455, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9080857038497925, + "num_tokens": 716100623.0, + "step": 18446 + }, + { + "epoch": 2.34664800915914, + "grad_norm": 2.756028652191162, + "learning_rate": 1e-06, + "loss": 0.2339, + "mean_token_accuracy": 0.9129592180252075, + "num_tokens": 716139266.0, + "step": 18447 + }, + { + "epoch": 2.3467752194377307, + "grad_norm": 2.7794277667999268, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9064157009124756, + "num_tokens": 716180267.0, + "step": 18448 + }, + { + "epoch": 2.3469024297163212, + "grad_norm": 2.7231714725494385, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9062379598617554, + "num_tokens": 716219613.0, + "step": 18449 + }, + { + "epoch": 2.3470296399949118, + "grad_norm": 2.378139019012451, + "learning_rate": 1e-06, + "loss": 0.2268, + "mean_token_accuracy": 0.9118978977203369, + "num_tokens": 716264008.0, + "step": 18450 + }, + { + "epoch": 2.3471568502735023, + "grad_norm": 3.031817674636841, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9051153659820557, + "num_tokens": 716297987.0, + "step": 18451 + }, + { + "epoch": 2.3472840605520924, + "grad_norm": 2.7371044158935547, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9111628532409668, + "num_tokens": 716338445.0, + "step": 18452 + }, + { + "epoch": 2.3474112708306833, + "grad_norm": 3.0734028816223145, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.903249979019165, + "num_tokens": 716372085.0, + "step": 18453 + }, + { + "epoch": 2.3475384811092734, + "grad_norm": 2.4347875118255615, + "learning_rate": 1e-06, + "loss": 0.2267, + "mean_token_accuracy": 0.9151113629341125, + "num_tokens": 716419098.0, + "step": 18454 + }, + { + "epoch": 2.347665691387864, + "grad_norm": 3.06980562210083, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.8988913297653198, + "num_tokens": 716456735.0, + "step": 18455 + }, + { + "epoch": 2.3477929016664545, + "grad_norm": 2.5795931816101074, + "learning_rate": 1e-06, + "loss": 0.2322, + "mean_token_accuracy": 0.9121630191802979, + "num_tokens": 716495500.0, + "step": 18456 + }, + { + "epoch": 2.347920111945045, + "grad_norm": 2.9644598960876465, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9031973481178284, + "num_tokens": 716533519.0, + "step": 18457 + }, + { + "epoch": 2.3480473222236355, + "grad_norm": 3.1093645095825195, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9088366031646729, + "num_tokens": 716564945.0, + "step": 18458 + }, + { + "epoch": 2.348174532502226, + "grad_norm": 3.0215470790863037, + "learning_rate": 1e-06, + "loss": 0.2291, + "mean_token_accuracy": 0.914562463760376, + "num_tokens": 716595261.0, + "step": 18459 + }, + { + "epoch": 2.3483017427808166, + "grad_norm": 2.914608955383301, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9052048325538635, + "num_tokens": 716632921.0, + "step": 18460 + }, + { + "epoch": 2.348428953059407, + "grad_norm": 2.6488993167877197, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9063946008682251, + "num_tokens": 716674060.0, + "step": 18461 + }, + { + "epoch": 2.3485561633379977, + "grad_norm": 2.752223014831543, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9033717513084412, + "num_tokens": 716715458.0, + "step": 18462 + }, + { + "epoch": 2.348683373616588, + "grad_norm": 2.9600131511688232, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8946493864059448, + "num_tokens": 716758415.0, + "step": 18463 + }, + { + "epoch": 2.3488105838951787, + "grad_norm": 2.871474266052246, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9068798422813416, + "num_tokens": 716799467.0, + "step": 18464 + }, + { + "epoch": 2.3489377941737692, + "grad_norm": 3.0038676261901855, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8974360227584839, + "num_tokens": 716837049.0, + "step": 18465 + }, + { + "epoch": 2.3490650044523598, + "grad_norm": 2.7324469089508057, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9086555242538452, + "num_tokens": 716875721.0, + "step": 18466 + }, + { + "epoch": 2.3491922147309503, + "grad_norm": 2.989107131958008, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9067132472991943, + "num_tokens": 716912213.0, + "step": 18467 + }, + { + "epoch": 2.349319425009541, + "grad_norm": 2.906219244003296, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9110739231109619, + "num_tokens": 716948940.0, + "step": 18468 + }, + { + "epoch": 2.3494466352881314, + "grad_norm": 3.0413405895233154, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8963346481323242, + "num_tokens": 716987532.0, + "step": 18469 + }, + { + "epoch": 2.349573845566722, + "grad_norm": 2.9676923751831055, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8906433582305908, + "num_tokens": 717030145.0, + "step": 18470 + }, + { + "epoch": 2.3497010558453124, + "grad_norm": 2.8082115650177, + "learning_rate": 1e-06, + "loss": 0.2273, + "mean_token_accuracy": 0.9138325452804565, + "num_tokens": 717066885.0, + "step": 18471 + }, + { + "epoch": 2.349828266123903, + "grad_norm": 3.266376256942749, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9022786617279053, + "num_tokens": 717100880.0, + "step": 18472 + }, + { + "epoch": 2.3499554764024935, + "grad_norm": 3.156266450881958, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8930104970932007, + "num_tokens": 717140161.0, + "step": 18473 + }, + { + "epoch": 2.350082686681084, + "grad_norm": 2.939150810241699, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.900719404220581, + "num_tokens": 717182262.0, + "step": 18474 + }, + { + "epoch": 2.3502098969596745, + "grad_norm": 3.187208890914917, + "learning_rate": 1e-06, + "loss": 0.3033, + "mean_token_accuracy": 0.8890097141265869, + "num_tokens": 717220056.0, + "step": 18475 + }, + { + "epoch": 2.350337107238265, + "grad_norm": 3.0145530700683594, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.9046412706375122, + "num_tokens": 717257425.0, + "step": 18476 + }, + { + "epoch": 2.350464317516855, + "grad_norm": 3.2443206310272217, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.909500002861023, + "num_tokens": 717287868.0, + "step": 18477 + }, + { + "epoch": 2.350591527795446, + "grad_norm": 2.6284072399139404, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9069063663482666, + "num_tokens": 717329818.0, + "step": 18478 + }, + { + "epoch": 2.350718738074036, + "grad_norm": 2.817404270172119, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9064019918441772, + "num_tokens": 717371869.0, + "step": 18479 + }, + { + "epoch": 2.3508459483526267, + "grad_norm": 2.92012882232666, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9028581380844116, + "num_tokens": 717410971.0, + "step": 18480 + }, + { + "epoch": 2.3509731586312173, + "grad_norm": 2.709665060043335, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.9072451591491699, + "num_tokens": 717450692.0, + "step": 18481 + }, + { + "epoch": 2.351100368909808, + "grad_norm": 2.785893440246582, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9093577265739441, + "num_tokens": 717490993.0, + "step": 18482 + }, + { + "epoch": 2.3512275791883983, + "grad_norm": 3.0719449520111084, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8955414295196533, + "num_tokens": 717532391.0, + "step": 18483 + }, + { + "epoch": 2.351354789466989, + "grad_norm": 3.0908987522125244, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9021631479263306, + "num_tokens": 717570550.0, + "step": 18484 + }, + { + "epoch": 2.3514819997455794, + "grad_norm": 3.0165822505950928, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9029324054718018, + "num_tokens": 717606027.0, + "step": 18485 + }, + { + "epoch": 2.35160921002417, + "grad_norm": 3.039783239364624, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9027557969093323, + "num_tokens": 717642383.0, + "step": 18486 + }, + { + "epoch": 2.3517364203027604, + "grad_norm": 2.852813720703125, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8981302976608276, + "num_tokens": 717686638.0, + "step": 18487 + }, + { + "epoch": 2.351863630581351, + "grad_norm": 2.6855733394622803, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9116032123565674, + "num_tokens": 717729400.0, + "step": 18488 + }, + { + "epoch": 2.3519908408599415, + "grad_norm": 2.8369333744049072, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9066687226295471, + "num_tokens": 717771150.0, + "step": 18489 + }, + { + "epoch": 2.352118051138532, + "grad_norm": 3.074627637863159, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9046918749809265, + "num_tokens": 717808059.0, + "step": 18490 + }, + { + "epoch": 2.3522452614171225, + "grad_norm": 2.8323636054992676, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9096825122833252, + "num_tokens": 717848009.0, + "step": 18491 + }, + { + "epoch": 2.352372471695713, + "grad_norm": 3.108463764190674, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9022288918495178, + "num_tokens": 717882581.0, + "step": 18492 + }, + { + "epoch": 2.3524996819743036, + "grad_norm": 2.6057069301605225, + "learning_rate": 1e-06, + "loss": 0.2325, + "mean_token_accuracy": 0.9139952659606934, + "num_tokens": 717924643.0, + "step": 18493 + }, + { + "epoch": 2.352626892252894, + "grad_norm": 2.9833762645721436, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8957195281982422, + "num_tokens": 717963128.0, + "step": 18494 + }, + { + "epoch": 2.3527541025314846, + "grad_norm": 2.794309377670288, + "learning_rate": 1e-06, + "loss": 0.2398, + "mean_token_accuracy": 0.908146858215332, + "num_tokens": 718002586.0, + "step": 18495 + }, + { + "epoch": 2.352881312810075, + "grad_norm": 3.016146659851074, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9055879712104797, + "num_tokens": 718042715.0, + "step": 18496 + }, + { + "epoch": 2.3530085230886657, + "grad_norm": 2.8171634674072266, + "learning_rate": 1e-06, + "loss": 0.2435, + "mean_token_accuracy": 0.9093969464302063, + "num_tokens": 718077306.0, + "step": 18497 + }, + { + "epoch": 2.3531357333672562, + "grad_norm": 2.96488618850708, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9080450534820557, + "num_tokens": 718116120.0, + "step": 18498 + }, + { + "epoch": 2.3532629436458468, + "grad_norm": 2.7973737716674805, + "learning_rate": 1e-06, + "loss": 0.2432, + "mean_token_accuracy": 0.9135376811027527, + "num_tokens": 718156285.0, + "step": 18499 + }, + { + "epoch": 2.353390153924437, + "grad_norm": 2.860170602798462, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9044710397720337, + "num_tokens": 718194617.0, + "step": 18500 + }, + { + "epoch": 2.353517364203028, + "grad_norm": 3.1749379634857178, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8977935314178467, + "num_tokens": 718229090.0, + "step": 18501 + }, + { + "epoch": 2.353644574481618, + "grad_norm": 2.8817665576934814, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9118039011955261, + "num_tokens": 718266004.0, + "step": 18502 + }, + { + "epoch": 2.353771784760209, + "grad_norm": 3.1129565238952637, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9143574237823486, + "num_tokens": 718299042.0, + "step": 18503 + }, + { + "epoch": 2.353898995038799, + "grad_norm": 2.8672070503234863, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.894897997379303, + "num_tokens": 718342859.0, + "step": 18504 + }, + { + "epoch": 2.3540262053173895, + "grad_norm": 2.963613986968994, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9015215635299683, + "num_tokens": 718381384.0, + "step": 18505 + }, + { + "epoch": 2.35415341559598, + "grad_norm": 3.2061896324157715, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8843818306922913, + "num_tokens": 718420676.0, + "step": 18506 + }, + { + "epoch": 2.3542806258745705, + "grad_norm": 2.733290195465088, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.907474935054779, + "num_tokens": 718466487.0, + "step": 18507 + }, + { + "epoch": 2.354407836153161, + "grad_norm": 2.788642168045044, + "learning_rate": 1e-06, + "loss": 0.2275, + "mean_token_accuracy": 0.9141607284545898, + "num_tokens": 718507580.0, + "step": 18508 + }, + { + "epoch": 2.3545350464317516, + "grad_norm": 3.2391793727874756, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.905495822429657, + "num_tokens": 718540906.0, + "step": 18509 + }, + { + "epoch": 2.354662256710342, + "grad_norm": 3.131518840789795, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9054657220840454, + "num_tokens": 718573092.0, + "step": 18510 + }, + { + "epoch": 2.3547894669889327, + "grad_norm": 3.2334020137786865, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9014551043510437, + "num_tokens": 718612511.0, + "step": 18511 + }, + { + "epoch": 2.354916677267523, + "grad_norm": 2.7133421897888184, + "learning_rate": 1e-06, + "loss": 0.2178, + "mean_token_accuracy": 0.9178031086921692, + "num_tokens": 718648896.0, + "step": 18512 + }, + { + "epoch": 2.3550438875461137, + "grad_norm": 3.0638420581817627, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.889879584312439, + "num_tokens": 718685568.0, + "step": 18513 + }, + { + "epoch": 2.3551710978247042, + "grad_norm": 2.843825578689575, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9084255695343018, + "num_tokens": 718721972.0, + "step": 18514 + }, + { + "epoch": 2.3552983081032948, + "grad_norm": 3.122972249984741, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9003191590309143, + "num_tokens": 718757938.0, + "step": 18515 + }, + { + "epoch": 2.3554255183818853, + "grad_norm": 3.00223970413208, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9028047919273376, + "num_tokens": 718795649.0, + "step": 18516 + }, + { + "epoch": 2.355552728660476, + "grad_norm": 2.945556163787842, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9011850357055664, + "num_tokens": 718831921.0, + "step": 18517 + }, + { + "epoch": 2.3556799389390664, + "grad_norm": 2.9303736686706543, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9049631357192993, + "num_tokens": 718870780.0, + "step": 18518 + }, + { + "epoch": 2.355807149217657, + "grad_norm": 2.659106969833374, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9107683897018433, + "num_tokens": 718913678.0, + "step": 18519 + }, + { + "epoch": 2.3559343594962474, + "grad_norm": 3.0531814098358154, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.8990157246589661, + "num_tokens": 718952912.0, + "step": 18520 + }, + { + "epoch": 2.356061569774838, + "grad_norm": 3.149048089981079, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8947261571884155, + "num_tokens": 718989889.0, + "step": 18521 + }, + { + "epoch": 2.3561887800534285, + "grad_norm": 3.0989716053009033, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9068424105644226, + "num_tokens": 719027059.0, + "step": 18522 + }, + { + "epoch": 2.356315990332019, + "grad_norm": 2.849947929382324, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9037046432495117, + "num_tokens": 719067387.0, + "step": 18523 + }, + { + "epoch": 2.3564432006106095, + "grad_norm": 2.7870566844940186, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9116288423538208, + "num_tokens": 719106994.0, + "step": 18524 + }, + { + "epoch": 2.3565704108891996, + "grad_norm": 3.1474273204803467, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9060845375061035, + "num_tokens": 719147221.0, + "step": 18525 + }, + { + "epoch": 2.3566976211677906, + "grad_norm": 2.788788080215454, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.8981886506080627, + "num_tokens": 719191320.0, + "step": 18526 + }, + { + "epoch": 2.3568248314463807, + "grad_norm": 2.9105448722839355, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8963425159454346, + "num_tokens": 719230796.0, + "step": 18527 + }, + { + "epoch": 2.356952041724971, + "grad_norm": 3.1084821224212646, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.8997500538825989, + "num_tokens": 719268299.0, + "step": 18528 + }, + { + "epoch": 2.3570792520035617, + "grad_norm": 2.9960312843322754, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.901362955570221, + "num_tokens": 719306865.0, + "step": 18529 + }, + { + "epoch": 2.3572064622821522, + "grad_norm": 2.8441407680511475, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9106026887893677, + "num_tokens": 719346148.0, + "step": 18530 + }, + { + "epoch": 2.3573336725607428, + "grad_norm": 3.068877696990967, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8986127376556396, + "num_tokens": 719386104.0, + "step": 18531 + }, + { + "epoch": 2.3574608828393333, + "grad_norm": 2.843564033508301, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9118306040763855, + "num_tokens": 719425005.0, + "step": 18532 + }, + { + "epoch": 2.357588093117924, + "grad_norm": 2.88447904586792, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.8997313976287842, + "num_tokens": 719468642.0, + "step": 18533 + }, + { + "epoch": 2.3577153033965144, + "grad_norm": 2.894012689590454, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9041491746902466, + "num_tokens": 719505402.0, + "step": 18534 + }, + { + "epoch": 2.357842513675105, + "grad_norm": 2.892460346221924, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9068142771720886, + "num_tokens": 719544775.0, + "step": 18535 + }, + { + "epoch": 2.3579697239536954, + "grad_norm": 2.7212212085723877, + "learning_rate": 1e-06, + "loss": 0.2323, + "mean_token_accuracy": 0.9152566194534302, + "num_tokens": 719579605.0, + "step": 18536 + }, + { + "epoch": 2.358096934232286, + "grad_norm": 2.9437623023986816, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.9002892971038818, + "num_tokens": 719625750.0, + "step": 18537 + }, + { + "epoch": 2.3582241445108765, + "grad_norm": 2.822375774383545, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.9033188819885254, + "num_tokens": 719667234.0, + "step": 18538 + }, + { + "epoch": 2.358351354789467, + "grad_norm": 2.9462730884552, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9124983549118042, + "num_tokens": 719701666.0, + "step": 18539 + }, + { + "epoch": 2.3584785650680575, + "grad_norm": 2.997370719909668, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9069237112998962, + "num_tokens": 719738206.0, + "step": 18540 + }, + { + "epoch": 2.358605775346648, + "grad_norm": 2.772181987762451, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9033825397491455, + "num_tokens": 719779352.0, + "step": 18541 + }, + { + "epoch": 2.3587329856252386, + "grad_norm": 2.8776192665100098, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9010088443756104, + "num_tokens": 719819848.0, + "step": 18542 + }, + { + "epoch": 2.358860195903829, + "grad_norm": 2.871537208557129, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8969308137893677, + "num_tokens": 719859875.0, + "step": 18543 + }, + { + "epoch": 2.3589874061824196, + "grad_norm": 2.6947062015533447, + "learning_rate": 1e-06, + "loss": 0.2157, + "mean_token_accuracy": 0.9190391302108765, + "num_tokens": 719897714.0, + "step": 18544 + }, + { + "epoch": 2.35911461646101, + "grad_norm": 2.7987942695617676, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9068901538848877, + "num_tokens": 719937294.0, + "step": 18545 + }, + { + "epoch": 2.3592418267396007, + "grad_norm": 3.30867338180542, + "learning_rate": 1e-06, + "loss": 0.2304, + "mean_token_accuracy": 0.9143362045288086, + "num_tokens": 719963963.0, + "step": 18546 + }, + { + "epoch": 2.3593690370181912, + "grad_norm": 2.5734407901763916, + "learning_rate": 1e-06, + "loss": 0.2271, + "mean_token_accuracy": 0.9152562618255615, + "num_tokens": 720006252.0, + "step": 18547 + }, + { + "epoch": 2.3594962472967818, + "grad_norm": 2.7473697662353516, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9104579091072083, + "num_tokens": 720042672.0, + "step": 18548 + }, + { + "epoch": 2.3596234575753723, + "grad_norm": 2.831864595413208, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9116387963294983, + "num_tokens": 720080563.0, + "step": 18549 + }, + { + "epoch": 2.3597506678539624, + "grad_norm": 3.084841251373291, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8938454985618591, + "num_tokens": 720118204.0, + "step": 18550 + }, + { + "epoch": 2.3598778781325533, + "grad_norm": 2.9079878330230713, + "learning_rate": 1e-06, + "loss": 0.2337, + "mean_token_accuracy": 0.913524866104126, + "num_tokens": 720153540.0, + "step": 18551 + }, + { + "epoch": 2.3600050884111434, + "grad_norm": 2.9249305725097656, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9076499342918396, + "num_tokens": 720191896.0, + "step": 18552 + }, + { + "epoch": 2.360132298689734, + "grad_norm": 2.8588781356811523, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9009649753570557, + "num_tokens": 720232812.0, + "step": 18553 + }, + { + "epoch": 2.3602595089683245, + "grad_norm": 3.182469367980957, + "learning_rate": 1e-06, + "loss": 0.2945, + "mean_token_accuracy": 0.8917828798294067, + "num_tokens": 720269130.0, + "step": 18554 + }, + { + "epoch": 2.360386719246915, + "grad_norm": 2.8088109493255615, + "learning_rate": 1e-06, + "loss": 0.2283, + "mean_token_accuracy": 0.915458083152771, + "num_tokens": 720306181.0, + "step": 18555 + }, + { + "epoch": 2.3605139295255055, + "grad_norm": 2.8405096530914307, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9071547985076904, + "num_tokens": 720349902.0, + "step": 18556 + }, + { + "epoch": 2.360641139804096, + "grad_norm": 3.0817272663116455, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.8983150720596313, + "num_tokens": 720386554.0, + "step": 18557 + }, + { + "epoch": 2.3607683500826866, + "grad_norm": 2.739563465118408, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9102972745895386, + "num_tokens": 720423505.0, + "step": 18558 + }, + { + "epoch": 2.360895560361277, + "grad_norm": 2.8825271129608154, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.8985900282859802, + "num_tokens": 720465470.0, + "step": 18559 + }, + { + "epoch": 2.3610227706398677, + "grad_norm": 3.0134456157684326, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8967368602752686, + "num_tokens": 720505557.0, + "step": 18560 + }, + { + "epoch": 2.361149980918458, + "grad_norm": 3.197154998779297, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9036887884140015, + "num_tokens": 720539605.0, + "step": 18561 + }, + { + "epoch": 2.3612771911970487, + "grad_norm": 2.7860536575317383, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9061205983161926, + "num_tokens": 720582968.0, + "step": 18562 + }, + { + "epoch": 2.3614044014756392, + "grad_norm": 3.0590615272521973, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9061499834060669, + "num_tokens": 720619087.0, + "step": 18563 + }, + { + "epoch": 2.3615316117542298, + "grad_norm": 2.9159722328186035, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9053442478179932, + "num_tokens": 720656278.0, + "step": 18564 + }, + { + "epoch": 2.3616588220328203, + "grad_norm": 2.7277843952178955, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9059916734695435, + "num_tokens": 720699935.0, + "step": 18565 + }, + { + "epoch": 2.361786032311411, + "grad_norm": 2.987962007522583, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9122697114944458, + "num_tokens": 720735124.0, + "step": 18566 + }, + { + "epoch": 2.3619132425900013, + "grad_norm": 2.86757493019104, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9010343551635742, + "num_tokens": 720773684.0, + "step": 18567 + }, + { + "epoch": 2.362040452868592, + "grad_norm": 2.8090837001800537, + "learning_rate": 1e-06, + "loss": 0.2295, + "mean_token_accuracy": 0.9138830900192261, + "num_tokens": 720807961.0, + "step": 18568 + }, + { + "epoch": 2.3621676631471824, + "grad_norm": 3.0814783573150635, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9094262719154358, + "num_tokens": 720840879.0, + "step": 18569 + }, + { + "epoch": 2.362294873425773, + "grad_norm": 2.6702489852905273, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8942064642906189, + "num_tokens": 720889551.0, + "step": 18570 + }, + { + "epoch": 2.3624220837043635, + "grad_norm": 2.9147777557373047, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8864061832427979, + "num_tokens": 720933767.0, + "step": 18571 + }, + { + "epoch": 2.362549293982954, + "grad_norm": 2.8345842361450195, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9107053279876709, + "num_tokens": 720971699.0, + "step": 18572 + }, + { + "epoch": 2.3626765042615445, + "grad_norm": 2.6654648780822754, + "learning_rate": 1e-06, + "loss": 0.2186, + "mean_token_accuracy": 0.9171460866928101, + "num_tokens": 721011809.0, + "step": 18573 + }, + { + "epoch": 2.362803714540135, + "grad_norm": 3.1376242637634277, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9064444303512573, + "num_tokens": 721046896.0, + "step": 18574 + }, + { + "epoch": 2.362930924818725, + "grad_norm": 3.093648910522461, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8945672512054443, + "num_tokens": 721090705.0, + "step": 18575 + }, + { + "epoch": 2.363058135097316, + "grad_norm": 2.8639018535614014, + "learning_rate": 1e-06, + "loss": 0.2169, + "mean_token_accuracy": 0.918606162071228, + "num_tokens": 721126275.0, + "step": 18576 + }, + { + "epoch": 2.363185345375906, + "grad_norm": 2.920668840408325, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.89140385389328, + "num_tokens": 721166548.0, + "step": 18577 + }, + { + "epoch": 2.3633125556544967, + "grad_norm": 2.8993184566497803, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9098441004753113, + "num_tokens": 721203647.0, + "step": 18578 + }, + { + "epoch": 2.3634397659330872, + "grad_norm": 2.822340965270996, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9058019518852234, + "num_tokens": 721242590.0, + "step": 18579 + }, + { + "epoch": 2.3635669762116778, + "grad_norm": 3.0651514530181885, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9084101915359497, + "num_tokens": 721275849.0, + "step": 18580 + }, + { + "epoch": 2.3636941864902683, + "grad_norm": 2.9284515380859375, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9100888967514038, + "num_tokens": 721312810.0, + "step": 18581 + }, + { + "epoch": 2.363821396768859, + "grad_norm": 3.151654005050659, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9002555012702942, + "num_tokens": 721348685.0, + "step": 18582 + }, + { + "epoch": 2.3639486070474494, + "grad_norm": 2.9075918197631836, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9010940194129944, + "num_tokens": 721387470.0, + "step": 18583 + }, + { + "epoch": 2.36407581732604, + "grad_norm": 3.0359861850738525, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9005987644195557, + "num_tokens": 721425066.0, + "step": 18584 + }, + { + "epoch": 2.3642030276046304, + "grad_norm": 3.019030809402466, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9049751162528992, + "num_tokens": 721458820.0, + "step": 18585 + }, + { + "epoch": 2.364330237883221, + "grad_norm": 2.9673547744750977, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9052928686141968, + "num_tokens": 721493561.0, + "step": 18586 + }, + { + "epoch": 2.3644574481618115, + "grad_norm": 2.874701499938965, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9098600149154663, + "num_tokens": 721529451.0, + "step": 18587 + }, + { + "epoch": 2.364584658440402, + "grad_norm": 2.9102203845977783, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8976696133613586, + "num_tokens": 721569014.0, + "step": 18588 + }, + { + "epoch": 2.3647118687189925, + "grad_norm": 2.9241943359375, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9015617370605469, + "num_tokens": 721606423.0, + "step": 18589 + }, + { + "epoch": 2.364839078997583, + "grad_norm": 2.9815332889556885, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9044779539108276, + "num_tokens": 721641077.0, + "step": 18590 + }, + { + "epoch": 2.3649662892761736, + "grad_norm": 2.8553922176361084, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9116363525390625, + "num_tokens": 721674958.0, + "step": 18591 + }, + { + "epoch": 2.365093499554764, + "grad_norm": 2.5393688678741455, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9101055264472961, + "num_tokens": 721722403.0, + "step": 18592 + }, + { + "epoch": 2.3652207098333546, + "grad_norm": 2.8153560161590576, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8979545831680298, + "num_tokens": 721761414.0, + "step": 18593 + }, + { + "epoch": 2.365347920111945, + "grad_norm": 2.8615121841430664, + "learning_rate": 1e-06, + "loss": 0.2411, + "mean_token_accuracy": 0.910589873790741, + "num_tokens": 721799699.0, + "step": 18594 + }, + { + "epoch": 2.3654751303905357, + "grad_norm": 3.0985753536224365, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9030869007110596, + "num_tokens": 721834147.0, + "step": 18595 + }, + { + "epoch": 2.3656023406691262, + "grad_norm": 3.1626482009887695, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.9008309841156006, + "num_tokens": 721870588.0, + "step": 18596 + }, + { + "epoch": 2.3657295509477168, + "grad_norm": 3.2669501304626465, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.894853413105011, + "num_tokens": 721904429.0, + "step": 18597 + }, + { + "epoch": 2.365856761226307, + "grad_norm": 3.0736243724823, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9039759635925293, + "num_tokens": 721941353.0, + "step": 18598 + }, + { + "epoch": 2.365983971504898, + "grad_norm": 3.0628321170806885, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8988224267959595, + "num_tokens": 721978721.0, + "step": 18599 + }, + { + "epoch": 2.366111181783488, + "grad_norm": 3.0467896461486816, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8983950614929199, + "num_tokens": 722017301.0, + "step": 18600 + }, + { + "epoch": 2.3662383920620784, + "grad_norm": 3.076845407485962, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9088022112846375, + "num_tokens": 722051600.0, + "step": 18601 + }, + { + "epoch": 2.366365602340669, + "grad_norm": 3.2860829830169678, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8976039290428162, + "num_tokens": 722084721.0, + "step": 18602 + }, + { + "epoch": 2.3664928126192595, + "grad_norm": 3.1524581909179688, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8910788297653198, + "num_tokens": 722120826.0, + "step": 18603 + }, + { + "epoch": 2.36662002289785, + "grad_norm": 2.6689207553863525, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.906426191329956, + "num_tokens": 722165521.0, + "step": 18604 + }, + { + "epoch": 2.3667472331764405, + "grad_norm": 2.729015588760376, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.906456470489502, + "num_tokens": 722209419.0, + "step": 18605 + }, + { + "epoch": 2.366874443455031, + "grad_norm": 2.882685661315918, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.899729311466217, + "num_tokens": 722247393.0, + "step": 18606 + }, + { + "epoch": 2.3670016537336216, + "grad_norm": 2.9289822578430176, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9021504521369934, + "num_tokens": 722285963.0, + "step": 18607 + }, + { + "epoch": 2.367128864012212, + "grad_norm": 3.081502676010132, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8978078365325928, + "num_tokens": 722321639.0, + "step": 18608 + }, + { + "epoch": 2.3672560742908026, + "grad_norm": 3.0024824142456055, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9098122119903564, + "num_tokens": 722357004.0, + "step": 18609 + }, + { + "epoch": 2.367383284569393, + "grad_norm": 2.712219715118408, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9095380306243896, + "num_tokens": 722394926.0, + "step": 18610 + }, + { + "epoch": 2.3675104948479837, + "grad_norm": 2.9109818935394287, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9088797569274902, + "num_tokens": 722433886.0, + "step": 18611 + }, + { + "epoch": 2.3676377051265742, + "grad_norm": 2.5475986003875732, + "learning_rate": 1e-06, + "loss": 0.2217, + "mean_token_accuracy": 0.9162386655807495, + "num_tokens": 722476752.0, + "step": 18612 + }, + { + "epoch": 2.3677649154051648, + "grad_norm": 3.098693370819092, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8993125557899475, + "num_tokens": 722511936.0, + "step": 18613 + }, + { + "epoch": 2.3678921256837553, + "grad_norm": 3.0675716400146484, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8967984318733215, + "num_tokens": 722550521.0, + "step": 18614 + }, + { + "epoch": 2.368019335962346, + "grad_norm": 2.9917614459991455, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9108226299285889, + "num_tokens": 722585262.0, + "step": 18615 + }, + { + "epoch": 2.3681465462409363, + "grad_norm": 2.744983196258545, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9065520763397217, + "num_tokens": 722626024.0, + "step": 18616 + }, + { + "epoch": 2.368273756519527, + "grad_norm": 2.927804470062256, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9055715203285217, + "num_tokens": 722662770.0, + "step": 18617 + }, + { + "epoch": 2.3684009667981174, + "grad_norm": 2.7378830909729004, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9101325273513794, + "num_tokens": 722698373.0, + "step": 18618 + }, + { + "epoch": 2.368528177076708, + "grad_norm": 3.039950370788574, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9126964807510376, + "num_tokens": 722729418.0, + "step": 18619 + }, + { + "epoch": 2.3686553873552985, + "grad_norm": 2.8457789421081543, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9071992635726929, + "num_tokens": 722767854.0, + "step": 18620 + }, + { + "epoch": 2.368782597633889, + "grad_norm": 2.7753164768218994, + "learning_rate": 1e-06, + "loss": 0.222, + "mean_token_accuracy": 0.9164008498191833, + "num_tokens": 722804494.0, + "step": 18621 + }, + { + "epoch": 2.3689098079124795, + "grad_norm": 2.576338052749634, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9115217924118042, + "num_tokens": 722844550.0, + "step": 18622 + }, + { + "epoch": 2.3690370181910696, + "grad_norm": 2.768585443496704, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9038376212120056, + "num_tokens": 722884076.0, + "step": 18623 + }, + { + "epoch": 2.3691642284696606, + "grad_norm": 2.8729608058929443, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.8998890519142151, + "num_tokens": 722922420.0, + "step": 18624 + }, + { + "epoch": 2.3692914387482507, + "grad_norm": 2.976682186126709, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9081423282623291, + "num_tokens": 722959839.0, + "step": 18625 + }, + { + "epoch": 2.369418649026841, + "grad_norm": 2.8981308937072754, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9066689014434814, + "num_tokens": 722996711.0, + "step": 18626 + }, + { + "epoch": 2.3695458593054317, + "grad_norm": 3.1206889152526855, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9038369655609131, + "num_tokens": 723033424.0, + "step": 18627 + }, + { + "epoch": 2.3696730695840222, + "grad_norm": 2.910139322280884, + "learning_rate": 1e-06, + "loss": 0.2277, + "mean_token_accuracy": 0.9155327081680298, + "num_tokens": 723067612.0, + "step": 18628 + }, + { + "epoch": 2.3698002798626128, + "grad_norm": 2.912529706954956, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.909634530544281, + "num_tokens": 723106770.0, + "step": 18629 + }, + { + "epoch": 2.3699274901412033, + "grad_norm": 2.992795467376709, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9008768796920776, + "num_tokens": 723146209.0, + "step": 18630 + }, + { + "epoch": 2.370054700419794, + "grad_norm": 2.733208417892456, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.9141656756401062, + "num_tokens": 723183747.0, + "step": 18631 + }, + { + "epoch": 2.3701819106983844, + "grad_norm": 2.9314377307891846, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9050861597061157, + "num_tokens": 723221572.0, + "step": 18632 + }, + { + "epoch": 2.370309120976975, + "grad_norm": 3.005575180053711, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8957900404930115, + "num_tokens": 723261339.0, + "step": 18633 + }, + { + "epoch": 2.3704363312555654, + "grad_norm": 2.9516537189483643, + "learning_rate": 1e-06, + "loss": 0.2392, + "mean_token_accuracy": 0.912929892539978, + "num_tokens": 723298727.0, + "step": 18634 + }, + { + "epoch": 2.370563541534156, + "grad_norm": 2.754329204559326, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8976205587387085, + "num_tokens": 723342140.0, + "step": 18635 + }, + { + "epoch": 2.3706907518127465, + "grad_norm": 2.8953051567077637, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9046343564987183, + "num_tokens": 723380791.0, + "step": 18636 + }, + { + "epoch": 2.370817962091337, + "grad_norm": 3.0423080921173096, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9075608253479004, + "num_tokens": 723414372.0, + "step": 18637 + }, + { + "epoch": 2.3709451723699275, + "grad_norm": 2.785571575164795, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8984256982803345, + "num_tokens": 723456937.0, + "step": 18638 + }, + { + "epoch": 2.371072382648518, + "grad_norm": 2.829846143722534, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9113726615905762, + "num_tokens": 723494712.0, + "step": 18639 + }, + { + "epoch": 2.3711995929271086, + "grad_norm": 2.6165919303894043, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.909172773361206, + "num_tokens": 723535508.0, + "step": 18640 + }, + { + "epoch": 2.371326803205699, + "grad_norm": 3.2413928508758545, + "learning_rate": 1e-06, + "loss": 0.3279, + "mean_token_accuracy": 0.8814927339553833, + "num_tokens": 723574585.0, + "step": 18641 + }, + { + "epoch": 2.3714540134842896, + "grad_norm": 3.1585593223571777, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9093252420425415, + "num_tokens": 723608581.0, + "step": 18642 + }, + { + "epoch": 2.37158122376288, + "grad_norm": 2.6921420097351074, + "learning_rate": 1e-06, + "loss": 0.2066, + "mean_token_accuracy": 0.9228555560112, + "num_tokens": 723649710.0, + "step": 18643 + }, + { + "epoch": 2.3717084340414707, + "grad_norm": 2.8932182788848877, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8967071771621704, + "num_tokens": 723690510.0, + "step": 18644 + }, + { + "epoch": 2.371835644320061, + "grad_norm": 3.0878002643585205, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9033218622207642, + "num_tokens": 723726621.0, + "step": 18645 + }, + { + "epoch": 2.3719628545986517, + "grad_norm": 2.6610331535339355, + "learning_rate": 1e-06, + "loss": 0.223, + "mean_token_accuracy": 0.9171019792556763, + "num_tokens": 723767913.0, + "step": 18646 + }, + { + "epoch": 2.3720900648772423, + "grad_norm": 2.813873529434204, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9018563032150269, + "num_tokens": 723808845.0, + "step": 18647 + }, + { + "epoch": 2.3722172751558324, + "grad_norm": 2.8934950828552246, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9018231630325317, + "num_tokens": 723850314.0, + "step": 18648 + }, + { + "epoch": 2.3723444854344233, + "grad_norm": 2.957388162612915, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9059491753578186, + "num_tokens": 723886145.0, + "step": 18649 + }, + { + "epoch": 2.3724716957130134, + "grad_norm": 2.7131736278533936, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.9068626165390015, + "num_tokens": 723928176.0, + "step": 18650 + }, + { + "epoch": 2.372598905991604, + "grad_norm": 2.972153902053833, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9010549187660217, + "num_tokens": 723968475.0, + "step": 18651 + }, + { + "epoch": 2.3727261162701945, + "grad_norm": 3.283196210861206, + "learning_rate": 1e-06, + "loss": 0.2413, + "mean_token_accuracy": 0.9093847274780273, + "num_tokens": 723997870.0, + "step": 18652 + }, + { + "epoch": 2.372853326548785, + "grad_norm": 2.5533549785614014, + "learning_rate": 1e-06, + "loss": 0.2191, + "mean_token_accuracy": 0.9185134768486023, + "num_tokens": 724038740.0, + "step": 18653 + }, + { + "epoch": 2.3729805368273755, + "grad_norm": 2.982795238494873, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.900811493396759, + "num_tokens": 724079795.0, + "step": 18654 + }, + { + "epoch": 2.373107747105966, + "grad_norm": 2.9610636234283447, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9057909250259399, + "num_tokens": 724116377.0, + "step": 18655 + }, + { + "epoch": 2.3732349573845566, + "grad_norm": 2.787569046020508, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8999652862548828, + "num_tokens": 724161045.0, + "step": 18656 + }, + { + "epoch": 2.373362167663147, + "grad_norm": 2.6969707012176514, + "learning_rate": 1e-06, + "loss": 0.2318, + "mean_token_accuracy": 0.9148459434509277, + "num_tokens": 724202802.0, + "step": 18657 + }, + { + "epoch": 2.3734893779417376, + "grad_norm": 2.7908105850219727, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9061068892478943, + "num_tokens": 724243903.0, + "step": 18658 + }, + { + "epoch": 2.373616588220328, + "grad_norm": 2.613063097000122, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9036312103271484, + "num_tokens": 724291666.0, + "step": 18659 + }, + { + "epoch": 2.3737437984989187, + "grad_norm": 3.081587076187134, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8948668241500854, + "num_tokens": 724330118.0, + "step": 18660 + }, + { + "epoch": 2.3738710087775092, + "grad_norm": 2.7844667434692383, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9040212631225586, + "num_tokens": 724370131.0, + "step": 18661 + }, + { + "epoch": 2.3739982190560998, + "grad_norm": 2.8693959712982178, + "learning_rate": 1e-06, + "loss": 0.2177, + "mean_token_accuracy": 0.9193808436393738, + "num_tokens": 724403451.0, + "step": 18662 + }, + { + "epoch": 2.3741254293346903, + "grad_norm": 3.2260031700134277, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.890133261680603, + "num_tokens": 724441835.0, + "step": 18663 + }, + { + "epoch": 2.374252639613281, + "grad_norm": 2.8561148643493652, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.912529468536377, + "num_tokens": 724480089.0, + "step": 18664 + }, + { + "epoch": 2.3743798498918713, + "grad_norm": 2.8433806896209717, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.902569055557251, + "num_tokens": 724518385.0, + "step": 18665 + }, + { + "epoch": 2.374507060170462, + "grad_norm": 2.8539140224456787, + "learning_rate": 1e-06, + "loss": 0.2221, + "mean_token_accuracy": 0.9172054529190063, + "num_tokens": 724554078.0, + "step": 18666 + }, + { + "epoch": 2.3746342704490524, + "grad_norm": 2.9710886478424072, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9071614742279053, + "num_tokens": 724594005.0, + "step": 18667 + }, + { + "epoch": 2.374761480727643, + "grad_norm": 3.140947103500366, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8965145945549011, + "num_tokens": 724627151.0, + "step": 18668 + }, + { + "epoch": 2.3748886910062335, + "grad_norm": 2.699873208999634, + "learning_rate": 1e-06, + "loss": 0.2303, + "mean_token_accuracy": 0.9145569801330566, + "num_tokens": 724666192.0, + "step": 18669 + }, + { + "epoch": 2.375015901284824, + "grad_norm": 2.7609498500823975, + "learning_rate": 1e-06, + "loss": 0.2066, + "mean_token_accuracy": 0.9205647706985474, + "num_tokens": 724699125.0, + "step": 18670 + }, + { + "epoch": 2.3751431115634145, + "grad_norm": 2.8561789989471436, + "learning_rate": 1e-06, + "loss": 0.2443, + "mean_token_accuracy": 0.9060338139533997, + "num_tokens": 724734832.0, + "step": 18671 + }, + { + "epoch": 2.375270321842005, + "grad_norm": 3.023059606552124, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9005569815635681, + "num_tokens": 724770178.0, + "step": 18672 + }, + { + "epoch": 2.375397532120595, + "grad_norm": 2.7476277351379395, + "learning_rate": 1e-06, + "loss": 0.2306, + "mean_token_accuracy": 0.9118923544883728, + "num_tokens": 724807877.0, + "step": 18673 + }, + { + "epoch": 2.375524742399186, + "grad_norm": 2.7340803146362305, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9052473902702332, + "num_tokens": 724846216.0, + "step": 18674 + }, + { + "epoch": 2.375651952677776, + "grad_norm": 2.984705686569214, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.9012982845306396, + "num_tokens": 724882480.0, + "step": 18675 + }, + { + "epoch": 2.3757791629563667, + "grad_norm": 2.9313151836395264, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8970212936401367, + "num_tokens": 724923865.0, + "step": 18676 + }, + { + "epoch": 2.3759063732349572, + "grad_norm": 2.7128865718841553, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9085780382156372, + "num_tokens": 724966393.0, + "step": 18677 + }, + { + "epoch": 2.3760335835135478, + "grad_norm": 3.116398572921753, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9058432579040527, + "num_tokens": 724999293.0, + "step": 18678 + }, + { + "epoch": 2.3761607937921383, + "grad_norm": 2.795605182647705, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9110054969787598, + "num_tokens": 725037435.0, + "step": 18679 + }, + { + "epoch": 2.376288004070729, + "grad_norm": 3.0125372409820557, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9028409719467163, + "num_tokens": 725074646.0, + "step": 18680 + }, + { + "epoch": 2.3764152143493193, + "grad_norm": 3.0870325565338135, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9050711989402771, + "num_tokens": 725109772.0, + "step": 18681 + }, + { + "epoch": 2.37654242462791, + "grad_norm": 2.951899290084839, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8982462882995605, + "num_tokens": 725151028.0, + "step": 18682 + }, + { + "epoch": 2.3766696349065004, + "grad_norm": 3.1284358501434326, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9111095666885376, + "num_tokens": 725184458.0, + "step": 18683 + }, + { + "epoch": 2.376796845185091, + "grad_norm": 2.813718795776367, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9125157594680786, + "num_tokens": 725221297.0, + "step": 18684 + }, + { + "epoch": 2.3769240554636815, + "grad_norm": 2.871788263320923, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.899534285068512, + "num_tokens": 725264556.0, + "step": 18685 + }, + { + "epoch": 2.377051265742272, + "grad_norm": 2.7290844917297363, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9137713313102722, + "num_tokens": 725305559.0, + "step": 18686 + }, + { + "epoch": 2.3771784760208625, + "grad_norm": 2.9753241539001465, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.8994625806808472, + "num_tokens": 725346007.0, + "step": 18687 + }, + { + "epoch": 2.377305686299453, + "grad_norm": 3.0425796508789062, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.89923095703125, + "num_tokens": 725384755.0, + "step": 18688 + }, + { + "epoch": 2.3774328965780436, + "grad_norm": 2.610503911972046, + "learning_rate": 1e-06, + "loss": 0.2176, + "mean_token_accuracy": 0.9192473888397217, + "num_tokens": 725425511.0, + "step": 18689 + }, + { + "epoch": 2.377560106856634, + "grad_norm": 2.976041555404663, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9022903442382812, + "num_tokens": 725465189.0, + "step": 18690 + }, + { + "epoch": 2.3776873171352246, + "grad_norm": 2.978750228881836, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9069075584411621, + "num_tokens": 725502162.0, + "step": 18691 + }, + { + "epoch": 2.377814527413815, + "grad_norm": 2.7678325176239014, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9028534293174744, + "num_tokens": 725543507.0, + "step": 18692 + }, + { + "epoch": 2.3779417376924057, + "grad_norm": 2.969393730163574, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.889408528804779, + "num_tokens": 725587535.0, + "step": 18693 + }, + { + "epoch": 2.378068947970996, + "grad_norm": 2.96687912940979, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.8994406461715698, + "num_tokens": 725627518.0, + "step": 18694 + }, + { + "epoch": 2.3781961582495867, + "grad_norm": 2.6676549911499023, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.8991005420684814, + "num_tokens": 725675522.0, + "step": 18695 + }, + { + "epoch": 2.378323368528177, + "grad_norm": 2.733168601989746, + "learning_rate": 1e-06, + "loss": 0.242, + "mean_token_accuracy": 0.9094055891036987, + "num_tokens": 725713871.0, + "step": 18696 + }, + { + "epoch": 2.378450578806768, + "grad_norm": 2.7722015380859375, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9121111035346985, + "num_tokens": 725751415.0, + "step": 18697 + }, + { + "epoch": 2.378577789085358, + "grad_norm": 2.9176650047302246, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.912891149520874, + "num_tokens": 725787728.0, + "step": 18698 + }, + { + "epoch": 2.3787049993639484, + "grad_norm": 2.764498233795166, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9068078994750977, + "num_tokens": 725827194.0, + "step": 18699 + }, + { + "epoch": 2.378832209642539, + "grad_norm": 2.9745073318481445, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.907207727432251, + "num_tokens": 725862571.0, + "step": 18700 + }, + { + "epoch": 2.3789594199211295, + "grad_norm": 2.6472878456115723, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9076100587844849, + "num_tokens": 725905144.0, + "step": 18701 + }, + { + "epoch": 2.37908663019972, + "grad_norm": 2.9801411628723145, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9034913778305054, + "num_tokens": 725942017.0, + "step": 18702 + }, + { + "epoch": 2.3792138404783105, + "grad_norm": 2.8122990131378174, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9115664958953857, + "num_tokens": 725980429.0, + "step": 18703 + }, + { + "epoch": 2.379341050756901, + "grad_norm": 2.772238254547119, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9112904667854309, + "num_tokens": 726017741.0, + "step": 18704 + }, + { + "epoch": 2.3794682610354916, + "grad_norm": 2.9480745792388916, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9024364948272705, + "num_tokens": 726051416.0, + "step": 18705 + }, + { + "epoch": 2.379595471314082, + "grad_norm": 2.9672117233276367, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.9117337465286255, + "num_tokens": 726084618.0, + "step": 18706 + }, + { + "epoch": 2.3797226815926726, + "grad_norm": 2.6780009269714355, + "learning_rate": 1e-06, + "loss": 0.2232, + "mean_token_accuracy": 0.9181926250457764, + "num_tokens": 726120555.0, + "step": 18707 + }, + { + "epoch": 2.379849891871263, + "grad_norm": 3.06591534614563, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8923872709274292, + "num_tokens": 726160370.0, + "step": 18708 + }, + { + "epoch": 2.3799771021498537, + "grad_norm": 3.3602781295776367, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9009337425231934, + "num_tokens": 726191610.0, + "step": 18709 + }, + { + "epoch": 2.3801043124284442, + "grad_norm": 2.673768997192383, + "learning_rate": 1e-06, + "loss": 0.2318, + "mean_token_accuracy": 0.9140161275863647, + "num_tokens": 726231422.0, + "step": 18710 + }, + { + "epoch": 2.3802315227070348, + "grad_norm": 2.7230396270751953, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9071428775787354, + "num_tokens": 726273674.0, + "step": 18711 + }, + { + "epoch": 2.3803587329856253, + "grad_norm": 2.6760671138763428, + "learning_rate": 1e-06, + "loss": 0.217, + "mean_token_accuracy": 0.9164876937866211, + "num_tokens": 726311051.0, + "step": 18712 + }, + { + "epoch": 2.380485943264216, + "grad_norm": 2.7854206562042236, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9092646241188049, + "num_tokens": 726352773.0, + "step": 18713 + }, + { + "epoch": 2.3806131535428063, + "grad_norm": 2.979308843612671, + "learning_rate": 1e-06, + "loss": 0.2337, + "mean_token_accuracy": 0.9136060476303101, + "num_tokens": 726386748.0, + "step": 18714 + }, + { + "epoch": 2.380740363821397, + "grad_norm": 3.2088661193847656, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9033122062683105, + "num_tokens": 726420224.0, + "step": 18715 + }, + { + "epoch": 2.3808675740999874, + "grad_norm": 2.912484645843506, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8931283950805664, + "num_tokens": 726462320.0, + "step": 18716 + }, + { + "epoch": 2.380994784378578, + "grad_norm": 3.13191819190979, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8967248797416687, + "num_tokens": 726499385.0, + "step": 18717 + }, + { + "epoch": 2.3811219946571685, + "grad_norm": 2.912280321121216, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9069397449493408, + "num_tokens": 726538631.0, + "step": 18718 + }, + { + "epoch": 2.381249204935759, + "grad_norm": 3.106525182723999, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9022753238677979, + "num_tokens": 726575226.0, + "step": 18719 + }, + { + "epoch": 2.3813764152143495, + "grad_norm": 2.7571933269500732, + "learning_rate": 1e-06, + "loss": 0.2432, + "mean_token_accuracy": 0.9086587429046631, + "num_tokens": 726614057.0, + "step": 18720 + }, + { + "epoch": 2.3815036254929396, + "grad_norm": 2.754647970199585, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.907279372215271, + "num_tokens": 726655063.0, + "step": 18721 + }, + { + "epoch": 2.3816308357715306, + "grad_norm": 2.7087035179138184, + "learning_rate": 1e-06, + "loss": 0.2261, + "mean_token_accuracy": 0.91583251953125, + "num_tokens": 726693856.0, + "step": 18722 + }, + { + "epoch": 2.3817580460501206, + "grad_norm": 2.494818687438965, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9143958687782288, + "num_tokens": 726741456.0, + "step": 18723 + }, + { + "epoch": 2.381885256328711, + "grad_norm": 3.1061503887176514, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9075251817703247, + "num_tokens": 726776771.0, + "step": 18724 + }, + { + "epoch": 2.3820124666073017, + "grad_norm": 3.1529037952423096, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9047256708145142, + "num_tokens": 726812053.0, + "step": 18725 + }, + { + "epoch": 2.3821396768858922, + "grad_norm": 3.030829906463623, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8970912098884583, + "num_tokens": 726850369.0, + "step": 18726 + }, + { + "epoch": 2.3822668871644828, + "grad_norm": 3.061220645904541, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.8995357751846313, + "num_tokens": 726887083.0, + "step": 18727 + }, + { + "epoch": 2.3823940974430733, + "grad_norm": 2.7941670417785645, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9051934480667114, + "num_tokens": 726926149.0, + "step": 18728 + }, + { + "epoch": 2.382521307721664, + "grad_norm": 3.128109931945801, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.9002136588096619, + "num_tokens": 726964089.0, + "step": 18729 + }, + { + "epoch": 2.3826485180002543, + "grad_norm": 3.2504494190216064, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8897020816802979, + "num_tokens": 726998640.0, + "step": 18730 + }, + { + "epoch": 2.382775728278845, + "grad_norm": 3.3146071434020996, + "learning_rate": 1e-06, + "loss": 0.2973, + "mean_token_accuracy": 0.8926774263381958, + "num_tokens": 727034979.0, + "step": 18731 + }, + { + "epoch": 2.3829029385574354, + "grad_norm": 2.8252665996551514, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9049511551856995, + "num_tokens": 727079804.0, + "step": 18732 + }, + { + "epoch": 2.383030148836026, + "grad_norm": 2.8108773231506348, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8911560773849487, + "num_tokens": 727121290.0, + "step": 18733 + }, + { + "epoch": 2.3831573591146165, + "grad_norm": 2.859139919281006, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9060285091400146, + "num_tokens": 727158721.0, + "step": 18734 + }, + { + "epoch": 2.383284569393207, + "grad_norm": 2.816009283065796, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.910813570022583, + "num_tokens": 727197511.0, + "step": 18735 + }, + { + "epoch": 2.3834117796717975, + "grad_norm": 2.8868865966796875, + "learning_rate": 1e-06, + "loss": 0.2393, + "mean_token_accuracy": 0.9105483293533325, + "num_tokens": 727233068.0, + "step": 18736 + }, + { + "epoch": 2.383538989950388, + "grad_norm": 3.0945663452148438, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9018957018852234, + "num_tokens": 727267897.0, + "step": 18737 + }, + { + "epoch": 2.3836662002289786, + "grad_norm": 2.8110482692718506, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8931936025619507, + "num_tokens": 727310591.0, + "step": 18738 + }, + { + "epoch": 2.383793410507569, + "grad_norm": 2.5963785648345947, + "learning_rate": 1e-06, + "loss": 0.2249, + "mean_token_accuracy": 0.9165347814559937, + "num_tokens": 727351471.0, + "step": 18739 + }, + { + "epoch": 2.3839206207861596, + "grad_norm": 2.8097734451293945, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.906034529209137, + "num_tokens": 727391108.0, + "step": 18740 + }, + { + "epoch": 2.38404783106475, + "grad_norm": 2.9852702617645264, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9030683040618896, + "num_tokens": 727428549.0, + "step": 18741 + }, + { + "epoch": 2.3841750413433407, + "grad_norm": 2.958564281463623, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.9082794785499573, + "num_tokens": 727468207.0, + "step": 18742 + }, + { + "epoch": 2.384302251621931, + "grad_norm": 3.038505792617798, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9089332222938538, + "num_tokens": 727503941.0, + "step": 18743 + }, + { + "epoch": 2.3844294619005217, + "grad_norm": 2.7971906661987305, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9101099967956543, + "num_tokens": 727545286.0, + "step": 18744 + }, + { + "epoch": 2.3845566721791123, + "grad_norm": 2.76401686668396, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9035425782203674, + "num_tokens": 727590743.0, + "step": 18745 + }, + { + "epoch": 2.3846838824577024, + "grad_norm": 3.1256299018859863, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8911054134368896, + "num_tokens": 727627415.0, + "step": 18746 + }, + { + "epoch": 2.3848110927362933, + "grad_norm": 2.906200885772705, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9060021042823792, + "num_tokens": 727668410.0, + "step": 18747 + }, + { + "epoch": 2.3849383030148834, + "grad_norm": 2.8450732231140137, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9004621505737305, + "num_tokens": 727707762.0, + "step": 18748 + }, + { + "epoch": 2.385065513293474, + "grad_norm": 2.7631256580352783, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9035755395889282, + "num_tokens": 727747853.0, + "step": 18749 + }, + { + "epoch": 2.3851927235720645, + "grad_norm": 2.894026041030884, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9033882021903992, + "num_tokens": 727789314.0, + "step": 18750 + }, + { + "epoch": 2.385319933850655, + "grad_norm": 2.849644422531128, + "learning_rate": 1e-06, + "loss": 0.2316, + "mean_token_accuracy": 0.9131293892860413, + "num_tokens": 727825132.0, + "step": 18751 + }, + { + "epoch": 2.3854471441292455, + "grad_norm": 3.009594678878784, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8980432152748108, + "num_tokens": 727863265.0, + "step": 18752 + }, + { + "epoch": 2.385574354407836, + "grad_norm": 2.4947993755340576, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.9108816385269165, + "num_tokens": 727909749.0, + "step": 18753 + }, + { + "epoch": 2.3857015646864266, + "grad_norm": 2.7427375316619873, + "learning_rate": 1e-06, + "loss": 0.215, + "mean_token_accuracy": 0.9215861558914185, + "num_tokens": 727945576.0, + "step": 18754 + }, + { + "epoch": 2.385828774965017, + "grad_norm": 3.0123648643493652, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.899170994758606, + "num_tokens": 727985454.0, + "step": 18755 + }, + { + "epoch": 2.3859559852436076, + "grad_norm": 2.9036307334899902, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.900881826877594, + "num_tokens": 728025106.0, + "step": 18756 + }, + { + "epoch": 2.386083195522198, + "grad_norm": 2.894496202468872, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8926931619644165, + "num_tokens": 728072430.0, + "step": 18757 + }, + { + "epoch": 2.3862104058007887, + "grad_norm": 2.9570224285125732, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.906936526298523, + "num_tokens": 728107251.0, + "step": 18758 + }, + { + "epoch": 2.386337616079379, + "grad_norm": 2.5035479068756104, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9062072038650513, + "num_tokens": 728156155.0, + "step": 18759 + }, + { + "epoch": 2.3864648263579697, + "grad_norm": 2.79823637008667, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9060352444648743, + "num_tokens": 728200347.0, + "step": 18760 + }, + { + "epoch": 2.3865920366365603, + "grad_norm": 2.9642369747161865, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9026665687561035, + "num_tokens": 728236727.0, + "step": 18761 + }, + { + "epoch": 2.386719246915151, + "grad_norm": 2.857131242752075, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.910115122795105, + "num_tokens": 728276151.0, + "step": 18762 + }, + { + "epoch": 2.3868464571937413, + "grad_norm": 2.840618848800659, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9017010927200317, + "num_tokens": 728318984.0, + "step": 18763 + }, + { + "epoch": 2.386973667472332, + "grad_norm": 2.9153575897216797, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9073665142059326, + "num_tokens": 728355094.0, + "step": 18764 + }, + { + "epoch": 2.3871008777509224, + "grad_norm": 3.1741418838500977, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9014671444892883, + "num_tokens": 728388371.0, + "step": 18765 + }, + { + "epoch": 2.387228088029513, + "grad_norm": 2.9053938388824463, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.904503583908081, + "num_tokens": 728425238.0, + "step": 18766 + }, + { + "epoch": 2.3873552983081034, + "grad_norm": 2.7236430644989014, + "learning_rate": 1e-06, + "loss": 0.2224, + "mean_token_accuracy": 0.9152494668960571, + "num_tokens": 728463211.0, + "step": 18767 + }, + { + "epoch": 2.387482508586694, + "grad_norm": 2.6427760124206543, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9010679721832275, + "num_tokens": 728511661.0, + "step": 18768 + }, + { + "epoch": 2.3876097188652845, + "grad_norm": 2.87723970413208, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9068689346313477, + "num_tokens": 728549986.0, + "step": 18769 + }, + { + "epoch": 2.387736929143875, + "grad_norm": 3.1530869007110596, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8988956212997437, + "num_tokens": 728587056.0, + "step": 18770 + }, + { + "epoch": 2.387864139422465, + "grad_norm": 2.711915969848633, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8981961607933044, + "num_tokens": 728632611.0, + "step": 18771 + }, + { + "epoch": 2.387991349701056, + "grad_norm": 2.913193702697754, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9065432548522949, + "num_tokens": 728670430.0, + "step": 18772 + }, + { + "epoch": 2.388118559979646, + "grad_norm": 3.1702048778533936, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8954178094863892, + "num_tokens": 728707774.0, + "step": 18773 + }, + { + "epoch": 2.3882457702582367, + "grad_norm": 2.9133849143981934, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8911809325218201, + "num_tokens": 728750025.0, + "step": 18774 + }, + { + "epoch": 2.3883729805368272, + "grad_norm": 3.0832061767578125, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.9006586074829102, + "num_tokens": 728784657.0, + "step": 18775 + }, + { + "epoch": 2.3885001908154178, + "grad_norm": 2.9992470741271973, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8926562666893005, + "num_tokens": 728823699.0, + "step": 18776 + }, + { + "epoch": 2.3886274010940083, + "grad_norm": 2.90295147895813, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9028139114379883, + "num_tokens": 728861790.0, + "step": 18777 + }, + { + "epoch": 2.388754611372599, + "grad_norm": 2.7869832515716553, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9157352447509766, + "num_tokens": 728900266.0, + "step": 18778 + }, + { + "epoch": 2.3888818216511893, + "grad_norm": 2.938598155975342, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.900955319404602, + "num_tokens": 728936145.0, + "step": 18779 + }, + { + "epoch": 2.38900903192978, + "grad_norm": 2.986599922180176, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9024240374565125, + "num_tokens": 728975044.0, + "step": 18780 + }, + { + "epoch": 2.3891362422083704, + "grad_norm": 2.6358916759490967, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9043786525726318, + "num_tokens": 729018385.0, + "step": 18781 + }, + { + "epoch": 2.389263452486961, + "grad_norm": 2.7635934352874756, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9020770192146301, + "num_tokens": 729062264.0, + "step": 18782 + }, + { + "epoch": 2.3893906627655515, + "grad_norm": 3.261187791824341, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9047626852989197, + "num_tokens": 729092269.0, + "step": 18783 + }, + { + "epoch": 2.389517873044142, + "grad_norm": 3.4859747886657715, + "learning_rate": 1e-06, + "loss": 0.3259, + "mean_token_accuracy": 0.8830258250236511, + "num_tokens": 729128454.0, + "step": 18784 + }, + { + "epoch": 2.3896450833227325, + "grad_norm": 3.374770164489746, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8938760757446289, + "num_tokens": 729160659.0, + "step": 18785 + }, + { + "epoch": 2.389772293601323, + "grad_norm": 3.0309760570526123, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9038352966308594, + "num_tokens": 729196875.0, + "step": 18786 + }, + { + "epoch": 2.3898995038799136, + "grad_norm": 2.7471814155578613, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9095655083656311, + "num_tokens": 729240792.0, + "step": 18787 + }, + { + "epoch": 2.390026714158504, + "grad_norm": 2.7848103046417236, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9054457545280457, + "num_tokens": 729279087.0, + "step": 18788 + }, + { + "epoch": 2.3901539244370946, + "grad_norm": 2.786548376083374, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.916201114654541, + "num_tokens": 729317885.0, + "step": 18789 + }, + { + "epoch": 2.390281134715685, + "grad_norm": 2.5956218242645264, + "learning_rate": 1e-06, + "loss": 0.2322, + "mean_token_accuracy": 0.9125293493270874, + "num_tokens": 729360503.0, + "step": 18790 + }, + { + "epoch": 2.3904083449942757, + "grad_norm": 3.205623149871826, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9012008905410767, + "num_tokens": 729397362.0, + "step": 18791 + }, + { + "epoch": 2.390535555272866, + "grad_norm": 3.0325398445129395, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9077587127685547, + "num_tokens": 729433050.0, + "step": 18792 + }, + { + "epoch": 2.3906627655514567, + "grad_norm": 2.689617395401001, + "learning_rate": 1e-06, + "loss": 0.2437, + "mean_token_accuracy": 0.9086984992027283, + "num_tokens": 729471563.0, + "step": 18793 + }, + { + "epoch": 2.390789975830047, + "grad_norm": 2.9012534618377686, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9043364524841309, + "num_tokens": 729509147.0, + "step": 18794 + }, + { + "epoch": 2.390917186108638, + "grad_norm": 2.769050359725952, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9111654758453369, + "num_tokens": 729548336.0, + "step": 18795 + }, + { + "epoch": 2.391044396387228, + "grad_norm": 2.954765796661377, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9030489921569824, + "num_tokens": 729587625.0, + "step": 18796 + }, + { + "epoch": 2.3911716066658184, + "grad_norm": 2.9419002532958984, + "learning_rate": 1e-06, + "loss": 0.2091, + "mean_token_accuracy": 0.9211475849151611, + "num_tokens": 729622371.0, + "step": 18797 + }, + { + "epoch": 2.391298816944409, + "grad_norm": 2.982658863067627, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8969418406486511, + "num_tokens": 729662067.0, + "step": 18798 + }, + { + "epoch": 2.3914260272229995, + "grad_norm": 3.1813957691192627, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9080313444137573, + "num_tokens": 729693107.0, + "step": 18799 + }, + { + "epoch": 2.39155323750159, + "grad_norm": 3.0571582317352295, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8914559483528137, + "num_tokens": 729735074.0, + "step": 18800 + }, + { + "epoch": 2.3916804477801805, + "grad_norm": 3.034224271774292, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.8977702856063843, + "num_tokens": 729773631.0, + "step": 18801 + }, + { + "epoch": 2.391807658058771, + "grad_norm": 2.935912847518921, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9003484845161438, + "num_tokens": 729813151.0, + "step": 18802 + }, + { + "epoch": 2.3919348683373616, + "grad_norm": 2.770625591278076, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9092835783958435, + "num_tokens": 729852365.0, + "step": 18803 + }, + { + "epoch": 2.392062078615952, + "grad_norm": 2.9331746101379395, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9106107950210571, + "num_tokens": 729888241.0, + "step": 18804 + }, + { + "epoch": 2.3921892888945426, + "grad_norm": 2.6240546703338623, + "learning_rate": 1e-06, + "loss": 0.2411, + "mean_token_accuracy": 0.910024881362915, + "num_tokens": 729930628.0, + "step": 18805 + }, + { + "epoch": 2.392316499173133, + "grad_norm": 2.8715338706970215, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.910463273525238, + "num_tokens": 729966677.0, + "step": 18806 + }, + { + "epoch": 2.3924437094517237, + "grad_norm": 2.7291369438171387, + "learning_rate": 1e-06, + "loss": 0.2217, + "mean_token_accuracy": 0.9171276092529297, + "num_tokens": 730003610.0, + "step": 18807 + }, + { + "epoch": 2.392570919730314, + "grad_norm": 2.9648988246917725, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9072521924972534, + "num_tokens": 730042421.0, + "step": 18808 + }, + { + "epoch": 2.3926981300089047, + "grad_norm": 3.354768991470337, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9066667556762695, + "num_tokens": 730073439.0, + "step": 18809 + }, + { + "epoch": 2.3928253402874953, + "grad_norm": 2.9581117630004883, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9100337624549866, + "num_tokens": 730110044.0, + "step": 18810 + }, + { + "epoch": 2.392952550566086, + "grad_norm": 2.726484775543213, + "learning_rate": 1e-06, + "loss": 0.2378, + "mean_token_accuracy": 0.9124277830123901, + "num_tokens": 730151401.0, + "step": 18811 + }, + { + "epoch": 2.3930797608446763, + "grad_norm": 2.78007173538208, + "learning_rate": 1e-06, + "loss": 0.2325, + "mean_token_accuracy": 0.9152846336364746, + "num_tokens": 730189872.0, + "step": 18812 + }, + { + "epoch": 2.393206971123267, + "grad_norm": 2.725734233856201, + "learning_rate": 1e-06, + "loss": 0.2204, + "mean_token_accuracy": 0.9163799285888672, + "num_tokens": 730224714.0, + "step": 18813 + }, + { + "epoch": 2.3933341814018574, + "grad_norm": 2.786724328994751, + "learning_rate": 1e-06, + "loss": 0.2314, + "mean_token_accuracy": 0.9140263199806213, + "num_tokens": 730260208.0, + "step": 18814 + }, + { + "epoch": 2.393461391680448, + "grad_norm": 2.9434571266174316, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9044852256774902, + "num_tokens": 730299586.0, + "step": 18815 + }, + { + "epoch": 2.3935886019590384, + "grad_norm": 2.7851438522338867, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.8997638821601868, + "num_tokens": 730342832.0, + "step": 18816 + }, + { + "epoch": 2.393715812237629, + "grad_norm": 2.9645142555236816, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9002043008804321, + "num_tokens": 730387810.0, + "step": 18817 + }, + { + "epoch": 2.3938430225162195, + "grad_norm": 2.68915057182312, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9140496253967285, + "num_tokens": 730427909.0, + "step": 18818 + }, + { + "epoch": 2.3939702327948096, + "grad_norm": 2.9685235023498535, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9032745361328125, + "num_tokens": 730467240.0, + "step": 18819 + }, + { + "epoch": 2.3940974430734006, + "grad_norm": 2.8311893939971924, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9057374596595764, + "num_tokens": 730503964.0, + "step": 18820 + }, + { + "epoch": 2.3942246533519906, + "grad_norm": 3.151533603668213, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9061930179595947, + "num_tokens": 730536036.0, + "step": 18821 + }, + { + "epoch": 2.394351863630581, + "grad_norm": 2.6316592693328857, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.91330486536026, + "num_tokens": 730574483.0, + "step": 18822 + }, + { + "epoch": 2.3944790739091717, + "grad_norm": 2.9908533096313477, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9014899134635925, + "num_tokens": 730614987.0, + "step": 18823 + }, + { + "epoch": 2.3946062841877622, + "grad_norm": 2.5918149948120117, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.8987679481506348, + "num_tokens": 730660539.0, + "step": 18824 + }, + { + "epoch": 2.3947334944663528, + "grad_norm": 2.8917183876037598, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9065942168235779, + "num_tokens": 730696746.0, + "step": 18825 + }, + { + "epoch": 2.3948607047449433, + "grad_norm": 2.948005199432373, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9016480445861816, + "num_tokens": 730732555.0, + "step": 18826 + }, + { + "epoch": 2.394987915023534, + "grad_norm": 2.9570000171661377, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9010709524154663, + "num_tokens": 730770922.0, + "step": 18827 + }, + { + "epoch": 2.3951151253021243, + "grad_norm": 2.7865357398986816, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9070463180541992, + "num_tokens": 730810384.0, + "step": 18828 + }, + { + "epoch": 2.395242335580715, + "grad_norm": 2.661377191543579, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9026825428009033, + "num_tokens": 730854074.0, + "step": 18829 + }, + { + "epoch": 2.3953695458593054, + "grad_norm": 2.945949077606201, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9061968326568604, + "num_tokens": 730891528.0, + "step": 18830 + }, + { + "epoch": 2.395496756137896, + "grad_norm": 3.260122060775757, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8946652412414551, + "num_tokens": 730925873.0, + "step": 18831 + }, + { + "epoch": 2.3956239664164865, + "grad_norm": 3.136444091796875, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.8983224034309387, + "num_tokens": 730963191.0, + "step": 18832 + }, + { + "epoch": 2.395751176695077, + "grad_norm": 2.7886712551116943, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9117355942726135, + "num_tokens": 731000734.0, + "step": 18833 + }, + { + "epoch": 2.3958783869736675, + "grad_norm": 3.1106505393981934, + "learning_rate": 1e-06, + "loss": 0.3059, + "mean_token_accuracy": 0.8905652165412903, + "num_tokens": 731040621.0, + "step": 18834 + }, + { + "epoch": 2.396005597252258, + "grad_norm": 2.9046404361724854, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9056960344314575, + "num_tokens": 731080224.0, + "step": 18835 + }, + { + "epoch": 2.3961328075308486, + "grad_norm": 2.8527004718780518, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9107261300086975, + "num_tokens": 731116739.0, + "step": 18836 + }, + { + "epoch": 2.396260017809439, + "grad_norm": 2.86319899559021, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9057006239891052, + "num_tokens": 731154546.0, + "step": 18837 + }, + { + "epoch": 2.3963872280880296, + "grad_norm": 3.130619525909424, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8926246166229248, + "num_tokens": 731190720.0, + "step": 18838 + }, + { + "epoch": 2.39651443836662, + "grad_norm": 2.9201912879943848, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.905964732170105, + "num_tokens": 731229015.0, + "step": 18839 + }, + { + "epoch": 2.3966416486452107, + "grad_norm": 2.9456820487976074, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9054126739501953, + "num_tokens": 731265909.0, + "step": 18840 + }, + { + "epoch": 2.396768858923801, + "grad_norm": 2.943516254425049, + "learning_rate": 1e-06, + "loss": 0.2133, + "mean_token_accuracy": 0.9193437099456787, + "num_tokens": 731297879.0, + "step": 18841 + }, + { + "epoch": 2.3968960692023917, + "grad_norm": 2.698333978652954, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9033831357955933, + "num_tokens": 731340116.0, + "step": 18842 + }, + { + "epoch": 2.3970232794809823, + "grad_norm": 2.959932804107666, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9084811210632324, + "num_tokens": 731380961.0, + "step": 18843 + }, + { + "epoch": 2.3971504897595723, + "grad_norm": 2.6902167797088623, + "learning_rate": 1e-06, + "loss": 0.2146, + "mean_token_accuracy": 0.9187256693840027, + "num_tokens": 731419302.0, + "step": 18844 + }, + { + "epoch": 2.3972777000381633, + "grad_norm": 2.89807391166687, + "learning_rate": 1e-06, + "loss": 0.234, + "mean_token_accuracy": 0.9118671417236328, + "num_tokens": 731453800.0, + "step": 18845 + }, + { + "epoch": 2.3974049103167534, + "grad_norm": 2.8847503662109375, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.9133262634277344, + "num_tokens": 731488642.0, + "step": 18846 + }, + { + "epoch": 2.397532120595344, + "grad_norm": 3.0659124851226807, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8945556879043579, + "num_tokens": 731526263.0, + "step": 18847 + }, + { + "epoch": 2.3976593308739345, + "grad_norm": 2.749791145324707, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9091781377792358, + "num_tokens": 731568052.0, + "step": 18848 + }, + { + "epoch": 2.397786541152525, + "grad_norm": 2.629218578338623, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9088033437728882, + "num_tokens": 731610024.0, + "step": 18849 + }, + { + "epoch": 2.3979137514311155, + "grad_norm": 2.932427406311035, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9139257073402405, + "num_tokens": 731644521.0, + "step": 18850 + }, + { + "epoch": 2.398040961709706, + "grad_norm": 3.1931967735290527, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.8860899209976196, + "num_tokens": 731683957.0, + "step": 18851 + }, + { + "epoch": 2.3981681719882966, + "grad_norm": 2.9808571338653564, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9068781733512878, + "num_tokens": 731719564.0, + "step": 18852 + }, + { + "epoch": 2.398295382266887, + "grad_norm": 2.9194576740264893, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.911689281463623, + "num_tokens": 731758303.0, + "step": 18853 + }, + { + "epoch": 2.3984225925454776, + "grad_norm": 2.787527084350586, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.8982733488082886, + "num_tokens": 731800376.0, + "step": 18854 + }, + { + "epoch": 2.398549802824068, + "grad_norm": 2.9002039432525635, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9086965918540955, + "num_tokens": 731837936.0, + "step": 18855 + }, + { + "epoch": 2.3986770131026587, + "grad_norm": 2.81030011177063, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8998878598213196, + "num_tokens": 731881241.0, + "step": 18856 + }, + { + "epoch": 2.398804223381249, + "grad_norm": 2.828979253768921, + "learning_rate": 1e-06, + "loss": 0.237, + "mean_token_accuracy": 0.9116905927658081, + "num_tokens": 731917368.0, + "step": 18857 + }, + { + "epoch": 2.3989314336598397, + "grad_norm": 2.9581408500671387, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9019811153411865, + "num_tokens": 731955149.0, + "step": 18858 + }, + { + "epoch": 2.3990586439384303, + "grad_norm": 2.973141670227051, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9071614742279053, + "num_tokens": 731991856.0, + "step": 18859 + }, + { + "epoch": 2.399185854217021, + "grad_norm": 2.724071979522705, + "learning_rate": 1e-06, + "loss": 0.2267, + "mean_token_accuracy": 0.9139944314956665, + "num_tokens": 732029189.0, + "step": 18860 + }, + { + "epoch": 2.3993130644956113, + "grad_norm": 2.917273759841919, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9104905724525452, + "num_tokens": 732067127.0, + "step": 18861 + }, + { + "epoch": 2.399440274774202, + "grad_norm": 3.0011537075042725, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8962388038635254, + "num_tokens": 732103781.0, + "step": 18862 + }, + { + "epoch": 2.3995674850527924, + "grad_norm": 2.8321123123168945, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9022700190544128, + "num_tokens": 732149743.0, + "step": 18863 + }, + { + "epoch": 2.399694695331383, + "grad_norm": 3.361100196838379, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.8989400267601013, + "num_tokens": 732185809.0, + "step": 18864 + }, + { + "epoch": 2.3998219056099734, + "grad_norm": 2.9374966621398926, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9094024896621704, + "num_tokens": 732220047.0, + "step": 18865 + }, + { + "epoch": 2.399949115888564, + "grad_norm": 2.911888360977173, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9027401804924011, + "num_tokens": 732259882.0, + "step": 18866 + }, + { + "epoch": 2.4000763261671545, + "grad_norm": 3.048450469970703, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8972211480140686, + "num_tokens": 732296941.0, + "step": 18867 + }, + { + "epoch": 2.400203536445745, + "grad_norm": 2.639249086380005, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9086676239967346, + "num_tokens": 732340471.0, + "step": 18868 + }, + { + "epoch": 2.400330746724335, + "grad_norm": 3.3522908687591553, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8955709338188171, + "num_tokens": 732373223.0, + "step": 18869 + }, + { + "epoch": 2.400457957002926, + "grad_norm": 2.9139134883880615, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9068169593811035, + "num_tokens": 732416368.0, + "step": 18870 + }, + { + "epoch": 2.400585167281516, + "grad_norm": 2.9777188301086426, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.9128387570381165, + "num_tokens": 732451926.0, + "step": 18871 + }, + { + "epoch": 2.4007123775601067, + "grad_norm": 2.82332444190979, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9051402807235718, + "num_tokens": 732491644.0, + "step": 18872 + }, + { + "epoch": 2.400839587838697, + "grad_norm": 2.994776725769043, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9054686427116394, + "num_tokens": 732527972.0, + "step": 18873 + }, + { + "epoch": 2.4009667981172877, + "grad_norm": 2.86859393119812, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8974276781082153, + "num_tokens": 732573680.0, + "step": 18874 + }, + { + "epoch": 2.4010940083958783, + "grad_norm": 3.079188346862793, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.902422308921814, + "num_tokens": 732613518.0, + "step": 18875 + }, + { + "epoch": 2.401221218674469, + "grad_norm": 2.6211800575256348, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9090884923934937, + "num_tokens": 732651915.0, + "step": 18876 + }, + { + "epoch": 2.4013484289530593, + "grad_norm": 2.5538418292999268, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9048526287078857, + "num_tokens": 732697239.0, + "step": 18877 + }, + { + "epoch": 2.40147563923165, + "grad_norm": 2.85878849029541, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9124324321746826, + "num_tokens": 732734699.0, + "step": 18878 + }, + { + "epoch": 2.4016028495102404, + "grad_norm": 2.796527147293091, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9050473570823669, + "num_tokens": 732772052.0, + "step": 18879 + }, + { + "epoch": 2.401730059788831, + "grad_norm": 2.9021670818328857, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9030691385269165, + "num_tokens": 732813107.0, + "step": 18880 + }, + { + "epoch": 2.4018572700674214, + "grad_norm": 3.0879807472229004, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.8993343114852905, + "num_tokens": 732846626.0, + "step": 18881 + }, + { + "epoch": 2.401984480346012, + "grad_norm": 2.945142984390259, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8944608569145203, + "num_tokens": 732886169.0, + "step": 18882 + }, + { + "epoch": 2.4021116906246025, + "grad_norm": 2.804205894470215, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9051266312599182, + "num_tokens": 732926906.0, + "step": 18883 + }, + { + "epoch": 2.402238900903193, + "grad_norm": 2.6303870677948, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9068697094917297, + "num_tokens": 732968492.0, + "step": 18884 + }, + { + "epoch": 2.4023661111817836, + "grad_norm": 3.3156871795654297, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8999344110488892, + "num_tokens": 733002480.0, + "step": 18885 + }, + { + "epoch": 2.402493321460374, + "grad_norm": 2.6317007541656494, + "learning_rate": 1e-06, + "loss": 0.2321, + "mean_token_accuracy": 0.9122633934020996, + "num_tokens": 733043931.0, + "step": 18886 + }, + { + "epoch": 2.4026205317389646, + "grad_norm": 2.9915761947631836, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.8990650177001953, + "num_tokens": 733082310.0, + "step": 18887 + }, + { + "epoch": 2.402747742017555, + "grad_norm": 2.983689546585083, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9102972745895386, + "num_tokens": 733117234.0, + "step": 18888 + }, + { + "epoch": 2.4028749522961457, + "grad_norm": 2.9156553745269775, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9049162864685059, + "num_tokens": 733156646.0, + "step": 18889 + }, + { + "epoch": 2.403002162574736, + "grad_norm": 3.147503137588501, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8875280618667603, + "num_tokens": 733195819.0, + "step": 18890 + }, + { + "epoch": 2.4031293728533267, + "grad_norm": 3.098283529281616, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9000577926635742, + "num_tokens": 733233631.0, + "step": 18891 + }, + { + "epoch": 2.403256583131917, + "grad_norm": 2.6971817016601562, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9025334119796753, + "num_tokens": 733274836.0, + "step": 18892 + }, + { + "epoch": 2.403383793410508, + "grad_norm": 3.031820774078369, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9053815603256226, + "num_tokens": 733309751.0, + "step": 18893 + }, + { + "epoch": 2.403511003689098, + "grad_norm": 3.1191020011901855, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.8986676335334778, + "num_tokens": 733346470.0, + "step": 18894 + }, + { + "epoch": 2.4036382139676884, + "grad_norm": 2.991337537765503, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9029591083526611, + "num_tokens": 733380440.0, + "step": 18895 + }, + { + "epoch": 2.403765424246279, + "grad_norm": 2.855884075164795, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.903894305229187, + "num_tokens": 733419255.0, + "step": 18896 + }, + { + "epoch": 2.4038926345248695, + "grad_norm": 2.768885850906372, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9090496301651001, + "num_tokens": 733457351.0, + "step": 18897 + }, + { + "epoch": 2.40401984480346, + "grad_norm": 3.23431134223938, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.8969229459762573, + "num_tokens": 733492836.0, + "step": 18898 + }, + { + "epoch": 2.4041470550820505, + "grad_norm": 2.900047540664673, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9105063676834106, + "num_tokens": 733530049.0, + "step": 18899 + }, + { + "epoch": 2.404274265360641, + "grad_norm": 2.8711159229278564, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.905400276184082, + "num_tokens": 733570635.0, + "step": 18900 + }, + { + "epoch": 2.4044014756392316, + "grad_norm": 2.941063165664673, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8942945599555969, + "num_tokens": 733609677.0, + "step": 18901 + }, + { + "epoch": 2.404528685917822, + "grad_norm": 3.0755813121795654, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8980588316917419, + "num_tokens": 733645179.0, + "step": 18902 + }, + { + "epoch": 2.4046558961964126, + "grad_norm": 2.657346725463867, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9014118909835815, + "num_tokens": 733690870.0, + "step": 18903 + }, + { + "epoch": 2.404783106475003, + "grad_norm": 2.810335636138916, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9105005860328674, + "num_tokens": 733727339.0, + "step": 18904 + }, + { + "epoch": 2.4049103167535937, + "grad_norm": 3.081495761871338, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8943704962730408, + "num_tokens": 733762836.0, + "step": 18905 + }, + { + "epoch": 2.405037527032184, + "grad_norm": 2.838334321975708, + "learning_rate": 1e-06, + "loss": 0.237, + "mean_token_accuracy": 0.913522481918335, + "num_tokens": 733801559.0, + "step": 18906 + }, + { + "epoch": 2.4051647373107747, + "grad_norm": 2.698033571243286, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.9074091911315918, + "num_tokens": 733843234.0, + "step": 18907 + }, + { + "epoch": 2.4052919475893653, + "grad_norm": 2.7053253650665283, + "learning_rate": 1e-06, + "loss": 0.2342, + "mean_token_accuracy": 0.9122740030288696, + "num_tokens": 733884959.0, + "step": 18908 + }, + { + "epoch": 2.405419157867956, + "grad_norm": 2.9668731689453125, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.8979316353797913, + "num_tokens": 733924542.0, + "step": 18909 + }, + { + "epoch": 2.4055463681465463, + "grad_norm": 2.93533992767334, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9095578193664551, + "num_tokens": 733961637.0, + "step": 18910 + }, + { + "epoch": 2.405673578425137, + "grad_norm": 2.941199779510498, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.892541229724884, + "num_tokens": 734001821.0, + "step": 18911 + }, + { + "epoch": 2.4058007887037274, + "grad_norm": 2.7409262657165527, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9098289608955383, + "num_tokens": 734041896.0, + "step": 18912 + }, + { + "epoch": 2.405927998982318, + "grad_norm": 2.8177597522735596, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.8997083306312561, + "num_tokens": 734084309.0, + "step": 18913 + }, + { + "epoch": 2.4060552092609084, + "grad_norm": 2.855757713317871, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9057413339614868, + "num_tokens": 734125701.0, + "step": 18914 + }, + { + "epoch": 2.406182419539499, + "grad_norm": 2.793704032897949, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9045535326004028, + "num_tokens": 734165287.0, + "step": 18915 + }, + { + "epoch": 2.4063096298180895, + "grad_norm": 3.4271020889282227, + "learning_rate": 1e-06, + "loss": 0.3169, + "mean_token_accuracy": 0.8894884586334229, + "num_tokens": 734198434.0, + "step": 18916 + }, + { + "epoch": 2.4064368400966796, + "grad_norm": 3.024146795272827, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.8990048766136169, + "num_tokens": 734235145.0, + "step": 18917 + }, + { + "epoch": 2.4065640503752705, + "grad_norm": 3.0314888954162598, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9016925096511841, + "num_tokens": 734274221.0, + "step": 18918 + }, + { + "epoch": 2.4066912606538606, + "grad_norm": 3.3129234313964844, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9054623246192932, + "num_tokens": 734307218.0, + "step": 18919 + }, + { + "epoch": 2.406818470932451, + "grad_norm": 2.8421554565429688, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9085649251937866, + "num_tokens": 734346798.0, + "step": 18920 + }, + { + "epoch": 2.4069456812110417, + "grad_norm": 2.6446943283081055, + "learning_rate": 1e-06, + "loss": 0.2154, + "mean_token_accuracy": 0.9195414781570435, + "num_tokens": 734383671.0, + "step": 18921 + }, + { + "epoch": 2.407072891489632, + "grad_norm": 2.7779886722564697, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9106743335723877, + "num_tokens": 734421238.0, + "step": 18922 + }, + { + "epoch": 2.4072001017682227, + "grad_norm": 2.965407133102417, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9052071571350098, + "num_tokens": 734460161.0, + "step": 18923 + }, + { + "epoch": 2.4073273120468133, + "grad_norm": 2.9502968788146973, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9018388986587524, + "num_tokens": 734497253.0, + "step": 18924 + }, + { + "epoch": 2.407454522325404, + "grad_norm": 2.675933361053467, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9097332954406738, + "num_tokens": 734537537.0, + "step": 18925 + }, + { + "epoch": 2.4075817326039943, + "grad_norm": 2.664293050765991, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9072009325027466, + "num_tokens": 734581512.0, + "step": 18926 + }, + { + "epoch": 2.407708942882585, + "grad_norm": 3.365489959716797, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9011809229850769, + "num_tokens": 734618758.0, + "step": 18927 + }, + { + "epoch": 2.4078361531611754, + "grad_norm": 2.73264479637146, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9075303673744202, + "num_tokens": 734664197.0, + "step": 18928 + }, + { + "epoch": 2.407963363439766, + "grad_norm": 2.745706796646118, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9073551893234253, + "num_tokens": 734707504.0, + "step": 18929 + }, + { + "epoch": 2.4080905737183564, + "grad_norm": 2.9525108337402344, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9060505032539368, + "num_tokens": 734744223.0, + "step": 18930 + }, + { + "epoch": 2.408217783996947, + "grad_norm": 2.6862921714782715, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9007738828659058, + "num_tokens": 734787871.0, + "step": 18931 + }, + { + "epoch": 2.4083449942755375, + "grad_norm": 2.992995023727417, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9004544019699097, + "num_tokens": 734827690.0, + "step": 18932 + }, + { + "epoch": 2.408472204554128, + "grad_norm": 2.8272738456726074, + "learning_rate": 1e-06, + "loss": 0.2347, + "mean_token_accuracy": 0.9100704193115234, + "num_tokens": 734869349.0, + "step": 18933 + }, + { + "epoch": 2.4085994148327186, + "grad_norm": 3.4092586040496826, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8866304159164429, + "num_tokens": 734901611.0, + "step": 18934 + }, + { + "epoch": 2.408726625111309, + "grad_norm": 2.6599464416503906, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9115539789199829, + "num_tokens": 734940930.0, + "step": 18935 + }, + { + "epoch": 2.4088538353898996, + "grad_norm": 2.64391827583313, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9059337973594666, + "num_tokens": 734984605.0, + "step": 18936 + }, + { + "epoch": 2.40898104566849, + "grad_norm": 2.7725512981414795, + "learning_rate": 1e-06, + "loss": 0.2264, + "mean_token_accuracy": 0.9163012504577637, + "num_tokens": 735025481.0, + "step": 18937 + }, + { + "epoch": 2.4091082559470807, + "grad_norm": 2.8600914478302, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8905693292617798, + "num_tokens": 735068140.0, + "step": 18938 + }, + { + "epoch": 2.409235466225671, + "grad_norm": 3.036425828933716, + "learning_rate": 1e-06, + "loss": 0.3284, + "mean_token_accuracy": 0.8845953345298767, + "num_tokens": 735112687.0, + "step": 18939 + }, + { + "epoch": 2.4093626765042617, + "grad_norm": 2.8430285453796387, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9046120047569275, + "num_tokens": 735151769.0, + "step": 18940 + }, + { + "epoch": 2.4094898867828523, + "grad_norm": 2.614447832107544, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9072717428207397, + "num_tokens": 735191952.0, + "step": 18941 + }, + { + "epoch": 2.4096170970614423, + "grad_norm": 2.858126401901245, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9083307385444641, + "num_tokens": 735229836.0, + "step": 18942 + }, + { + "epoch": 2.4097443073400333, + "grad_norm": 3.046515703201294, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9008990526199341, + "num_tokens": 735267121.0, + "step": 18943 + }, + { + "epoch": 2.4098715176186234, + "grad_norm": 2.7247300148010254, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9111701846122742, + "num_tokens": 735304193.0, + "step": 18944 + }, + { + "epoch": 2.409998727897214, + "grad_norm": 2.747833251953125, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9030424356460571, + "num_tokens": 735345803.0, + "step": 18945 + }, + { + "epoch": 2.4101259381758044, + "grad_norm": 2.786953926086426, + "learning_rate": 1e-06, + "loss": 0.2165, + "mean_token_accuracy": 0.9168683290481567, + "num_tokens": 735380146.0, + "step": 18946 + }, + { + "epoch": 2.410253148454395, + "grad_norm": 2.8062121868133545, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.9097322225570679, + "num_tokens": 735418656.0, + "step": 18947 + }, + { + "epoch": 2.4103803587329855, + "grad_norm": 2.994868755340576, + "learning_rate": 1e-06, + "loss": 0.2416, + "mean_token_accuracy": 0.9077341556549072, + "num_tokens": 735453786.0, + "step": 18948 + }, + { + "epoch": 2.410507569011576, + "grad_norm": 3.001016855239868, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9102576375007629, + "num_tokens": 735488457.0, + "step": 18949 + }, + { + "epoch": 2.4106347792901666, + "grad_norm": 2.5656800270080566, + "learning_rate": 1e-06, + "loss": 0.237, + "mean_token_accuracy": 0.9120359420776367, + "num_tokens": 735533108.0, + "step": 18950 + }, + { + "epoch": 2.410761989568757, + "grad_norm": 2.985321044921875, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9035357236862183, + "num_tokens": 735574346.0, + "step": 18951 + }, + { + "epoch": 2.4108891998473476, + "grad_norm": 2.897475004196167, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9066015481948853, + "num_tokens": 735611799.0, + "step": 18952 + }, + { + "epoch": 2.411016410125938, + "grad_norm": 2.9418346881866455, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.901793360710144, + "num_tokens": 735646465.0, + "step": 18953 + }, + { + "epoch": 2.4111436204045287, + "grad_norm": 3.0110549926757812, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.911621630191803, + "num_tokens": 735682668.0, + "step": 18954 + }, + { + "epoch": 2.411270830683119, + "grad_norm": 2.6843104362487793, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9100993871688843, + "num_tokens": 735720874.0, + "step": 18955 + }, + { + "epoch": 2.4113980409617097, + "grad_norm": 3.2388596534729004, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8967360258102417, + "num_tokens": 735755265.0, + "step": 18956 + }, + { + "epoch": 2.4115252512403003, + "grad_norm": 2.8654558658599854, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9039300680160522, + "num_tokens": 735795836.0, + "step": 18957 + }, + { + "epoch": 2.411652461518891, + "grad_norm": 3.0013153553009033, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9051589369773865, + "num_tokens": 735829333.0, + "step": 18958 + }, + { + "epoch": 2.4117796717974813, + "grad_norm": 3.3992369174957275, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8958336114883423, + "num_tokens": 735862384.0, + "step": 18959 + }, + { + "epoch": 2.411906882076072, + "grad_norm": 2.8154385089874268, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9065195918083191, + "num_tokens": 735900984.0, + "step": 18960 + }, + { + "epoch": 2.4120340923546624, + "grad_norm": 2.89593768119812, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8995879888534546, + "num_tokens": 735939857.0, + "step": 18961 + }, + { + "epoch": 2.412161302633253, + "grad_norm": 2.9052741527557373, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9029746651649475, + "num_tokens": 735980423.0, + "step": 18962 + }, + { + "epoch": 2.4122885129118434, + "grad_norm": 3.3442935943603516, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8977579474449158, + "num_tokens": 736014893.0, + "step": 18963 + }, + { + "epoch": 2.412415723190434, + "grad_norm": 2.962509870529175, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9116855263710022, + "num_tokens": 736052160.0, + "step": 18964 + }, + { + "epoch": 2.4125429334690245, + "grad_norm": 2.9403092861175537, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8991483449935913, + "num_tokens": 736092827.0, + "step": 18965 + }, + { + "epoch": 2.412670143747615, + "grad_norm": 3.1408536434173584, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9123167991638184, + "num_tokens": 736122956.0, + "step": 18966 + }, + { + "epoch": 2.412797354026205, + "grad_norm": 2.9970381259918213, + "learning_rate": 1e-06, + "loss": 0.2211, + "mean_token_accuracy": 0.9156918525695801, + "num_tokens": 736155266.0, + "step": 18967 + }, + { + "epoch": 2.412924564304796, + "grad_norm": 3.0115163326263428, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.893207848072052, + "num_tokens": 736197113.0, + "step": 18968 + }, + { + "epoch": 2.413051774583386, + "grad_norm": 2.8682456016540527, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9032949209213257, + "num_tokens": 736236815.0, + "step": 18969 + }, + { + "epoch": 2.4131789848619767, + "grad_norm": 2.751547336578369, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9067794680595398, + "num_tokens": 736279785.0, + "step": 18970 + }, + { + "epoch": 2.413306195140567, + "grad_norm": 2.7378995418548584, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9028096199035645, + "num_tokens": 736319325.0, + "step": 18971 + }, + { + "epoch": 2.4134334054191577, + "grad_norm": 3.0086381435394287, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.9103010892868042, + "num_tokens": 736353862.0, + "step": 18972 + }, + { + "epoch": 2.4135606156977483, + "grad_norm": 3.1055691242218018, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.905624508857727, + "num_tokens": 736389896.0, + "step": 18973 + }, + { + "epoch": 2.413687825976339, + "grad_norm": 3.0004210472106934, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8960636854171753, + "num_tokens": 736428310.0, + "step": 18974 + }, + { + "epoch": 2.4138150362549293, + "grad_norm": 2.7596871852874756, + "learning_rate": 1e-06, + "loss": 0.2295, + "mean_token_accuracy": 0.9132121801376343, + "num_tokens": 736465538.0, + "step": 18975 + }, + { + "epoch": 2.41394224653352, + "grad_norm": 2.792877674102783, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9088441729545593, + "num_tokens": 736504885.0, + "step": 18976 + }, + { + "epoch": 2.4140694568121104, + "grad_norm": 2.6833722591400146, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.9007496237754822, + "num_tokens": 736549930.0, + "step": 18977 + }, + { + "epoch": 2.414196667090701, + "grad_norm": 2.8373897075653076, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9113469123840332, + "num_tokens": 736587247.0, + "step": 18978 + }, + { + "epoch": 2.4143238773692914, + "grad_norm": 3.1164114475250244, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8964293003082275, + "num_tokens": 736624793.0, + "step": 18979 + }, + { + "epoch": 2.414451087647882, + "grad_norm": 3.03790020942688, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9033670425415039, + "num_tokens": 736660186.0, + "step": 18980 + }, + { + "epoch": 2.4145782979264725, + "grad_norm": 2.7639222145080566, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.905478298664093, + "num_tokens": 736704714.0, + "step": 18981 + }, + { + "epoch": 2.414705508205063, + "grad_norm": 2.76755428314209, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9103056192398071, + "num_tokens": 736742272.0, + "step": 18982 + }, + { + "epoch": 2.4148327184836536, + "grad_norm": 2.6281394958496094, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9098369479179382, + "num_tokens": 736781177.0, + "step": 18983 + }, + { + "epoch": 2.414959928762244, + "grad_norm": 3.0540523529052734, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.9113405346870422, + "num_tokens": 736810958.0, + "step": 18984 + }, + { + "epoch": 2.4150871390408346, + "grad_norm": 2.7164700031280518, + "learning_rate": 1e-06, + "loss": 0.2195, + "mean_token_accuracy": 0.9191826581954956, + "num_tokens": 736846460.0, + "step": 18985 + }, + { + "epoch": 2.415214349319425, + "grad_norm": 3.2310497760772705, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8993408679962158, + "num_tokens": 736881351.0, + "step": 18986 + }, + { + "epoch": 2.4153415595980157, + "grad_norm": 3.0083062648773193, + "learning_rate": 1e-06, + "loss": 0.2978, + "mean_token_accuracy": 0.8918924331665039, + "num_tokens": 736927937.0, + "step": 18987 + }, + { + "epoch": 2.415468769876606, + "grad_norm": 3.056680679321289, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9053862690925598, + "num_tokens": 736963023.0, + "step": 18988 + }, + { + "epoch": 2.4155959801551967, + "grad_norm": 2.8342320919036865, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9094264507293701, + "num_tokens": 737001066.0, + "step": 18989 + }, + { + "epoch": 2.415723190433787, + "grad_norm": 2.9072818756103516, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9016810059547424, + "num_tokens": 737037430.0, + "step": 18990 + }, + { + "epoch": 2.4158504007123778, + "grad_norm": 2.6447064876556396, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8921051621437073, + "num_tokens": 737083981.0, + "step": 18991 + }, + { + "epoch": 2.415977610990968, + "grad_norm": 2.9828526973724365, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9030812978744507, + "num_tokens": 737121901.0, + "step": 18992 + }, + { + "epoch": 2.4161048212695584, + "grad_norm": 2.680117130279541, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9107598066329956, + "num_tokens": 737163821.0, + "step": 18993 + }, + { + "epoch": 2.416232031548149, + "grad_norm": 2.803158760070801, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9095479249954224, + "num_tokens": 737200466.0, + "step": 18994 + }, + { + "epoch": 2.4163592418267394, + "grad_norm": 2.8693933486938477, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9084458351135254, + "num_tokens": 737241058.0, + "step": 18995 + }, + { + "epoch": 2.41648645210533, + "grad_norm": 2.725193977355957, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9010571241378784, + "num_tokens": 737284332.0, + "step": 18996 + }, + { + "epoch": 2.4166136623839205, + "grad_norm": 3.129446268081665, + "learning_rate": 1e-06, + "loss": 0.2425, + "mean_token_accuracy": 0.9118277430534363, + "num_tokens": 737318531.0, + "step": 18997 + }, + { + "epoch": 2.416740872662511, + "grad_norm": 2.654967784881592, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9054524302482605, + "num_tokens": 737362210.0, + "step": 18998 + }, + { + "epoch": 2.4168680829411016, + "grad_norm": 3.0082449913024902, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9003590941429138, + "num_tokens": 737401215.0, + "step": 18999 + }, + { + "epoch": 2.416995293219692, + "grad_norm": 2.6455135345458984, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.911329448223114, + "num_tokens": 737438994.0, + "step": 19000 + }, + { + "epoch": 2.4171225034982826, + "grad_norm": 2.972745656967163, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9087173938751221, + "num_tokens": 737472635.0, + "step": 19001 + }, + { + "epoch": 2.417249713776873, + "grad_norm": 3.1589667797088623, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8946250677108765, + "num_tokens": 737507684.0, + "step": 19002 + }, + { + "epoch": 2.4173769240554637, + "grad_norm": 2.9275949001312256, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.8989224433898926, + "num_tokens": 737544684.0, + "step": 19003 + }, + { + "epoch": 2.417504134334054, + "grad_norm": 2.929619073867798, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9075732827186584, + "num_tokens": 737580016.0, + "step": 19004 + }, + { + "epoch": 2.4176313446126447, + "grad_norm": 2.78287672996521, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8987114429473877, + "num_tokens": 737625975.0, + "step": 19005 + }, + { + "epoch": 2.4177585548912353, + "grad_norm": 3.0448391437530518, + "learning_rate": 1e-06, + "loss": 0.2271, + "mean_token_accuracy": 0.914945662021637, + "num_tokens": 737658271.0, + "step": 19006 + }, + { + "epoch": 2.417885765169826, + "grad_norm": 2.806009531021118, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9110082983970642, + "num_tokens": 737698815.0, + "step": 19007 + }, + { + "epoch": 2.4180129754484163, + "grad_norm": 2.9128212928771973, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9092859625816345, + "num_tokens": 737735968.0, + "step": 19008 + }, + { + "epoch": 2.418140185727007, + "grad_norm": 2.8721678256988525, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9059368968009949, + "num_tokens": 737776987.0, + "step": 19009 + }, + { + "epoch": 2.4182673960055974, + "grad_norm": 2.748082399368286, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.914231538772583, + "num_tokens": 737812929.0, + "step": 19010 + }, + { + "epoch": 2.418394606284188, + "grad_norm": 2.7420527935028076, + "learning_rate": 1e-06, + "loss": 0.2267, + "mean_token_accuracy": 0.9152069091796875, + "num_tokens": 737850304.0, + "step": 19011 + }, + { + "epoch": 2.4185218165627784, + "grad_norm": 2.9748873710632324, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9013659358024597, + "num_tokens": 737892653.0, + "step": 19012 + }, + { + "epoch": 2.418649026841369, + "grad_norm": 2.8849804401397705, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9035739898681641, + "num_tokens": 737933264.0, + "step": 19013 + }, + { + "epoch": 2.4187762371199595, + "grad_norm": 2.7034189701080322, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9052311182022095, + "num_tokens": 737977559.0, + "step": 19014 + }, + { + "epoch": 2.4189034473985496, + "grad_norm": 2.805467128753662, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9169188737869263, + "num_tokens": 738013017.0, + "step": 19015 + }, + { + "epoch": 2.4190306576771405, + "grad_norm": 2.804476737976074, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9044179916381836, + "num_tokens": 738054867.0, + "step": 19016 + }, + { + "epoch": 2.4191578679557306, + "grad_norm": 2.894967794418335, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.911016583442688, + "num_tokens": 738089925.0, + "step": 19017 + }, + { + "epoch": 2.419285078234321, + "grad_norm": 2.8079497814178467, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.8994503617286682, + "num_tokens": 738135122.0, + "step": 19018 + }, + { + "epoch": 2.4194122885129117, + "grad_norm": 2.9309277534484863, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9099840521812439, + "num_tokens": 738172800.0, + "step": 19019 + }, + { + "epoch": 2.419539498791502, + "grad_norm": 2.8160736560821533, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9085993766784668, + "num_tokens": 738214058.0, + "step": 19020 + }, + { + "epoch": 2.4196667090700927, + "grad_norm": 2.745333194732666, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9073545932769775, + "num_tokens": 738253646.0, + "step": 19021 + }, + { + "epoch": 2.4197939193486833, + "grad_norm": 3.021500825881958, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9001309275627136, + "num_tokens": 738290466.0, + "step": 19022 + }, + { + "epoch": 2.419921129627274, + "grad_norm": 2.851670742034912, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9016693830490112, + "num_tokens": 738332156.0, + "step": 19023 + }, + { + "epoch": 2.4200483399058643, + "grad_norm": 3.0251224040985107, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.8992263078689575, + "num_tokens": 738372152.0, + "step": 19024 + }, + { + "epoch": 2.420175550184455, + "grad_norm": 3.154146671295166, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9008930325508118, + "num_tokens": 738409566.0, + "step": 19025 + }, + { + "epoch": 2.4203027604630454, + "grad_norm": 2.8981525897979736, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.898385226726532, + "num_tokens": 738450076.0, + "step": 19026 + }, + { + "epoch": 2.420429970741636, + "grad_norm": 3.003066062927246, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9060512185096741, + "num_tokens": 738485205.0, + "step": 19027 + }, + { + "epoch": 2.4205571810202264, + "grad_norm": 2.853942394256592, + "learning_rate": 1e-06, + "loss": 0.2392, + "mean_token_accuracy": 0.9095715284347534, + "num_tokens": 738522039.0, + "step": 19028 + }, + { + "epoch": 2.420684391298817, + "grad_norm": 2.9125330448150635, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.90788733959198, + "num_tokens": 738560024.0, + "step": 19029 + }, + { + "epoch": 2.4208116015774075, + "grad_norm": 2.9962124824523926, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9040356874465942, + "num_tokens": 738595808.0, + "step": 19030 + }, + { + "epoch": 2.420938811855998, + "grad_norm": 2.9808149337768555, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8960058689117432, + "num_tokens": 738632151.0, + "step": 19031 + }, + { + "epoch": 2.4210660221345885, + "grad_norm": 2.592986822128296, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9119946360588074, + "num_tokens": 738671164.0, + "step": 19032 + }, + { + "epoch": 2.421193232413179, + "grad_norm": 2.6211371421813965, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9117292165756226, + "num_tokens": 738710989.0, + "step": 19033 + }, + { + "epoch": 2.4213204426917696, + "grad_norm": 3.0986509323120117, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9046263694763184, + "num_tokens": 738743841.0, + "step": 19034 + }, + { + "epoch": 2.42144765297036, + "grad_norm": 2.8642358779907227, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9052460193634033, + "num_tokens": 738785861.0, + "step": 19035 + }, + { + "epoch": 2.4215748632489507, + "grad_norm": 2.798656702041626, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9052668809890747, + "num_tokens": 738827771.0, + "step": 19036 + }, + { + "epoch": 2.421702073527541, + "grad_norm": 3.106228828430176, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8926783800125122, + "num_tokens": 738868809.0, + "step": 19037 + }, + { + "epoch": 2.4218292838061317, + "grad_norm": 3.0295419692993164, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9051341414451599, + "num_tokens": 738904400.0, + "step": 19038 + }, + { + "epoch": 2.4219564940847222, + "grad_norm": 2.855405330657959, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9096286296844482, + "num_tokens": 738940110.0, + "step": 19039 + }, + { + "epoch": 2.4220837043633123, + "grad_norm": 3.111619234085083, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8965301513671875, + "num_tokens": 738979684.0, + "step": 19040 + }, + { + "epoch": 2.4222109146419033, + "grad_norm": 3.109145164489746, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.9005579948425293, + "num_tokens": 739016997.0, + "step": 19041 + }, + { + "epoch": 2.4223381249204934, + "grad_norm": 2.861788749694824, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9119431376457214, + "num_tokens": 739053051.0, + "step": 19042 + }, + { + "epoch": 2.422465335199084, + "grad_norm": 2.9162726402282715, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9119086265563965, + "num_tokens": 739089043.0, + "step": 19043 + }, + { + "epoch": 2.4225925454776744, + "grad_norm": 3.0495593547821045, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.898618757724762, + "num_tokens": 739125439.0, + "step": 19044 + }, + { + "epoch": 2.422719755756265, + "grad_norm": 2.8879618644714355, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9099209308624268, + "num_tokens": 739166313.0, + "step": 19045 + }, + { + "epoch": 2.4228469660348555, + "grad_norm": 3.04012131690979, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9052015542984009, + "num_tokens": 739202645.0, + "step": 19046 + }, + { + "epoch": 2.422974176313446, + "grad_norm": 2.9998295307159424, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8941688537597656, + "num_tokens": 739242777.0, + "step": 19047 + }, + { + "epoch": 2.4231013865920366, + "grad_norm": 2.912078380584717, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9026868343353271, + "num_tokens": 739283202.0, + "step": 19048 + }, + { + "epoch": 2.423228596870627, + "grad_norm": 2.7883291244506836, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9057930707931519, + "num_tokens": 739323343.0, + "step": 19049 + }, + { + "epoch": 2.4233558071492176, + "grad_norm": 3.0512053966522217, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8954620361328125, + "num_tokens": 739361744.0, + "step": 19050 + }, + { + "epoch": 2.423483017427808, + "grad_norm": 2.907255172729492, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.907219648361206, + "num_tokens": 739402535.0, + "step": 19051 + }, + { + "epoch": 2.4236102277063987, + "grad_norm": 2.883697986602783, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9008901119232178, + "num_tokens": 739439005.0, + "step": 19052 + }, + { + "epoch": 2.423737437984989, + "grad_norm": 2.8284130096435547, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8988509774208069, + "num_tokens": 739477510.0, + "step": 19053 + }, + { + "epoch": 2.4238646482635797, + "grad_norm": 3.1691195964813232, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.8997694253921509, + "num_tokens": 739510840.0, + "step": 19054 + }, + { + "epoch": 2.4239918585421703, + "grad_norm": 2.908067464828491, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9082108736038208, + "num_tokens": 739545075.0, + "step": 19055 + }, + { + "epoch": 2.424119068820761, + "grad_norm": 3.023498058319092, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9048277139663696, + "num_tokens": 739582110.0, + "step": 19056 + }, + { + "epoch": 2.4242462790993513, + "grad_norm": 3.1537740230560303, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9007686376571655, + "num_tokens": 739614757.0, + "step": 19057 + }, + { + "epoch": 2.424373489377942, + "grad_norm": 2.634829521179199, + "learning_rate": 1e-06, + "loss": 0.2219, + "mean_token_accuracy": 0.916588544845581, + "num_tokens": 739656253.0, + "step": 19058 + }, + { + "epoch": 2.4245006996565324, + "grad_norm": 2.686953067779541, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9087070226669312, + "num_tokens": 739700739.0, + "step": 19059 + }, + { + "epoch": 2.424627909935123, + "grad_norm": 2.962308406829834, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9010255336761475, + "num_tokens": 739741275.0, + "step": 19060 + }, + { + "epoch": 2.4247551202137134, + "grad_norm": 2.696774482727051, + "learning_rate": 1e-06, + "loss": 0.2097, + "mean_token_accuracy": 0.9205063581466675, + "num_tokens": 739779452.0, + "step": 19061 + }, + { + "epoch": 2.424882330492304, + "grad_norm": 2.9163901805877686, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9033286571502686, + "num_tokens": 739817404.0, + "step": 19062 + }, + { + "epoch": 2.4250095407708945, + "grad_norm": 2.8432655334472656, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9067120552062988, + "num_tokens": 739858482.0, + "step": 19063 + }, + { + "epoch": 2.425136751049485, + "grad_norm": 3.1603574752807617, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9073136448860168, + "num_tokens": 739891396.0, + "step": 19064 + }, + { + "epoch": 2.425263961328075, + "grad_norm": 2.8664205074310303, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9070755243301392, + "num_tokens": 739933420.0, + "step": 19065 + }, + { + "epoch": 2.425391171606666, + "grad_norm": 2.8228869438171387, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9073727130889893, + "num_tokens": 739972249.0, + "step": 19066 + }, + { + "epoch": 2.425518381885256, + "grad_norm": 2.927734375, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9078245759010315, + "num_tokens": 740009164.0, + "step": 19067 + }, + { + "epoch": 2.4256455921638467, + "grad_norm": 3.0942163467407227, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.906892716884613, + "num_tokens": 740042336.0, + "step": 19068 + }, + { + "epoch": 2.425772802442437, + "grad_norm": 2.9332733154296875, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9057577848434448, + "num_tokens": 740079099.0, + "step": 19069 + }, + { + "epoch": 2.4259000127210277, + "grad_norm": 2.952057123184204, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.905474066734314, + "num_tokens": 740117053.0, + "step": 19070 + }, + { + "epoch": 2.4260272229996183, + "grad_norm": 3.050147533416748, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8948386907577515, + "num_tokens": 740154505.0, + "step": 19071 + }, + { + "epoch": 2.426154433278209, + "grad_norm": 2.605867862701416, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9139650464057922, + "num_tokens": 740195258.0, + "step": 19072 + }, + { + "epoch": 2.4262816435567993, + "grad_norm": 2.776124954223633, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9034417867660522, + "num_tokens": 740234073.0, + "step": 19073 + }, + { + "epoch": 2.42640885383539, + "grad_norm": 3.123441219329834, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8978705406188965, + "num_tokens": 740271727.0, + "step": 19074 + }, + { + "epoch": 2.4265360641139804, + "grad_norm": 2.7553296089172363, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9002848863601685, + "num_tokens": 740313387.0, + "step": 19075 + }, + { + "epoch": 2.426663274392571, + "grad_norm": 2.82468843460083, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.913627564907074, + "num_tokens": 740352483.0, + "step": 19076 + }, + { + "epoch": 2.4267904846711614, + "grad_norm": 2.9705448150634766, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9047703742980957, + "num_tokens": 740396809.0, + "step": 19077 + }, + { + "epoch": 2.426917694949752, + "grad_norm": 2.9730632305145264, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9052426815032959, + "num_tokens": 740434616.0, + "step": 19078 + }, + { + "epoch": 2.4270449052283425, + "grad_norm": 2.978771209716797, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.8982212543487549, + "num_tokens": 740470948.0, + "step": 19079 + }, + { + "epoch": 2.427172115506933, + "grad_norm": 3.0687313079833984, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8945049047470093, + "num_tokens": 740507038.0, + "step": 19080 + }, + { + "epoch": 2.4272993257855235, + "grad_norm": 2.745979070663452, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9036756753921509, + "num_tokens": 740547114.0, + "step": 19081 + }, + { + "epoch": 2.427426536064114, + "grad_norm": 3.0354132652282715, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.8974714279174805, + "num_tokens": 740585117.0, + "step": 19082 + }, + { + "epoch": 2.4275537463427046, + "grad_norm": 3.031768798828125, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9096969366073608, + "num_tokens": 740619233.0, + "step": 19083 + }, + { + "epoch": 2.427680956621295, + "grad_norm": 3.1223599910736084, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9029937386512756, + "num_tokens": 740653597.0, + "step": 19084 + }, + { + "epoch": 2.4278081668998857, + "grad_norm": 2.758089780807495, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.908316969871521, + "num_tokens": 740688445.0, + "step": 19085 + }, + { + "epoch": 2.427935377178476, + "grad_norm": 2.9617388248443604, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9053005576133728, + "num_tokens": 740725895.0, + "step": 19086 + }, + { + "epoch": 2.4280625874570667, + "grad_norm": 3.0248215198516846, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9091755151748657, + "num_tokens": 740761745.0, + "step": 19087 + }, + { + "epoch": 2.428189797735657, + "grad_norm": 3.0311648845672607, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9066460132598877, + "num_tokens": 740799682.0, + "step": 19088 + }, + { + "epoch": 2.4283170080142478, + "grad_norm": 2.8768157958984375, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8953267931938171, + "num_tokens": 740841694.0, + "step": 19089 + }, + { + "epoch": 2.428444218292838, + "grad_norm": 2.6939401626586914, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9082074165344238, + "num_tokens": 740883027.0, + "step": 19090 + }, + { + "epoch": 2.4285714285714284, + "grad_norm": 2.737053871154785, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8948782682418823, + "num_tokens": 740926215.0, + "step": 19091 + }, + { + "epoch": 2.428698638850019, + "grad_norm": 3.1082634925842285, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9090379476547241, + "num_tokens": 740959751.0, + "step": 19092 + }, + { + "epoch": 2.4288258491286094, + "grad_norm": 2.860812187194824, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9055604934692383, + "num_tokens": 741002507.0, + "step": 19093 + }, + { + "epoch": 2.4289530594072, + "grad_norm": 2.9913394451141357, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9085537195205688, + "num_tokens": 741037488.0, + "step": 19094 + }, + { + "epoch": 2.4290802696857905, + "grad_norm": 2.995565891265869, + "learning_rate": 1e-06, + "loss": 0.2349, + "mean_token_accuracy": 0.9145364761352539, + "num_tokens": 741071655.0, + "step": 19095 + }, + { + "epoch": 2.429207479964381, + "grad_norm": 2.945629119873047, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9094946384429932, + "num_tokens": 741112615.0, + "step": 19096 + }, + { + "epoch": 2.4293346902429716, + "grad_norm": 2.910569667816162, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9126611948013306, + "num_tokens": 741151968.0, + "step": 19097 + }, + { + "epoch": 2.429461900521562, + "grad_norm": 2.9363129138946533, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9025843739509583, + "num_tokens": 741186857.0, + "step": 19098 + }, + { + "epoch": 2.4295891108001526, + "grad_norm": 2.6018550395965576, + "learning_rate": 1e-06, + "loss": 0.211, + "mean_token_accuracy": 0.9209321737289429, + "num_tokens": 741225150.0, + "step": 19099 + }, + { + "epoch": 2.429716321078743, + "grad_norm": 3.123619556427002, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.9072821140289307, + "num_tokens": 741257159.0, + "step": 19100 + }, + { + "epoch": 2.4298435313573337, + "grad_norm": 2.913323163986206, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9080126285552979, + "num_tokens": 741295496.0, + "step": 19101 + }, + { + "epoch": 2.429970741635924, + "grad_norm": 2.7727675437927246, + "learning_rate": 1e-06, + "loss": 0.2413, + "mean_token_accuracy": 0.9085631370544434, + "num_tokens": 741333358.0, + "step": 19102 + }, + { + "epoch": 2.4300979519145147, + "grad_norm": 2.7179746627807617, + "learning_rate": 1e-06, + "loss": 0.2217, + "mean_token_accuracy": 0.9186301231384277, + "num_tokens": 741372292.0, + "step": 19103 + }, + { + "epoch": 2.4302251621931052, + "grad_norm": 3.199032783508301, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.8985083699226379, + "num_tokens": 741408230.0, + "step": 19104 + }, + { + "epoch": 2.4303523724716958, + "grad_norm": 2.8755600452423096, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.8979113101959229, + "num_tokens": 741448242.0, + "step": 19105 + }, + { + "epoch": 2.4304795827502863, + "grad_norm": 2.9730899333953857, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9028884172439575, + "num_tokens": 741486777.0, + "step": 19106 + }, + { + "epoch": 2.430606793028877, + "grad_norm": 2.995370626449585, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8977407813072205, + "num_tokens": 741522092.0, + "step": 19107 + }, + { + "epoch": 2.4307340033074674, + "grad_norm": 2.7709336280822754, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9011805653572083, + "num_tokens": 741564762.0, + "step": 19108 + }, + { + "epoch": 2.430861213586058, + "grad_norm": 3.1067938804626465, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8965432643890381, + "num_tokens": 741604876.0, + "step": 19109 + }, + { + "epoch": 2.4309884238646484, + "grad_norm": 2.568772315979004, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9099929928779602, + "num_tokens": 741647508.0, + "step": 19110 + }, + { + "epoch": 2.431115634143239, + "grad_norm": 2.816453456878662, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9036084413528442, + "num_tokens": 741692685.0, + "step": 19111 + }, + { + "epoch": 2.4312428444218295, + "grad_norm": 2.7691543102264404, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9029263854026794, + "num_tokens": 741735803.0, + "step": 19112 + }, + { + "epoch": 2.4313700547004196, + "grad_norm": 2.880439519882202, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.8992758989334106, + "num_tokens": 741774162.0, + "step": 19113 + }, + { + "epoch": 2.4314972649790105, + "grad_norm": 2.909337282180786, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9036037921905518, + "num_tokens": 741809739.0, + "step": 19114 + }, + { + "epoch": 2.4316244752576006, + "grad_norm": 2.8541243076324463, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9028841257095337, + "num_tokens": 741849410.0, + "step": 19115 + }, + { + "epoch": 2.431751685536191, + "grad_norm": 3.0775246620178223, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.898807168006897, + "num_tokens": 741884680.0, + "step": 19116 + }, + { + "epoch": 2.4318788958147817, + "grad_norm": 2.7621376514434814, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9110541343688965, + "num_tokens": 741923416.0, + "step": 19117 + }, + { + "epoch": 2.432006106093372, + "grad_norm": 2.841567277908325, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.909687876701355, + "num_tokens": 741961275.0, + "step": 19118 + }, + { + "epoch": 2.4321333163719627, + "grad_norm": 2.6678919792175293, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9078839421272278, + "num_tokens": 742004258.0, + "step": 19119 + }, + { + "epoch": 2.4322605266505533, + "grad_norm": 3.17625093460083, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8937386870384216, + "num_tokens": 742040439.0, + "step": 19120 + }, + { + "epoch": 2.432387736929144, + "grad_norm": 2.6414942741394043, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.899879515171051, + "num_tokens": 742084271.0, + "step": 19121 + }, + { + "epoch": 2.4325149472077343, + "grad_norm": 2.78322434425354, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9085817337036133, + "num_tokens": 742127869.0, + "step": 19122 + }, + { + "epoch": 2.432642157486325, + "grad_norm": 2.8646275997161865, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9086354970932007, + "num_tokens": 742167820.0, + "step": 19123 + }, + { + "epoch": 2.4327693677649154, + "grad_norm": 3.295546293258667, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9042953252792358, + "num_tokens": 742198766.0, + "step": 19124 + }, + { + "epoch": 2.432896578043506, + "grad_norm": 2.7016756534576416, + "learning_rate": 1e-06, + "loss": 0.2331, + "mean_token_accuracy": 0.9130922555923462, + "num_tokens": 742238670.0, + "step": 19125 + }, + { + "epoch": 2.4330237883220964, + "grad_norm": 2.8603932857513428, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9037572145462036, + "num_tokens": 742277303.0, + "step": 19126 + }, + { + "epoch": 2.433150998600687, + "grad_norm": 2.6611523628234863, + "learning_rate": 1e-06, + "loss": 0.2345, + "mean_token_accuracy": 0.9127923250198364, + "num_tokens": 742319853.0, + "step": 19127 + }, + { + "epoch": 2.4332782088792775, + "grad_norm": 2.7263400554656982, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.903276801109314, + "num_tokens": 742360372.0, + "step": 19128 + }, + { + "epoch": 2.433405419157868, + "grad_norm": 2.9856553077697754, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9004557132720947, + "num_tokens": 742403474.0, + "step": 19129 + }, + { + "epoch": 2.4335326294364585, + "grad_norm": 2.776390552520752, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9096408486366272, + "num_tokens": 742446187.0, + "step": 19130 + }, + { + "epoch": 2.433659839715049, + "grad_norm": 2.9722306728363037, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8942537307739258, + "num_tokens": 742485431.0, + "step": 19131 + }, + { + "epoch": 2.4337870499936396, + "grad_norm": 3.271491050720215, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8963357210159302, + "num_tokens": 742526624.0, + "step": 19132 + }, + { + "epoch": 2.43391426027223, + "grad_norm": 2.882645845413208, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9047236442565918, + "num_tokens": 742567656.0, + "step": 19133 + }, + { + "epoch": 2.4340414705508207, + "grad_norm": 2.745366096496582, + "learning_rate": 1e-06, + "loss": 0.2241, + "mean_token_accuracy": 0.9141734838485718, + "num_tokens": 742605416.0, + "step": 19134 + }, + { + "epoch": 2.434168680829411, + "grad_norm": 2.807730197906494, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9076356887817383, + "num_tokens": 742642934.0, + "step": 19135 + }, + { + "epoch": 2.4342958911080017, + "grad_norm": 3.273343801498413, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.900984525680542, + "num_tokens": 742675775.0, + "step": 19136 + }, + { + "epoch": 2.4344231013865922, + "grad_norm": 3.323786497116089, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9016404151916504, + "num_tokens": 742711574.0, + "step": 19137 + }, + { + "epoch": 2.4345503116651823, + "grad_norm": 2.5551090240478516, + "learning_rate": 1e-06, + "loss": 0.2331, + "mean_token_accuracy": 0.9144026041030884, + "num_tokens": 742754504.0, + "step": 19138 + }, + { + "epoch": 2.4346775219437733, + "grad_norm": 2.9711012840270996, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.913294792175293, + "num_tokens": 742789150.0, + "step": 19139 + }, + { + "epoch": 2.4348047322223634, + "grad_norm": 2.902390956878662, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9076860547065735, + "num_tokens": 742825735.0, + "step": 19140 + }, + { + "epoch": 2.434931942500954, + "grad_norm": 3.011991262435913, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9026848077774048, + "num_tokens": 742863482.0, + "step": 19141 + }, + { + "epoch": 2.4350591527795444, + "grad_norm": 3.056867837905884, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9080313444137573, + "num_tokens": 742897117.0, + "step": 19142 + }, + { + "epoch": 2.435186363058135, + "grad_norm": 3.01125168800354, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9083388447761536, + "num_tokens": 742934192.0, + "step": 19143 + }, + { + "epoch": 2.4353135733367255, + "grad_norm": 2.534881830215454, + "learning_rate": 1e-06, + "loss": 0.1997, + "mean_token_accuracy": 0.9251170754432678, + "num_tokens": 742971393.0, + "step": 19144 + }, + { + "epoch": 2.435440783615316, + "grad_norm": 2.7937464714050293, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9090782403945923, + "num_tokens": 743010858.0, + "step": 19145 + }, + { + "epoch": 2.4355679938939065, + "grad_norm": 2.6281962394714355, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9091042280197144, + "num_tokens": 743055312.0, + "step": 19146 + }, + { + "epoch": 2.435695204172497, + "grad_norm": 2.8965044021606445, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9111087918281555, + "num_tokens": 743089633.0, + "step": 19147 + }, + { + "epoch": 2.4358224144510876, + "grad_norm": 2.8387532234191895, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9142535924911499, + "num_tokens": 743126666.0, + "step": 19148 + }, + { + "epoch": 2.435949624729678, + "grad_norm": 2.9084017276763916, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9060287475585938, + "num_tokens": 743164267.0, + "step": 19149 + }, + { + "epoch": 2.4360768350082687, + "grad_norm": 2.533351182937622, + "learning_rate": 1e-06, + "loss": 0.2221, + "mean_token_accuracy": 0.916835367679596, + "num_tokens": 743209330.0, + "step": 19150 + }, + { + "epoch": 2.436204045286859, + "grad_norm": 2.749197006225586, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9042458534240723, + "num_tokens": 743254001.0, + "step": 19151 + }, + { + "epoch": 2.4363312555654497, + "grad_norm": 2.8670456409454346, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9087479710578918, + "num_tokens": 743293899.0, + "step": 19152 + }, + { + "epoch": 2.4364584658440402, + "grad_norm": 2.70760440826416, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9141780138015747, + "num_tokens": 743332492.0, + "step": 19153 + }, + { + "epoch": 2.4365856761226308, + "grad_norm": 2.8341481685638428, + "learning_rate": 1e-06, + "loss": 0.2299, + "mean_token_accuracy": 0.915717601776123, + "num_tokens": 743368174.0, + "step": 19154 + }, + { + "epoch": 2.4367128864012213, + "grad_norm": 2.826073408126831, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9059771299362183, + "num_tokens": 743409851.0, + "step": 19155 + }, + { + "epoch": 2.436840096679812, + "grad_norm": 3.0213334560394287, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8896768093109131, + "num_tokens": 743451213.0, + "step": 19156 + }, + { + "epoch": 2.4369673069584024, + "grad_norm": 2.709958553314209, + "learning_rate": 1e-06, + "loss": 0.2253, + "mean_token_accuracy": 0.9132174253463745, + "num_tokens": 743490714.0, + "step": 19157 + }, + { + "epoch": 2.437094517236993, + "grad_norm": 2.9853620529174805, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9055356979370117, + "num_tokens": 743526428.0, + "step": 19158 + }, + { + "epoch": 2.4372217275155834, + "grad_norm": 3.440725326538086, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8950424790382385, + "num_tokens": 743558200.0, + "step": 19159 + }, + { + "epoch": 2.437348937794174, + "grad_norm": 2.9605178833007812, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9034115672111511, + "num_tokens": 743599084.0, + "step": 19160 + }, + { + "epoch": 2.4374761480727645, + "grad_norm": 2.881234645843506, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8970696330070496, + "num_tokens": 743640571.0, + "step": 19161 + }, + { + "epoch": 2.437603358351355, + "grad_norm": 3.1328792572021484, + "learning_rate": 1e-06, + "loss": 0.3003, + "mean_token_accuracy": 0.8912338614463806, + "num_tokens": 743678790.0, + "step": 19162 + }, + { + "epoch": 2.437730568629945, + "grad_norm": 2.9465699195861816, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9019505381584167, + "num_tokens": 743718798.0, + "step": 19163 + }, + { + "epoch": 2.437857778908536, + "grad_norm": 2.7288544178009033, + "learning_rate": 1e-06, + "loss": 0.2212, + "mean_token_accuracy": 0.9143095016479492, + "num_tokens": 743754808.0, + "step": 19164 + }, + { + "epoch": 2.437984989187126, + "grad_norm": 3.2649526596069336, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9006643891334534, + "num_tokens": 743787685.0, + "step": 19165 + }, + { + "epoch": 2.4381121994657167, + "grad_norm": 2.5579428672790527, + "learning_rate": 1e-06, + "loss": 0.2309, + "mean_token_accuracy": 0.9148858189582825, + "num_tokens": 743833239.0, + "step": 19166 + }, + { + "epoch": 2.438239409744307, + "grad_norm": 2.976713180541992, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.8993271589279175, + "num_tokens": 743871774.0, + "step": 19167 + }, + { + "epoch": 2.4383666200228977, + "grad_norm": 2.9321932792663574, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9022790193557739, + "num_tokens": 743909923.0, + "step": 19168 + }, + { + "epoch": 2.4384938303014883, + "grad_norm": 2.8133578300476074, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9099104404449463, + "num_tokens": 743948919.0, + "step": 19169 + }, + { + "epoch": 2.438621040580079, + "grad_norm": 2.846038341522217, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9042426943778992, + "num_tokens": 743986771.0, + "step": 19170 + }, + { + "epoch": 2.4387482508586693, + "grad_norm": 2.9036247730255127, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9002952575683594, + "num_tokens": 744027495.0, + "step": 19171 + }, + { + "epoch": 2.43887546113726, + "grad_norm": 2.905564308166504, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9112930297851562, + "num_tokens": 744065378.0, + "step": 19172 + }, + { + "epoch": 2.4390026714158504, + "grad_norm": 2.687861680984497, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9081706404685974, + "num_tokens": 744103109.0, + "step": 19173 + }, + { + "epoch": 2.439129881694441, + "grad_norm": 2.9969401359558105, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8950344920158386, + "num_tokens": 744139243.0, + "step": 19174 + }, + { + "epoch": 2.4392570919730314, + "grad_norm": 2.686103105545044, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9108180999755859, + "num_tokens": 744180338.0, + "step": 19175 + }, + { + "epoch": 2.439384302251622, + "grad_norm": 2.7080132961273193, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9069809913635254, + "num_tokens": 744222290.0, + "step": 19176 + }, + { + "epoch": 2.4395115125302125, + "grad_norm": 2.477311372756958, + "learning_rate": 1e-06, + "loss": 0.2154, + "mean_token_accuracy": 0.9197432994842529, + "num_tokens": 744261707.0, + "step": 19177 + }, + { + "epoch": 2.439638722808803, + "grad_norm": 2.7808632850646973, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9085922241210938, + "num_tokens": 744302178.0, + "step": 19178 + }, + { + "epoch": 2.4397659330873935, + "grad_norm": 2.9952621459960938, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8997803926467896, + "num_tokens": 744338999.0, + "step": 19179 + }, + { + "epoch": 2.439893143365984, + "grad_norm": 2.9310712814331055, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9099571704864502, + "num_tokens": 744378491.0, + "step": 19180 + }, + { + "epoch": 2.4400203536445746, + "grad_norm": 2.9299228191375732, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8957600593566895, + "num_tokens": 744418188.0, + "step": 19181 + }, + { + "epoch": 2.440147563923165, + "grad_norm": 2.706670045852661, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9002290964126587, + "num_tokens": 744464533.0, + "step": 19182 + }, + { + "epoch": 2.4402747742017556, + "grad_norm": 2.8081777095794678, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9108498096466064, + "num_tokens": 744502786.0, + "step": 19183 + }, + { + "epoch": 2.440401984480346, + "grad_norm": 2.7523930072784424, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.911757230758667, + "num_tokens": 744541997.0, + "step": 19184 + }, + { + "epoch": 2.4405291947589367, + "grad_norm": 2.574047803878784, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9096336960792542, + "num_tokens": 744586783.0, + "step": 19185 + }, + { + "epoch": 2.440656405037527, + "grad_norm": 2.840691089630127, + "learning_rate": 1e-06, + "loss": 0.2442, + "mean_token_accuracy": 0.9074180126190186, + "num_tokens": 744623495.0, + "step": 19186 + }, + { + "epoch": 2.4407836153161178, + "grad_norm": 2.927703619003296, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9032117128372192, + "num_tokens": 744660044.0, + "step": 19187 + }, + { + "epoch": 2.440910825594708, + "grad_norm": 3.1710245609283447, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.904161810874939, + "num_tokens": 744693560.0, + "step": 19188 + }, + { + "epoch": 2.4410380358732984, + "grad_norm": 3.1016130447387695, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8899519443511963, + "num_tokens": 744730805.0, + "step": 19189 + }, + { + "epoch": 2.441165246151889, + "grad_norm": 2.6956353187561035, + "learning_rate": 1e-06, + "loss": 0.2306, + "mean_token_accuracy": 0.9147784113883972, + "num_tokens": 744769035.0, + "step": 19190 + }, + { + "epoch": 2.4412924564304794, + "grad_norm": 2.756131887435913, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.902292788028717, + "num_tokens": 744816646.0, + "step": 19191 + }, + { + "epoch": 2.44141966670907, + "grad_norm": 2.7582852840423584, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9099559783935547, + "num_tokens": 744857109.0, + "step": 19192 + }, + { + "epoch": 2.4415468769876605, + "grad_norm": 3.2005088329315186, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.902552604675293, + "num_tokens": 744891538.0, + "step": 19193 + }, + { + "epoch": 2.441674087266251, + "grad_norm": 2.9703333377838135, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9084831476211548, + "num_tokens": 744927874.0, + "step": 19194 + }, + { + "epoch": 2.4418012975448415, + "grad_norm": 2.782463788986206, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.903641939163208, + "num_tokens": 744968076.0, + "step": 19195 + }, + { + "epoch": 2.441928507823432, + "grad_norm": 2.5213611125946045, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9098110198974609, + "num_tokens": 745012728.0, + "step": 19196 + }, + { + "epoch": 2.4420557181020226, + "grad_norm": 2.864072561264038, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8925328850746155, + "num_tokens": 745054939.0, + "step": 19197 + }, + { + "epoch": 2.442182928380613, + "grad_norm": 2.965024709701538, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9040768146514893, + "num_tokens": 745092215.0, + "step": 19198 + }, + { + "epoch": 2.4423101386592037, + "grad_norm": 2.7446272373199463, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9032864570617676, + "num_tokens": 745133326.0, + "step": 19199 + }, + { + "epoch": 2.442437348937794, + "grad_norm": 2.736912965774536, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.904498279094696, + "num_tokens": 745176494.0, + "step": 19200 + }, + { + "epoch": 2.4425645592163847, + "grad_norm": 2.8396060466766357, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9038506150245667, + "num_tokens": 745216954.0, + "step": 19201 + }, + { + "epoch": 2.4426917694949752, + "grad_norm": 2.8487796783447266, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9008678197860718, + "num_tokens": 745256775.0, + "step": 19202 + }, + { + "epoch": 2.4428189797735658, + "grad_norm": 2.9458038806915283, + "learning_rate": 1e-06, + "loss": 0.2862, + "mean_token_accuracy": 0.8966812491416931, + "num_tokens": 745294704.0, + "step": 19203 + }, + { + "epoch": 2.4429461900521563, + "grad_norm": 3.074524164199829, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.8999313712120056, + "num_tokens": 745332178.0, + "step": 19204 + }, + { + "epoch": 2.443073400330747, + "grad_norm": 2.7360000610351562, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9012872576713562, + "num_tokens": 745371734.0, + "step": 19205 + }, + { + "epoch": 2.4432006106093374, + "grad_norm": 2.9098708629608154, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9017876982688904, + "num_tokens": 745408745.0, + "step": 19206 + }, + { + "epoch": 2.443327820887928, + "grad_norm": 2.7911438941955566, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.910312294960022, + "num_tokens": 745449931.0, + "step": 19207 + }, + { + "epoch": 2.4434550311665184, + "grad_norm": 2.5721092224121094, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9063757658004761, + "num_tokens": 745493860.0, + "step": 19208 + }, + { + "epoch": 2.443582241445109, + "grad_norm": 2.7876086235046387, + "learning_rate": 1e-06, + "loss": 0.2304, + "mean_token_accuracy": 0.9144421815872192, + "num_tokens": 745526106.0, + "step": 19209 + }, + { + "epoch": 2.4437094517236995, + "grad_norm": 2.663142681121826, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9105415344238281, + "num_tokens": 745568744.0, + "step": 19210 + }, + { + "epoch": 2.4438366620022896, + "grad_norm": 3.099005937576294, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.8997377753257751, + "num_tokens": 745607359.0, + "step": 19211 + }, + { + "epoch": 2.4439638722808805, + "grad_norm": 2.972424268722534, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8938363790512085, + "num_tokens": 745647055.0, + "step": 19212 + }, + { + "epoch": 2.4440910825594706, + "grad_norm": 2.7875545024871826, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9060674905776978, + "num_tokens": 745686928.0, + "step": 19213 + }, + { + "epoch": 2.444218292838061, + "grad_norm": 2.7735819816589355, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9049347639083862, + "num_tokens": 745728758.0, + "step": 19214 + }, + { + "epoch": 2.4443455031166517, + "grad_norm": 2.764296293258667, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9112430810928345, + "num_tokens": 745768559.0, + "step": 19215 + }, + { + "epoch": 2.444472713395242, + "grad_norm": 2.760077476501465, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9024137258529663, + "num_tokens": 745809917.0, + "step": 19216 + }, + { + "epoch": 2.4445999236738327, + "grad_norm": 2.523097038269043, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9102843403816223, + "num_tokens": 745857434.0, + "step": 19217 + }, + { + "epoch": 2.4447271339524232, + "grad_norm": 2.6625964641571045, + "learning_rate": 1e-06, + "loss": 0.2364, + "mean_token_accuracy": 0.9120015501976013, + "num_tokens": 745897074.0, + "step": 19218 + }, + { + "epoch": 2.4448543442310138, + "grad_norm": 2.8741798400878906, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8978133201599121, + "num_tokens": 745940441.0, + "step": 19219 + }, + { + "epoch": 2.4449815545096043, + "grad_norm": 2.8991148471832275, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8984563946723938, + "num_tokens": 745980055.0, + "step": 19220 + }, + { + "epoch": 2.445108764788195, + "grad_norm": 2.9493422508239746, + "learning_rate": 1e-06, + "loss": 0.2348, + "mean_token_accuracy": 0.9119312763214111, + "num_tokens": 746014993.0, + "step": 19221 + }, + { + "epoch": 2.4452359750667854, + "grad_norm": 3.0291898250579834, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8963606357574463, + "num_tokens": 746052650.0, + "step": 19222 + }, + { + "epoch": 2.445363185345376, + "grad_norm": 2.803788900375366, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9088317155838013, + "num_tokens": 746090508.0, + "step": 19223 + }, + { + "epoch": 2.4454903956239664, + "grad_norm": 2.827526092529297, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9083511233329773, + "num_tokens": 746133574.0, + "step": 19224 + }, + { + "epoch": 2.445617605902557, + "grad_norm": 2.8955230712890625, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.902180552482605, + "num_tokens": 746171504.0, + "step": 19225 + }, + { + "epoch": 2.4457448161811475, + "grad_norm": 2.850029468536377, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8947585225105286, + "num_tokens": 746214383.0, + "step": 19226 + }, + { + "epoch": 2.445872026459738, + "grad_norm": 3.4137213230133057, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9043564796447754, + "num_tokens": 746242800.0, + "step": 19227 + }, + { + "epoch": 2.4459992367383285, + "grad_norm": 2.951093912124634, + "learning_rate": 1e-06, + "loss": 0.2405, + "mean_token_accuracy": 0.9097815752029419, + "num_tokens": 746277511.0, + "step": 19228 + }, + { + "epoch": 2.446126447016919, + "grad_norm": 2.7800629138946533, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9079328775405884, + "num_tokens": 746317128.0, + "step": 19229 + }, + { + "epoch": 2.4462536572955096, + "grad_norm": 2.8031134605407715, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9106981754302979, + "num_tokens": 746355285.0, + "step": 19230 + }, + { + "epoch": 2.4463808675741, + "grad_norm": 2.869997978210449, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9057486057281494, + "num_tokens": 746398987.0, + "step": 19231 + }, + { + "epoch": 2.4465080778526906, + "grad_norm": 2.7747347354888916, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.902450442314148, + "num_tokens": 746439209.0, + "step": 19232 + }, + { + "epoch": 2.446635288131281, + "grad_norm": 2.8590071201324463, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9080202579498291, + "num_tokens": 746479737.0, + "step": 19233 + }, + { + "epoch": 2.4467624984098717, + "grad_norm": 3.007866859436035, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.8925774097442627, + "num_tokens": 746518924.0, + "step": 19234 + }, + { + "epoch": 2.4468897086884622, + "grad_norm": 2.926501750946045, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9109782576560974, + "num_tokens": 746555149.0, + "step": 19235 + }, + { + "epoch": 2.4470169189670523, + "grad_norm": 3.025585412979126, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9047718048095703, + "num_tokens": 746592159.0, + "step": 19236 + }, + { + "epoch": 2.4471441292456433, + "grad_norm": 2.8558123111724854, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8964738249778748, + "num_tokens": 746632718.0, + "step": 19237 + }, + { + "epoch": 2.4472713395242334, + "grad_norm": 2.573594331741333, + "learning_rate": 1e-06, + "loss": 0.2206, + "mean_token_accuracy": 0.9141926169395447, + "num_tokens": 746673692.0, + "step": 19238 + }, + { + "epoch": 2.447398549802824, + "grad_norm": 2.808976173400879, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9150238037109375, + "num_tokens": 746712266.0, + "step": 19239 + }, + { + "epoch": 2.4475257600814144, + "grad_norm": 2.720569133758545, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9100565910339355, + "num_tokens": 746752557.0, + "step": 19240 + }, + { + "epoch": 2.447652970360005, + "grad_norm": 2.758420467376709, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9057146310806274, + "num_tokens": 746794971.0, + "step": 19241 + }, + { + "epoch": 2.4477801806385955, + "grad_norm": 3.0269265174865723, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9035950303077698, + "num_tokens": 746831656.0, + "step": 19242 + }, + { + "epoch": 2.447907390917186, + "grad_norm": 2.787294387817383, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9012345671653748, + "num_tokens": 746872650.0, + "step": 19243 + }, + { + "epoch": 2.4480346011957765, + "grad_norm": 2.7871878147125244, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9009813070297241, + "num_tokens": 746916943.0, + "step": 19244 + }, + { + "epoch": 2.448161811474367, + "grad_norm": 2.988781452178955, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9076266884803772, + "num_tokens": 746953364.0, + "step": 19245 + }, + { + "epoch": 2.4482890217529576, + "grad_norm": 3.008390188217163, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9029330015182495, + "num_tokens": 746989144.0, + "step": 19246 + }, + { + "epoch": 2.448416232031548, + "grad_norm": 2.97243332862854, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9071244597434998, + "num_tokens": 747028961.0, + "step": 19247 + }, + { + "epoch": 2.4485434423101387, + "grad_norm": 2.898303985595703, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9126700758934021, + "num_tokens": 747064441.0, + "step": 19248 + }, + { + "epoch": 2.448670652588729, + "grad_norm": 2.8852148056030273, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9048168659210205, + "num_tokens": 747103917.0, + "step": 19249 + }, + { + "epoch": 2.4487978628673197, + "grad_norm": 2.8443892002105713, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9123430252075195, + "num_tokens": 747143000.0, + "step": 19250 + }, + { + "epoch": 2.4489250731459102, + "grad_norm": 2.863050699234009, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9114722013473511, + "num_tokens": 747180934.0, + "step": 19251 + }, + { + "epoch": 2.4490522834245008, + "grad_norm": 3.3295161724090576, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8899834156036377, + "num_tokens": 747214534.0, + "step": 19252 + }, + { + "epoch": 2.4491794937030913, + "grad_norm": 2.6520721912384033, + "learning_rate": 1e-06, + "loss": 0.2306, + "mean_token_accuracy": 0.9150441884994507, + "num_tokens": 747255839.0, + "step": 19253 + }, + { + "epoch": 2.449306703981682, + "grad_norm": 3.4347665309906006, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8978472352027893, + "num_tokens": 747289290.0, + "step": 19254 + }, + { + "epoch": 2.4494339142602723, + "grad_norm": 3.057311773300171, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8963493704795837, + "num_tokens": 747328945.0, + "step": 19255 + }, + { + "epoch": 2.449561124538863, + "grad_norm": 2.9414560794830322, + "learning_rate": 1e-06, + "loss": 0.2322, + "mean_token_accuracy": 0.9149540066719055, + "num_tokens": 747363817.0, + "step": 19256 + }, + { + "epoch": 2.4496883348174534, + "grad_norm": 2.9637115001678467, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.8996260166168213, + "num_tokens": 747403291.0, + "step": 19257 + }, + { + "epoch": 2.449815545096044, + "grad_norm": 2.8100571632385254, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9070701599121094, + "num_tokens": 747444540.0, + "step": 19258 + }, + { + "epoch": 2.4499427553746345, + "grad_norm": 2.708775758743286, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9035052061080933, + "num_tokens": 747489774.0, + "step": 19259 + }, + { + "epoch": 2.450069965653225, + "grad_norm": 2.7687602043151855, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9083622694015503, + "num_tokens": 747531783.0, + "step": 19260 + }, + { + "epoch": 2.450197175931815, + "grad_norm": 2.564246416091919, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9115100502967834, + "num_tokens": 747578834.0, + "step": 19261 + }, + { + "epoch": 2.450324386210406, + "grad_norm": 2.8143959045410156, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8978884816169739, + "num_tokens": 747620015.0, + "step": 19262 + }, + { + "epoch": 2.450451596488996, + "grad_norm": 2.7760655879974365, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9059937596321106, + "num_tokens": 747658675.0, + "step": 19263 + }, + { + "epoch": 2.4505788067675867, + "grad_norm": 2.8394463062286377, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.8991721868515015, + "num_tokens": 747700294.0, + "step": 19264 + }, + { + "epoch": 2.450706017046177, + "grad_norm": 2.689220428466797, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8933019042015076, + "num_tokens": 747745488.0, + "step": 19265 + }, + { + "epoch": 2.4508332273247677, + "grad_norm": 2.6575777530670166, + "learning_rate": 1e-06, + "loss": 0.2347, + "mean_token_accuracy": 0.9136896133422852, + "num_tokens": 747786900.0, + "step": 19266 + }, + { + "epoch": 2.4509604376033582, + "grad_norm": 3.0142157077789307, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8934557437896729, + "num_tokens": 747830025.0, + "step": 19267 + }, + { + "epoch": 2.4510876478819488, + "grad_norm": 3.1529147624969482, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.8996540904045105, + "num_tokens": 747866389.0, + "step": 19268 + }, + { + "epoch": 2.4512148581605393, + "grad_norm": 2.745835065841675, + "learning_rate": 1e-06, + "loss": 0.2188, + "mean_token_accuracy": 0.917405903339386, + "num_tokens": 747903106.0, + "step": 19269 + }, + { + "epoch": 2.45134206843913, + "grad_norm": 2.8672878742218018, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9034748673439026, + "num_tokens": 747942790.0, + "step": 19270 + }, + { + "epoch": 2.4514692787177204, + "grad_norm": 3.145127534866333, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8939248919487, + "num_tokens": 747978677.0, + "step": 19271 + }, + { + "epoch": 2.451596488996311, + "grad_norm": 3.0728065967559814, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.893774688243866, + "num_tokens": 748017080.0, + "step": 19272 + }, + { + "epoch": 2.4517236992749014, + "grad_norm": 2.994929313659668, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9003419280052185, + "num_tokens": 748057498.0, + "step": 19273 + }, + { + "epoch": 2.451850909553492, + "grad_norm": 2.8823611736297607, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9031758308410645, + "num_tokens": 748096163.0, + "step": 19274 + }, + { + "epoch": 2.4519781198320825, + "grad_norm": 2.6998844146728516, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9049144983291626, + "num_tokens": 748143186.0, + "step": 19275 + }, + { + "epoch": 2.452105330110673, + "grad_norm": 2.971045732498169, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9127069711685181, + "num_tokens": 748176749.0, + "step": 19276 + }, + { + "epoch": 2.4522325403892635, + "grad_norm": 3.1637518405914307, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8971284627914429, + "num_tokens": 748210480.0, + "step": 19277 + }, + { + "epoch": 2.452359750667854, + "grad_norm": 2.971616506576538, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9016245603561401, + "num_tokens": 748249158.0, + "step": 19278 + }, + { + "epoch": 2.4524869609464446, + "grad_norm": 2.763629198074341, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9099245071411133, + "num_tokens": 748289373.0, + "step": 19279 + }, + { + "epoch": 2.452614171225035, + "grad_norm": 2.905766487121582, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9095303416252136, + "num_tokens": 748325164.0, + "step": 19280 + }, + { + "epoch": 2.4527413815036256, + "grad_norm": 2.9766998291015625, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9094670414924622, + "num_tokens": 748359857.0, + "step": 19281 + }, + { + "epoch": 2.452868591782216, + "grad_norm": 3.083508014678955, + "learning_rate": 1e-06, + "loss": 0.2437, + "mean_token_accuracy": 0.9093852639198303, + "num_tokens": 748394534.0, + "step": 19282 + }, + { + "epoch": 2.4529958020608067, + "grad_norm": 2.7944767475128174, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9088062047958374, + "num_tokens": 748433928.0, + "step": 19283 + }, + { + "epoch": 2.453123012339397, + "grad_norm": 2.8329508304595947, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9057448506355286, + "num_tokens": 748474200.0, + "step": 19284 + }, + { + "epoch": 2.4532502226179878, + "grad_norm": 2.705239772796631, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9113798141479492, + "num_tokens": 748514162.0, + "step": 19285 + }, + { + "epoch": 2.453377432896578, + "grad_norm": 2.910496950149536, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9020178318023682, + "num_tokens": 748553212.0, + "step": 19286 + }, + { + "epoch": 2.4535046431751684, + "grad_norm": 2.9396400451660156, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.91208815574646, + "num_tokens": 748589043.0, + "step": 19287 + }, + { + "epoch": 2.453631853453759, + "grad_norm": 2.851393938064575, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9068343639373779, + "num_tokens": 748624426.0, + "step": 19288 + }, + { + "epoch": 2.4537590637323494, + "grad_norm": 2.7484567165374756, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.8997861742973328, + "num_tokens": 748664519.0, + "step": 19289 + }, + { + "epoch": 2.45388627401094, + "grad_norm": 2.933652877807617, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9045128226280212, + "num_tokens": 748699872.0, + "step": 19290 + }, + { + "epoch": 2.4540134842895305, + "grad_norm": 2.797492504119873, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9033281803131104, + "num_tokens": 748742592.0, + "step": 19291 + }, + { + "epoch": 2.454140694568121, + "grad_norm": 2.870582103729248, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8991737365722656, + "num_tokens": 748784490.0, + "step": 19292 + }, + { + "epoch": 2.4542679048467115, + "grad_norm": 2.6543781757354736, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9020235538482666, + "num_tokens": 748829843.0, + "step": 19293 + }, + { + "epoch": 2.454395115125302, + "grad_norm": 2.734471321105957, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9017810225486755, + "num_tokens": 748873630.0, + "step": 19294 + }, + { + "epoch": 2.4545223254038926, + "grad_norm": 2.977480173110962, + "learning_rate": 1e-06, + "loss": 0.296, + "mean_token_accuracy": 0.8905704021453857, + "num_tokens": 748911778.0, + "step": 19295 + }, + { + "epoch": 2.454649535682483, + "grad_norm": 2.9491124153137207, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9050566554069519, + "num_tokens": 748949646.0, + "step": 19296 + }, + { + "epoch": 2.4547767459610736, + "grad_norm": 3.0778257846832275, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9019741415977478, + "num_tokens": 748987106.0, + "step": 19297 + }, + { + "epoch": 2.454903956239664, + "grad_norm": 2.795227527618408, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.907713770866394, + "num_tokens": 749026162.0, + "step": 19298 + }, + { + "epoch": 2.4550311665182547, + "grad_norm": 3.1170198917388916, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9016736745834351, + "num_tokens": 749063179.0, + "step": 19299 + }, + { + "epoch": 2.4551583767968452, + "grad_norm": 3.1708221435546875, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9003971815109253, + "num_tokens": 749096519.0, + "step": 19300 + }, + { + "epoch": 2.4552855870754358, + "grad_norm": 3.1596009731292725, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8992607593536377, + "num_tokens": 749130645.0, + "step": 19301 + }, + { + "epoch": 2.4554127973540263, + "grad_norm": 3.0222232341766357, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.90233314037323, + "num_tokens": 749165607.0, + "step": 19302 + }, + { + "epoch": 2.455540007632617, + "grad_norm": 2.6817703247070312, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9100393056869507, + "num_tokens": 749206488.0, + "step": 19303 + }, + { + "epoch": 2.4556672179112073, + "grad_norm": 2.8214123249053955, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9084038138389587, + "num_tokens": 749245879.0, + "step": 19304 + }, + { + "epoch": 2.455794428189798, + "grad_norm": 2.750995397567749, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9063096046447754, + "num_tokens": 749286947.0, + "step": 19305 + }, + { + "epoch": 2.4559216384683884, + "grad_norm": 3.1612319946289062, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9032403230667114, + "num_tokens": 749321598.0, + "step": 19306 + }, + { + "epoch": 2.456048848746979, + "grad_norm": 3.117790937423706, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9098026752471924, + "num_tokens": 749356530.0, + "step": 19307 + }, + { + "epoch": 2.4561760590255695, + "grad_norm": 2.9296488761901855, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.9003707766532898, + "num_tokens": 749395935.0, + "step": 19308 + }, + { + "epoch": 2.4563032693041595, + "grad_norm": 2.8784799575805664, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9019731283187866, + "num_tokens": 749434270.0, + "step": 19309 + }, + { + "epoch": 2.4564304795827505, + "grad_norm": 2.5807321071624756, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9058062434196472, + "num_tokens": 749479701.0, + "step": 19310 + }, + { + "epoch": 2.4565576898613406, + "grad_norm": 2.9796509742736816, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9102100133895874, + "num_tokens": 749516185.0, + "step": 19311 + }, + { + "epoch": 2.456684900139931, + "grad_norm": 3.037792682647705, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9080137014389038, + "num_tokens": 749550962.0, + "step": 19312 + }, + { + "epoch": 2.4568121104185217, + "grad_norm": 2.8473126888275146, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9063664078712463, + "num_tokens": 749590892.0, + "step": 19313 + }, + { + "epoch": 2.456939320697112, + "grad_norm": 2.798414707183838, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9086758494377136, + "num_tokens": 749628119.0, + "step": 19314 + }, + { + "epoch": 2.4570665309757027, + "grad_norm": 3.3146328926086426, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.901610255241394, + "num_tokens": 749658368.0, + "step": 19315 + }, + { + "epoch": 2.4571937412542932, + "grad_norm": 3.0977976322174072, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.897829532623291, + "num_tokens": 749698312.0, + "step": 19316 + }, + { + "epoch": 2.4573209515328838, + "grad_norm": 3.1387405395507812, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9078737497329712, + "num_tokens": 749732823.0, + "step": 19317 + }, + { + "epoch": 2.4574481618114743, + "grad_norm": 2.732192039489746, + "learning_rate": 1e-06, + "loss": 0.2195, + "mean_token_accuracy": 0.9193602800369263, + "num_tokens": 749773392.0, + "step": 19318 + }, + { + "epoch": 2.457575372090065, + "grad_norm": 2.7306995391845703, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9077964425086975, + "num_tokens": 749812848.0, + "step": 19319 + }, + { + "epoch": 2.4577025823686554, + "grad_norm": 2.842989444732666, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9007617235183716, + "num_tokens": 749854133.0, + "step": 19320 + }, + { + "epoch": 2.457829792647246, + "grad_norm": 2.8155720233917236, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9059306383132935, + "num_tokens": 749893605.0, + "step": 19321 + }, + { + "epoch": 2.4579570029258364, + "grad_norm": 2.9001219272613525, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.9110867977142334, + "num_tokens": 749929873.0, + "step": 19322 + }, + { + "epoch": 2.458084213204427, + "grad_norm": 2.7758266925811768, + "learning_rate": 1e-06, + "loss": 0.2127, + "mean_token_accuracy": 0.9183499813079834, + "num_tokens": 749966633.0, + "step": 19323 + }, + { + "epoch": 2.4582114234830175, + "grad_norm": 2.584768533706665, + "learning_rate": 1e-06, + "loss": 0.2283, + "mean_token_accuracy": 0.9162507057189941, + "num_tokens": 750009556.0, + "step": 19324 + }, + { + "epoch": 2.458338633761608, + "grad_norm": 2.8351986408233643, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.900539755821228, + "num_tokens": 750047579.0, + "step": 19325 + }, + { + "epoch": 2.4584658440401985, + "grad_norm": 3.08469820022583, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8919432759284973, + "num_tokens": 750086908.0, + "step": 19326 + }, + { + "epoch": 2.458593054318789, + "grad_norm": 2.962678909301758, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9037256240844727, + "num_tokens": 750123860.0, + "step": 19327 + }, + { + "epoch": 2.4587202645973796, + "grad_norm": 2.945491313934326, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9067471623420715, + "num_tokens": 750162116.0, + "step": 19328 + }, + { + "epoch": 2.45884747487597, + "grad_norm": 2.7641074657440186, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9085851311683655, + "num_tokens": 750201547.0, + "step": 19329 + }, + { + "epoch": 2.4589746851545606, + "grad_norm": 2.581228256225586, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9105768799781799, + "num_tokens": 750247066.0, + "step": 19330 + }, + { + "epoch": 2.459101895433151, + "grad_norm": 2.71347975730896, + "learning_rate": 1e-06, + "loss": 0.2315, + "mean_token_accuracy": 0.9112213850021362, + "num_tokens": 750289501.0, + "step": 19331 + }, + { + "epoch": 2.4592291057117417, + "grad_norm": 3.1963114738464355, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9002788066864014, + "num_tokens": 750322028.0, + "step": 19332 + }, + { + "epoch": 2.459356315990332, + "grad_norm": 3.107682704925537, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9044641852378845, + "num_tokens": 750355130.0, + "step": 19333 + }, + { + "epoch": 2.4594835262689223, + "grad_norm": 2.9966635704040527, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9099169969558716, + "num_tokens": 750388015.0, + "step": 19334 + }, + { + "epoch": 2.4596107365475133, + "grad_norm": 2.6138699054718018, + "learning_rate": 1e-06, + "loss": 0.214, + "mean_token_accuracy": 0.9208030104637146, + "num_tokens": 750427803.0, + "step": 19335 + }, + { + "epoch": 2.4597379468261034, + "grad_norm": 3.0899672508239746, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8922388553619385, + "num_tokens": 750467807.0, + "step": 19336 + }, + { + "epoch": 2.459865157104694, + "grad_norm": 2.96016526222229, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.912035346031189, + "num_tokens": 750503673.0, + "step": 19337 + }, + { + "epoch": 2.4599923673832844, + "grad_norm": 2.7537343502044678, + "learning_rate": 1e-06, + "loss": 0.2348, + "mean_token_accuracy": 0.9131056070327759, + "num_tokens": 750542761.0, + "step": 19338 + }, + { + "epoch": 2.460119577661875, + "grad_norm": 3.376148223876953, + "learning_rate": 1e-06, + "loss": 0.3143, + "mean_token_accuracy": 0.883449137210846, + "num_tokens": 750581151.0, + "step": 19339 + }, + { + "epoch": 2.4602467879404655, + "grad_norm": 3.047962188720703, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9098957777023315, + "num_tokens": 750615161.0, + "step": 19340 + }, + { + "epoch": 2.460373998219056, + "grad_norm": 2.975412607192993, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8891178369522095, + "num_tokens": 750658527.0, + "step": 19341 + }, + { + "epoch": 2.4605012084976465, + "grad_norm": 2.850729465484619, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9074894785881042, + "num_tokens": 750699454.0, + "step": 19342 + }, + { + "epoch": 2.460628418776237, + "grad_norm": 2.656881093978882, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9111709594726562, + "num_tokens": 750744472.0, + "step": 19343 + }, + { + "epoch": 2.4607556290548276, + "grad_norm": 3.199467420578003, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8993333578109741, + "num_tokens": 750781692.0, + "step": 19344 + }, + { + "epoch": 2.460882839333418, + "grad_norm": 2.714545965194702, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9100539684295654, + "num_tokens": 750821814.0, + "step": 19345 + }, + { + "epoch": 2.4610100496120086, + "grad_norm": 2.7923245429992676, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.9131300449371338, + "num_tokens": 750858061.0, + "step": 19346 + }, + { + "epoch": 2.461137259890599, + "grad_norm": 2.83594012260437, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9092646837234497, + "num_tokens": 750895705.0, + "step": 19347 + }, + { + "epoch": 2.4612644701691897, + "grad_norm": 2.6941637992858887, + "learning_rate": 1e-06, + "loss": 0.2272, + "mean_token_accuracy": 0.9166250228881836, + "num_tokens": 750935465.0, + "step": 19348 + }, + { + "epoch": 2.4613916804477802, + "grad_norm": 2.924741744995117, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8955289125442505, + "num_tokens": 750973960.0, + "step": 19349 + }, + { + "epoch": 2.4615188907263708, + "grad_norm": 2.9504265785217285, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9036245942115784, + "num_tokens": 751013434.0, + "step": 19350 + }, + { + "epoch": 2.4616461010049613, + "grad_norm": 3.408210277557373, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8966847658157349, + "num_tokens": 751043244.0, + "step": 19351 + }, + { + "epoch": 2.461773311283552, + "grad_norm": 3.173788070678711, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.9009982347488403, + "num_tokens": 751080050.0, + "step": 19352 + }, + { + "epoch": 2.4619005215621423, + "grad_norm": 3.0873095989227295, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9102808237075806, + "num_tokens": 751113306.0, + "step": 19353 + }, + { + "epoch": 2.462027731840733, + "grad_norm": 2.934354782104492, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9027015566825867, + "num_tokens": 751151368.0, + "step": 19354 + }, + { + "epoch": 2.4621549421193234, + "grad_norm": 2.6925582885742188, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.8989047408103943, + "num_tokens": 751197599.0, + "step": 19355 + }, + { + "epoch": 2.462282152397914, + "grad_norm": 3.194257974624634, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.902977466583252, + "num_tokens": 751229647.0, + "step": 19356 + }, + { + "epoch": 2.4624093626765045, + "grad_norm": 3.057171583175659, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.893466591835022, + "num_tokens": 751267515.0, + "step": 19357 + }, + { + "epoch": 2.462536572955095, + "grad_norm": 2.759437322616577, + "learning_rate": 1e-06, + "loss": 0.2405, + "mean_token_accuracy": 0.9116698503494263, + "num_tokens": 751307372.0, + "step": 19358 + }, + { + "epoch": 2.462663783233685, + "grad_norm": 2.896702766418457, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9037846326828003, + "num_tokens": 751349018.0, + "step": 19359 + }, + { + "epoch": 2.462790993512276, + "grad_norm": 3.183302879333496, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.8992425203323364, + "num_tokens": 751381158.0, + "step": 19360 + }, + { + "epoch": 2.462918203790866, + "grad_norm": 2.8134958744049072, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9094249606132507, + "num_tokens": 751418050.0, + "step": 19361 + }, + { + "epoch": 2.4630454140694567, + "grad_norm": 2.7654786109924316, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.899832010269165, + "num_tokens": 751462654.0, + "step": 19362 + }, + { + "epoch": 2.463172624348047, + "grad_norm": 2.7516703605651855, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9097471237182617, + "num_tokens": 751502694.0, + "step": 19363 + }, + { + "epoch": 2.4632998346266377, + "grad_norm": 3.0535006523132324, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8907195329666138, + "num_tokens": 751543231.0, + "step": 19364 + }, + { + "epoch": 2.4634270449052282, + "grad_norm": 2.903637170791626, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9087734818458557, + "num_tokens": 751579468.0, + "step": 19365 + }, + { + "epoch": 2.4635542551838188, + "grad_norm": 2.7625768184661865, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.9009347558021545, + "num_tokens": 751623499.0, + "step": 19366 + }, + { + "epoch": 2.4636814654624093, + "grad_norm": 2.782949686050415, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.900236964225769, + "num_tokens": 751663166.0, + "step": 19367 + }, + { + "epoch": 2.463808675741, + "grad_norm": 2.843125104904175, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.8993955850601196, + "num_tokens": 751704016.0, + "step": 19368 + }, + { + "epoch": 2.4639358860195903, + "grad_norm": 3.038926362991333, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8984217643737793, + "num_tokens": 751739671.0, + "step": 19369 + }, + { + "epoch": 2.464063096298181, + "grad_norm": 3.3065762519836426, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8954163789749146, + "num_tokens": 751773502.0, + "step": 19370 + }, + { + "epoch": 2.4641903065767714, + "grad_norm": 3.113252639770508, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9000424742698669, + "num_tokens": 751807460.0, + "step": 19371 + }, + { + "epoch": 2.464317516855362, + "grad_norm": 3.0517754554748535, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9041844010353088, + "num_tokens": 751840503.0, + "step": 19372 + }, + { + "epoch": 2.4644447271339525, + "grad_norm": 2.9422905445098877, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9000412821769714, + "num_tokens": 751879636.0, + "step": 19373 + }, + { + "epoch": 2.464571937412543, + "grad_norm": 3.287858009338379, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.897475004196167, + "num_tokens": 751917125.0, + "step": 19374 + }, + { + "epoch": 2.4646991476911335, + "grad_norm": 2.693120241165161, + "learning_rate": 1e-06, + "loss": 0.232, + "mean_token_accuracy": 0.913887619972229, + "num_tokens": 751955160.0, + "step": 19375 + }, + { + "epoch": 2.464826357969724, + "grad_norm": 3.1471478939056396, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8966529369354248, + "num_tokens": 751990722.0, + "step": 19376 + }, + { + "epoch": 2.4649535682483146, + "grad_norm": 3.065211534500122, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9011545181274414, + "num_tokens": 752022616.0, + "step": 19377 + }, + { + "epoch": 2.465080778526905, + "grad_norm": 2.805413007736206, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9064147472381592, + "num_tokens": 752062711.0, + "step": 19378 + }, + { + "epoch": 2.4652079888054956, + "grad_norm": 3.064868688583374, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8951100707054138, + "num_tokens": 752102810.0, + "step": 19379 + }, + { + "epoch": 2.465335199084086, + "grad_norm": 3.0644586086273193, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9022626280784607, + "num_tokens": 752137841.0, + "step": 19380 + }, + { + "epoch": 2.4654624093626767, + "grad_norm": 2.879394769668579, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9047553539276123, + "num_tokens": 752172953.0, + "step": 19381 + }, + { + "epoch": 2.4655896196412668, + "grad_norm": 2.868746280670166, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8995213508605957, + "num_tokens": 752214276.0, + "step": 19382 + }, + { + "epoch": 2.4657168299198577, + "grad_norm": 3.31728458404541, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8936395645141602, + "num_tokens": 752248846.0, + "step": 19383 + }, + { + "epoch": 2.465844040198448, + "grad_norm": 3.0102365016937256, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8962622284889221, + "num_tokens": 752287784.0, + "step": 19384 + }, + { + "epoch": 2.4659712504770384, + "grad_norm": 2.6928889751434326, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9037815928459167, + "num_tokens": 752331966.0, + "step": 19385 + }, + { + "epoch": 2.466098460755629, + "grad_norm": 2.717759370803833, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9118584394454956, + "num_tokens": 752369142.0, + "step": 19386 + }, + { + "epoch": 2.4662256710342194, + "grad_norm": 2.819777011871338, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9072673320770264, + "num_tokens": 752407259.0, + "step": 19387 + }, + { + "epoch": 2.46635288131281, + "grad_norm": 2.725048065185547, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9026289582252502, + "num_tokens": 752448923.0, + "step": 19388 + }, + { + "epoch": 2.4664800915914005, + "grad_norm": 2.883100986480713, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9033843874931335, + "num_tokens": 752487950.0, + "step": 19389 + }, + { + "epoch": 2.466607301869991, + "grad_norm": 2.673825740814209, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9044774770736694, + "num_tokens": 752531268.0, + "step": 19390 + }, + { + "epoch": 2.4667345121485815, + "grad_norm": 2.9067299365997314, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9057052135467529, + "num_tokens": 752566875.0, + "step": 19391 + }, + { + "epoch": 2.466861722427172, + "grad_norm": 2.7233595848083496, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.906898558139801, + "num_tokens": 752605357.0, + "step": 19392 + }, + { + "epoch": 2.4669889327057626, + "grad_norm": 2.6095921993255615, + "learning_rate": 1e-06, + "loss": 0.2241, + "mean_token_accuracy": 0.9158673882484436, + "num_tokens": 752647048.0, + "step": 19393 + }, + { + "epoch": 2.467116142984353, + "grad_norm": 2.6966724395751953, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9114137291908264, + "num_tokens": 752686128.0, + "step": 19394 + }, + { + "epoch": 2.4672433532629436, + "grad_norm": 2.7615230083465576, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9081497192382812, + "num_tokens": 752723532.0, + "step": 19395 + }, + { + "epoch": 2.467370563541534, + "grad_norm": 2.7227261066436768, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9008965492248535, + "num_tokens": 752765404.0, + "step": 19396 + }, + { + "epoch": 2.4674977738201247, + "grad_norm": 3.022510528564453, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8972021341323853, + "num_tokens": 752805485.0, + "step": 19397 + }, + { + "epoch": 2.4676249840987152, + "grad_norm": 3.0921120643615723, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9056264162063599, + "num_tokens": 752841946.0, + "step": 19398 + }, + { + "epoch": 2.4677521943773058, + "grad_norm": 3.0608084201812744, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.8995770812034607, + "num_tokens": 752878071.0, + "step": 19399 + }, + { + "epoch": 2.4678794046558963, + "grad_norm": 2.8526735305786133, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9017228484153748, + "num_tokens": 752922054.0, + "step": 19400 + }, + { + "epoch": 2.468006614934487, + "grad_norm": 2.8588128089904785, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9059819579124451, + "num_tokens": 752964976.0, + "step": 19401 + }, + { + "epoch": 2.4681338252130773, + "grad_norm": 3.169663190841675, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.895665168762207, + "num_tokens": 753001487.0, + "step": 19402 + }, + { + "epoch": 2.468261035491668, + "grad_norm": 2.871131658554077, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.903152585029602, + "num_tokens": 753041262.0, + "step": 19403 + }, + { + "epoch": 2.4683882457702584, + "grad_norm": 2.925910711288452, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9022998809814453, + "num_tokens": 753079083.0, + "step": 19404 + }, + { + "epoch": 2.468515456048849, + "grad_norm": 2.677633047103882, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9078648090362549, + "num_tokens": 753120516.0, + "step": 19405 + }, + { + "epoch": 2.4686426663274394, + "grad_norm": 2.86944580078125, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9113914966583252, + "num_tokens": 753157722.0, + "step": 19406 + }, + { + "epoch": 2.4687698766060295, + "grad_norm": 2.781952142715454, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.89863121509552, + "num_tokens": 753198950.0, + "step": 19407 + }, + { + "epoch": 2.4688970868846205, + "grad_norm": 3.222468137741089, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8966559171676636, + "num_tokens": 753238561.0, + "step": 19408 + }, + { + "epoch": 2.4690242971632106, + "grad_norm": 2.7366943359375, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9097568392753601, + "num_tokens": 753278701.0, + "step": 19409 + }, + { + "epoch": 2.469151507441801, + "grad_norm": 2.7507169246673584, + "learning_rate": 1e-06, + "loss": 0.2178, + "mean_token_accuracy": 0.9177719354629517, + "num_tokens": 753314675.0, + "step": 19410 + }, + { + "epoch": 2.4692787177203916, + "grad_norm": 2.734718084335327, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9003599882125854, + "num_tokens": 753359036.0, + "step": 19411 + }, + { + "epoch": 2.469405927998982, + "grad_norm": 2.7464487552642822, + "learning_rate": 1e-06, + "loss": 0.233, + "mean_token_accuracy": 0.9135366678237915, + "num_tokens": 753398838.0, + "step": 19412 + }, + { + "epoch": 2.4695331382775727, + "grad_norm": 2.967241048812866, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9105570912361145, + "num_tokens": 753432450.0, + "step": 19413 + }, + { + "epoch": 2.4696603485561632, + "grad_norm": 2.7791683673858643, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9019738435745239, + "num_tokens": 753475394.0, + "step": 19414 + }, + { + "epoch": 2.4697875588347538, + "grad_norm": 3.1372461318969727, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9035696387290955, + "num_tokens": 753512065.0, + "step": 19415 + }, + { + "epoch": 2.4699147691133443, + "grad_norm": 2.8095240592956543, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9078687429428101, + "num_tokens": 753552044.0, + "step": 19416 + }, + { + "epoch": 2.470041979391935, + "grad_norm": 2.8632216453552246, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8989390134811401, + "num_tokens": 753594060.0, + "step": 19417 + }, + { + "epoch": 2.4701691896705253, + "grad_norm": 2.999626636505127, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.897789478302002, + "num_tokens": 753636072.0, + "step": 19418 + }, + { + "epoch": 2.470296399949116, + "grad_norm": 2.7362921237945557, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9043149948120117, + "num_tokens": 753679833.0, + "step": 19419 + }, + { + "epoch": 2.4704236102277064, + "grad_norm": 2.8830056190490723, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9065961837768555, + "num_tokens": 753716836.0, + "step": 19420 + }, + { + "epoch": 2.470550820506297, + "grad_norm": 3.1277451515197754, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.897879958152771, + "num_tokens": 753753871.0, + "step": 19421 + }, + { + "epoch": 2.4706780307848875, + "grad_norm": 3.1705996990203857, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8958553075790405, + "num_tokens": 753787954.0, + "step": 19422 + }, + { + "epoch": 2.470805241063478, + "grad_norm": 2.9411282539367676, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.8973326683044434, + "num_tokens": 753828980.0, + "step": 19423 + }, + { + "epoch": 2.4709324513420685, + "grad_norm": 2.8856914043426514, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.912780225276947, + "num_tokens": 753865578.0, + "step": 19424 + }, + { + "epoch": 2.471059661620659, + "grad_norm": 2.683915615081787, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9098079204559326, + "num_tokens": 753909820.0, + "step": 19425 + }, + { + "epoch": 2.4711868718992496, + "grad_norm": 3.2423255443573, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8997432589530945, + "num_tokens": 753944533.0, + "step": 19426 + }, + { + "epoch": 2.47131408217784, + "grad_norm": 2.7718260288238525, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9020724892616272, + "num_tokens": 753987980.0, + "step": 19427 + }, + { + "epoch": 2.4714412924564306, + "grad_norm": 2.9278852939605713, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9077849984169006, + "num_tokens": 754024482.0, + "step": 19428 + }, + { + "epoch": 2.471568502735021, + "grad_norm": 2.7657153606414795, + "learning_rate": 1e-06, + "loss": 0.2337, + "mean_token_accuracy": 0.9120391607284546, + "num_tokens": 754062294.0, + "step": 19429 + }, + { + "epoch": 2.4716957130136117, + "grad_norm": 2.7124946117401123, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9086728096008301, + "num_tokens": 754102447.0, + "step": 19430 + }, + { + "epoch": 2.471822923292202, + "grad_norm": 3.1972084045410156, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8903014659881592, + "num_tokens": 754141893.0, + "step": 19431 + }, + { + "epoch": 2.4719501335707923, + "grad_norm": 2.910088300704956, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9124727249145508, + "num_tokens": 754177147.0, + "step": 19432 + }, + { + "epoch": 2.4720773438493833, + "grad_norm": 2.861307382583618, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.8987666368484497, + "num_tokens": 754216903.0, + "step": 19433 + }, + { + "epoch": 2.4722045541279734, + "grad_norm": 2.8962128162384033, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9042816162109375, + "num_tokens": 754252541.0, + "step": 19434 + }, + { + "epoch": 2.472331764406564, + "grad_norm": 3.297955274581909, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9022773504257202, + "num_tokens": 754283103.0, + "step": 19435 + }, + { + "epoch": 2.4724589746851544, + "grad_norm": 2.7327585220336914, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8979262113571167, + "num_tokens": 754329474.0, + "step": 19436 + }, + { + "epoch": 2.472586184963745, + "grad_norm": 2.9355595111846924, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8972284197807312, + "num_tokens": 754369058.0, + "step": 19437 + }, + { + "epoch": 2.4727133952423355, + "grad_norm": 3.124382257461548, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.8964474201202393, + "num_tokens": 754406459.0, + "step": 19438 + }, + { + "epoch": 2.472840605520926, + "grad_norm": 2.901059150695801, + "learning_rate": 1e-06, + "loss": 0.2294, + "mean_token_accuracy": 0.9145787954330444, + "num_tokens": 754441786.0, + "step": 19439 + }, + { + "epoch": 2.4729678157995165, + "grad_norm": 2.7535104751586914, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.8997737765312195, + "num_tokens": 754481055.0, + "step": 19440 + }, + { + "epoch": 2.473095026078107, + "grad_norm": 3.0144834518432617, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9120023846626282, + "num_tokens": 754516804.0, + "step": 19441 + }, + { + "epoch": 2.4732222363566976, + "grad_norm": 2.920811414718628, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9038325548171997, + "num_tokens": 754558664.0, + "step": 19442 + }, + { + "epoch": 2.473349446635288, + "grad_norm": 3.049614429473877, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9019697904586792, + "num_tokens": 754591666.0, + "step": 19443 + }, + { + "epoch": 2.4734766569138786, + "grad_norm": 2.9204258918762207, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9053955674171448, + "num_tokens": 754628054.0, + "step": 19444 + }, + { + "epoch": 2.473603867192469, + "grad_norm": 3.043189287185669, + "learning_rate": 1e-06, + "loss": 0.298, + "mean_token_accuracy": 0.892084002494812, + "num_tokens": 754669101.0, + "step": 19445 + }, + { + "epoch": 2.4737310774710597, + "grad_norm": 3.0635087490081787, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.8991809487342834, + "num_tokens": 754706656.0, + "step": 19446 + }, + { + "epoch": 2.47385828774965, + "grad_norm": 2.9906156063079834, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9135666489601135, + "num_tokens": 754739409.0, + "step": 19447 + }, + { + "epoch": 2.4739854980282407, + "grad_norm": 2.9645793437957764, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.9094917178153992, + "num_tokens": 754776601.0, + "step": 19448 + }, + { + "epoch": 2.4741127083068313, + "grad_norm": 2.739126205444336, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9067200422286987, + "num_tokens": 754821236.0, + "step": 19449 + }, + { + "epoch": 2.474239918585422, + "grad_norm": 3.033533811569214, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9034101963043213, + "num_tokens": 754855289.0, + "step": 19450 + }, + { + "epoch": 2.4743671288640123, + "grad_norm": 2.8478596210479736, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9063774347305298, + "num_tokens": 754895682.0, + "step": 19451 + }, + { + "epoch": 2.474494339142603, + "grad_norm": 3.0524325370788574, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9019806981086731, + "num_tokens": 754934996.0, + "step": 19452 + }, + { + "epoch": 2.4746215494211934, + "grad_norm": 2.817333459854126, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8973228335380554, + "num_tokens": 754979844.0, + "step": 19453 + }, + { + "epoch": 2.474748759699784, + "grad_norm": 3.160468816757202, + "learning_rate": 1e-06, + "loss": 0.2935, + "mean_token_accuracy": 0.8911706805229187, + "num_tokens": 755014725.0, + "step": 19454 + }, + { + "epoch": 2.4748759699783744, + "grad_norm": 3.0509181022644043, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8971575498580933, + "num_tokens": 755054769.0, + "step": 19455 + }, + { + "epoch": 2.475003180256965, + "grad_norm": 3.1379153728485107, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9048386812210083, + "num_tokens": 755090991.0, + "step": 19456 + }, + { + "epoch": 2.475130390535555, + "grad_norm": 2.756664752960205, + "learning_rate": 1e-06, + "loss": 0.2338, + "mean_token_accuracy": 0.9134795665740967, + "num_tokens": 755130441.0, + "step": 19457 + }, + { + "epoch": 2.475257600814146, + "grad_norm": 2.8116509914398193, + "learning_rate": 1e-06, + "loss": 0.2208, + "mean_token_accuracy": 0.9189469218254089, + "num_tokens": 755166503.0, + "step": 19458 + }, + { + "epoch": 2.475384811092736, + "grad_norm": 2.5479161739349365, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9032472968101501, + "num_tokens": 755213449.0, + "step": 19459 + }, + { + "epoch": 2.4755120213713266, + "grad_norm": 2.394333600997925, + "learning_rate": 1e-06, + "loss": 0.2158, + "mean_token_accuracy": 0.9194977283477783, + "num_tokens": 755261783.0, + "step": 19460 + }, + { + "epoch": 2.475639231649917, + "grad_norm": 2.861283540725708, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9075169563293457, + "num_tokens": 755297114.0, + "step": 19461 + }, + { + "epoch": 2.4757664419285077, + "grad_norm": 2.792398452758789, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.910521388053894, + "num_tokens": 755335644.0, + "step": 19462 + }, + { + "epoch": 2.4758936522070982, + "grad_norm": 2.767343282699585, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9050968885421753, + "num_tokens": 755377410.0, + "step": 19463 + }, + { + "epoch": 2.4760208624856888, + "grad_norm": 3.173924684524536, + "learning_rate": 1e-06, + "loss": 0.2331, + "mean_token_accuracy": 0.9119383096694946, + "num_tokens": 755407687.0, + "step": 19464 + }, + { + "epoch": 2.4761480727642793, + "grad_norm": 2.8333730697631836, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9084450006484985, + "num_tokens": 755443825.0, + "step": 19465 + }, + { + "epoch": 2.47627528304287, + "grad_norm": 3.198662519454956, + "learning_rate": 1e-06, + "loss": 0.2956, + "mean_token_accuracy": 0.8930344581604004, + "num_tokens": 755479108.0, + "step": 19466 + }, + { + "epoch": 2.4764024933214603, + "grad_norm": 3.166700601577759, + "learning_rate": 1e-06, + "loss": 0.3072, + "mean_token_accuracy": 0.8857906460762024, + "num_tokens": 755518047.0, + "step": 19467 + }, + { + "epoch": 2.476529703600051, + "grad_norm": 2.8508172035217285, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9027361869812012, + "num_tokens": 755562323.0, + "step": 19468 + }, + { + "epoch": 2.4766569138786414, + "grad_norm": 2.7402710914611816, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9113620519638062, + "num_tokens": 755599055.0, + "step": 19469 + }, + { + "epoch": 2.476784124157232, + "grad_norm": 3.061789035797119, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9053329229354858, + "num_tokens": 755632504.0, + "step": 19470 + }, + { + "epoch": 2.4769113344358225, + "grad_norm": 2.9873156547546387, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9005938172340393, + "num_tokens": 755670327.0, + "step": 19471 + }, + { + "epoch": 2.477038544714413, + "grad_norm": 3.1469969749450684, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8998972773551941, + "num_tokens": 755706718.0, + "step": 19472 + }, + { + "epoch": 2.4771657549930035, + "grad_norm": 2.8785951137542725, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9045738577842712, + "num_tokens": 755750233.0, + "step": 19473 + }, + { + "epoch": 2.477292965271594, + "grad_norm": 2.975724458694458, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9030318260192871, + "num_tokens": 755784857.0, + "step": 19474 + }, + { + "epoch": 2.4774201755501846, + "grad_norm": 2.6749327182769775, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9070983529090881, + "num_tokens": 755830137.0, + "step": 19475 + }, + { + "epoch": 2.477547385828775, + "grad_norm": 2.6622934341430664, + "learning_rate": 1e-06, + "loss": 0.2442, + "mean_token_accuracy": 0.9098771214485168, + "num_tokens": 755871132.0, + "step": 19476 + }, + { + "epoch": 2.4776745961073656, + "grad_norm": 2.698740005493164, + "learning_rate": 1e-06, + "loss": 0.2435, + "mean_token_accuracy": 0.9108279943466187, + "num_tokens": 755910750.0, + "step": 19477 + }, + { + "epoch": 2.477801806385956, + "grad_norm": 2.86651349067688, + "learning_rate": 1e-06, + "loss": 0.3018, + "mean_token_accuracy": 0.8869670629501343, + "num_tokens": 755954244.0, + "step": 19478 + }, + { + "epoch": 2.4779290166645467, + "grad_norm": 2.823425769805908, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9044691324234009, + "num_tokens": 755994784.0, + "step": 19479 + }, + { + "epoch": 2.4780562269431368, + "grad_norm": 2.8884353637695312, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9039713144302368, + "num_tokens": 756035916.0, + "step": 19480 + }, + { + "epoch": 2.4781834372217277, + "grad_norm": 2.7798900604248047, + "learning_rate": 1e-06, + "loss": 0.2307, + "mean_token_accuracy": 0.9138433933258057, + "num_tokens": 756071848.0, + "step": 19481 + }, + { + "epoch": 2.478310647500318, + "grad_norm": 3.3948497772216797, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9085265398025513, + "num_tokens": 756104833.0, + "step": 19482 + }, + { + "epoch": 2.4784378577789083, + "grad_norm": 2.68717360496521, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9088820219039917, + "num_tokens": 756146216.0, + "step": 19483 + }, + { + "epoch": 2.478565068057499, + "grad_norm": 2.550398826599121, + "learning_rate": 1e-06, + "loss": 0.2133, + "mean_token_accuracy": 0.9182710647583008, + "num_tokens": 756186241.0, + "step": 19484 + }, + { + "epoch": 2.4786922783360894, + "grad_norm": 2.747097969055176, + "learning_rate": 1e-06, + "loss": 0.2251, + "mean_token_accuracy": 0.9155015349388123, + "num_tokens": 756223915.0, + "step": 19485 + }, + { + "epoch": 2.47881948861468, + "grad_norm": 2.762923240661621, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9123796224594116, + "num_tokens": 756264112.0, + "step": 19486 + }, + { + "epoch": 2.4789466988932705, + "grad_norm": 2.8765082359313965, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9010277986526489, + "num_tokens": 756306469.0, + "step": 19487 + }, + { + "epoch": 2.479073909171861, + "grad_norm": 2.9097211360931396, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8899572491645813, + "num_tokens": 756348511.0, + "step": 19488 + }, + { + "epoch": 2.4792011194504515, + "grad_norm": 2.9915215969085693, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9072213172912598, + "num_tokens": 756382613.0, + "step": 19489 + }, + { + "epoch": 2.479328329729042, + "grad_norm": 2.6931567192077637, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9062023162841797, + "num_tokens": 756422782.0, + "step": 19490 + }, + { + "epoch": 2.4794555400076326, + "grad_norm": 2.9777791500091553, + "learning_rate": 1e-06, + "loss": 0.2291, + "mean_token_accuracy": 0.9147650003433228, + "num_tokens": 756457499.0, + "step": 19491 + }, + { + "epoch": 2.479582750286223, + "grad_norm": 3.072253942489624, + "learning_rate": 1e-06, + "loss": 0.3025, + "mean_token_accuracy": 0.8905600309371948, + "num_tokens": 756496096.0, + "step": 19492 + }, + { + "epoch": 2.4797099605648136, + "grad_norm": 2.6283586025238037, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9097284078598022, + "num_tokens": 756536451.0, + "step": 19493 + }, + { + "epoch": 2.479837170843404, + "grad_norm": 3.1632540225982666, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8931761980056763, + "num_tokens": 756572342.0, + "step": 19494 + }, + { + "epoch": 2.4799643811219947, + "grad_norm": 2.8575401306152344, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.902347981929779, + "num_tokens": 756616840.0, + "step": 19495 + }, + { + "epoch": 2.480091591400585, + "grad_norm": 3.021019220352173, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9010441303253174, + "num_tokens": 756653895.0, + "step": 19496 + }, + { + "epoch": 2.4802188016791757, + "grad_norm": 2.9565489292144775, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9075173139572144, + "num_tokens": 756688642.0, + "step": 19497 + }, + { + "epoch": 2.4803460119577663, + "grad_norm": 2.9751107692718506, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9018450975418091, + "num_tokens": 756727338.0, + "step": 19498 + }, + { + "epoch": 2.480473222236357, + "grad_norm": 2.9510598182678223, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9049731492996216, + "num_tokens": 756769039.0, + "step": 19499 + }, + { + "epoch": 2.4806004325149473, + "grad_norm": 3.0764334201812744, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9033041000366211, + "num_tokens": 756804113.0, + "step": 19500 + }, + { + "epoch": 2.480727642793538, + "grad_norm": 2.89296555519104, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9005060195922852, + "num_tokens": 756842953.0, + "step": 19501 + }, + { + "epoch": 2.4808548530721284, + "grad_norm": 3.044140577316284, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9055471420288086, + "num_tokens": 756876200.0, + "step": 19502 + }, + { + "epoch": 2.480982063350719, + "grad_norm": 3.0387938022613525, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8953698873519897, + "num_tokens": 756913134.0, + "step": 19503 + }, + { + "epoch": 2.4811092736293094, + "grad_norm": 2.7664389610290527, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9094241261482239, + "num_tokens": 756952315.0, + "step": 19504 + }, + { + "epoch": 2.4812364839078995, + "grad_norm": 2.9881577491760254, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.908774197101593, + "num_tokens": 756988105.0, + "step": 19505 + }, + { + "epoch": 2.4813636941864905, + "grad_norm": 2.923459768295288, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9038169384002686, + "num_tokens": 757027588.0, + "step": 19506 + }, + { + "epoch": 2.4814909044650806, + "grad_norm": 2.787475347518921, + "learning_rate": 1e-06, + "loss": 0.242, + "mean_token_accuracy": 0.9115194082260132, + "num_tokens": 757066962.0, + "step": 19507 + }, + { + "epoch": 2.481618114743671, + "grad_norm": 2.684047222137451, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9084643125534058, + "num_tokens": 757109990.0, + "step": 19508 + }, + { + "epoch": 2.4817453250222616, + "grad_norm": 2.702683925628662, + "learning_rate": 1e-06, + "loss": 0.2842, + "mean_token_accuracy": 0.8966226577758789, + "num_tokens": 757155571.0, + "step": 19509 + }, + { + "epoch": 2.481872535300852, + "grad_norm": 3.1435317993164062, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8942445516586304, + "num_tokens": 757192763.0, + "step": 19510 + }, + { + "epoch": 2.4819997455794427, + "grad_norm": 3.0448336601257324, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9010642766952515, + "num_tokens": 757228694.0, + "step": 19511 + }, + { + "epoch": 2.4821269558580332, + "grad_norm": 2.94063401222229, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.9019672870635986, + "num_tokens": 757271009.0, + "step": 19512 + }, + { + "epoch": 2.4822541661366238, + "grad_norm": 2.5493907928466797, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9090403318405151, + "num_tokens": 757317313.0, + "step": 19513 + }, + { + "epoch": 2.4823813764152143, + "grad_norm": 2.9922633171081543, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.8995305299758911, + "num_tokens": 757353801.0, + "step": 19514 + }, + { + "epoch": 2.482508586693805, + "grad_norm": 2.640881299972534, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9106442928314209, + "num_tokens": 757399453.0, + "step": 19515 + }, + { + "epoch": 2.4826357969723953, + "grad_norm": 2.8830292224884033, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8972688317298889, + "num_tokens": 757440716.0, + "step": 19516 + }, + { + "epoch": 2.482763007250986, + "grad_norm": 2.8543715476989746, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.9116010069847107, + "num_tokens": 757477811.0, + "step": 19517 + }, + { + "epoch": 2.4828902175295764, + "grad_norm": 3.003006935119629, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9074704051017761, + "num_tokens": 757511571.0, + "step": 19518 + }, + { + "epoch": 2.483017427808167, + "grad_norm": 3.0942323207855225, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8930119276046753, + "num_tokens": 757548533.0, + "step": 19519 + }, + { + "epoch": 2.4831446380867574, + "grad_norm": 2.816807746887207, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8995968103408813, + "num_tokens": 757589205.0, + "step": 19520 + }, + { + "epoch": 2.483271848365348, + "grad_norm": 2.9553983211517334, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9054372310638428, + "num_tokens": 757624263.0, + "step": 19521 + }, + { + "epoch": 2.4833990586439385, + "grad_norm": 2.96047306060791, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8947933912277222, + "num_tokens": 757663712.0, + "step": 19522 + }, + { + "epoch": 2.483526268922529, + "grad_norm": 2.7472400665283203, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9002501964569092, + "num_tokens": 757708877.0, + "step": 19523 + }, + { + "epoch": 2.4836534792011196, + "grad_norm": 3.4546236991882324, + "learning_rate": 1e-06, + "loss": 0.2984, + "mean_token_accuracy": 0.8918516635894775, + "num_tokens": 757741045.0, + "step": 19524 + }, + { + "epoch": 2.48378068947971, + "grad_norm": 2.802269458770752, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9076627492904663, + "num_tokens": 757778382.0, + "step": 19525 + }, + { + "epoch": 2.4839078997583006, + "grad_norm": 3.016131639480591, + "learning_rate": 1e-06, + "loss": 0.3037, + "mean_token_accuracy": 0.8869651556015015, + "num_tokens": 757816091.0, + "step": 19526 + }, + { + "epoch": 2.484035110036891, + "grad_norm": 3.0066580772399902, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9006320834159851, + "num_tokens": 757853296.0, + "step": 19527 + }, + { + "epoch": 2.4841623203154817, + "grad_norm": 2.5519235134124756, + "learning_rate": 1e-06, + "loss": 0.2241, + "mean_token_accuracy": 0.9153578877449036, + "num_tokens": 757894400.0, + "step": 19528 + }, + { + "epoch": 2.484289530594072, + "grad_norm": 2.779376745223999, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9081505537033081, + "num_tokens": 757936415.0, + "step": 19529 + }, + { + "epoch": 2.4844167408726623, + "grad_norm": 2.980754852294922, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9092465043067932, + "num_tokens": 757970960.0, + "step": 19530 + }, + { + "epoch": 2.4845439511512533, + "grad_norm": 2.939894437789917, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9041610956192017, + "num_tokens": 758014691.0, + "step": 19531 + }, + { + "epoch": 2.4846711614298433, + "grad_norm": 2.677738666534424, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9144182205200195, + "num_tokens": 758055171.0, + "step": 19532 + }, + { + "epoch": 2.484798371708434, + "grad_norm": 2.7323005199432373, + "learning_rate": 1e-06, + "loss": 0.2325, + "mean_token_accuracy": 0.9124332666397095, + "num_tokens": 758095193.0, + "step": 19533 + }, + { + "epoch": 2.4849255819870244, + "grad_norm": 2.8432552814483643, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9043745398521423, + "num_tokens": 758139245.0, + "step": 19534 + }, + { + "epoch": 2.485052792265615, + "grad_norm": 3.0592284202575684, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9028975367546082, + "num_tokens": 758176457.0, + "step": 19535 + }, + { + "epoch": 2.4851800025442055, + "grad_norm": 3.0475826263427734, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8958805203437805, + "num_tokens": 758214275.0, + "step": 19536 + }, + { + "epoch": 2.485307212822796, + "grad_norm": 3.0245299339294434, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.903357982635498, + "num_tokens": 758249567.0, + "step": 19537 + }, + { + "epoch": 2.4854344231013865, + "grad_norm": 3.32177734375, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8984652757644653, + "num_tokens": 758285586.0, + "step": 19538 + }, + { + "epoch": 2.485561633379977, + "grad_norm": 3.2956085205078125, + "learning_rate": 1e-06, + "loss": 0.2443, + "mean_token_accuracy": 0.9104081392288208, + "num_tokens": 758318931.0, + "step": 19539 + }, + { + "epoch": 2.4856888436585676, + "grad_norm": 2.8435535430908203, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9077264666557312, + "num_tokens": 758356156.0, + "step": 19540 + }, + { + "epoch": 2.485816053937158, + "grad_norm": 2.8662631511688232, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.8996473550796509, + "num_tokens": 758394088.0, + "step": 19541 + }, + { + "epoch": 2.4859432642157486, + "grad_norm": 2.8119442462921143, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9092766046524048, + "num_tokens": 758433325.0, + "step": 19542 + }, + { + "epoch": 2.486070474494339, + "grad_norm": 3.0610179901123047, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9025225639343262, + "num_tokens": 758471024.0, + "step": 19543 + }, + { + "epoch": 2.4861976847729297, + "grad_norm": 2.9904232025146484, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9058279991149902, + "num_tokens": 758505631.0, + "step": 19544 + }, + { + "epoch": 2.48632489505152, + "grad_norm": 2.8272757530212402, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.8981882333755493, + "num_tokens": 758543654.0, + "step": 19545 + }, + { + "epoch": 2.4864521053301107, + "grad_norm": 2.762362003326416, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9093291759490967, + "num_tokens": 758584817.0, + "step": 19546 + }, + { + "epoch": 2.4865793156087013, + "grad_norm": 2.9187469482421875, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8967603445053101, + "num_tokens": 758625287.0, + "step": 19547 + }, + { + "epoch": 2.486706525887292, + "grad_norm": 2.572572708129883, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9166409969329834, + "num_tokens": 758666648.0, + "step": 19548 + }, + { + "epoch": 2.4868337361658823, + "grad_norm": 2.986610174179077, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9044079780578613, + "num_tokens": 758703199.0, + "step": 19549 + }, + { + "epoch": 2.486960946444473, + "grad_norm": 2.9641711711883545, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9085822105407715, + "num_tokens": 758738442.0, + "step": 19550 + }, + { + "epoch": 2.4870881567230634, + "grad_norm": 2.8422772884368896, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9080805778503418, + "num_tokens": 758779289.0, + "step": 19551 + }, + { + "epoch": 2.487215367001654, + "grad_norm": 3.00594162940979, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8952640891075134, + "num_tokens": 758818954.0, + "step": 19552 + }, + { + "epoch": 2.4873425772802444, + "grad_norm": 2.6255247592926025, + "learning_rate": 1e-06, + "loss": 0.221, + "mean_token_accuracy": 0.9193121194839478, + "num_tokens": 758858305.0, + "step": 19553 + }, + { + "epoch": 2.487469787558835, + "grad_norm": 2.8114328384399414, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9103438854217529, + "num_tokens": 758894450.0, + "step": 19554 + }, + { + "epoch": 2.487596997837425, + "grad_norm": 2.714876413345337, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9030672907829285, + "num_tokens": 758936412.0, + "step": 19555 + }, + { + "epoch": 2.487724208116016, + "grad_norm": 2.8859148025512695, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.901229739189148, + "num_tokens": 758975661.0, + "step": 19556 + }, + { + "epoch": 2.487851418394606, + "grad_norm": 2.899047374725342, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9094827175140381, + "num_tokens": 759013262.0, + "step": 19557 + }, + { + "epoch": 2.4879786286731966, + "grad_norm": 2.7099337577819824, + "learning_rate": 1e-06, + "loss": 0.2292, + "mean_token_accuracy": 0.9137473106384277, + "num_tokens": 759052819.0, + "step": 19558 + }, + { + "epoch": 2.488105838951787, + "grad_norm": 3.112610101699829, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8957290649414062, + "num_tokens": 759090013.0, + "step": 19559 + }, + { + "epoch": 2.4882330492303777, + "grad_norm": 2.6837918758392334, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9079540967941284, + "num_tokens": 759132806.0, + "step": 19560 + }, + { + "epoch": 2.488360259508968, + "grad_norm": 2.6982712745666504, + "learning_rate": 1e-06, + "loss": 0.2216, + "mean_token_accuracy": 0.9163750410079956, + "num_tokens": 759170822.0, + "step": 19561 + }, + { + "epoch": 2.4884874697875587, + "grad_norm": 3.0790624618530273, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9025202989578247, + "num_tokens": 759206900.0, + "step": 19562 + }, + { + "epoch": 2.4886146800661493, + "grad_norm": 2.9088733196258545, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9054914712905884, + "num_tokens": 759245248.0, + "step": 19563 + }, + { + "epoch": 2.48874189034474, + "grad_norm": 2.9020252227783203, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.91408771276474, + "num_tokens": 759281939.0, + "step": 19564 + }, + { + "epoch": 2.4888691006233303, + "grad_norm": 2.9299426078796387, + "learning_rate": 1e-06, + "loss": 0.2275, + "mean_token_accuracy": 0.9124336242675781, + "num_tokens": 759315769.0, + "step": 19565 + }, + { + "epoch": 2.488996310901921, + "grad_norm": 3.0768449306488037, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9071785807609558, + "num_tokens": 759351804.0, + "step": 19566 + }, + { + "epoch": 2.4891235211805114, + "grad_norm": 2.9442028999328613, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8949860334396362, + "num_tokens": 759393201.0, + "step": 19567 + }, + { + "epoch": 2.489250731459102, + "grad_norm": 3.219226837158203, + "learning_rate": 1e-06, + "loss": 0.3158, + "mean_token_accuracy": 0.8867027163505554, + "num_tokens": 759430320.0, + "step": 19568 + }, + { + "epoch": 2.4893779417376924, + "grad_norm": 2.904944896697998, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9008709192276001, + "num_tokens": 759469525.0, + "step": 19569 + }, + { + "epoch": 2.489505152016283, + "grad_norm": 2.862095832824707, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9046194553375244, + "num_tokens": 759509420.0, + "step": 19570 + }, + { + "epoch": 2.4896323622948735, + "grad_norm": 2.814164876937866, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9140976071357727, + "num_tokens": 759549097.0, + "step": 19571 + }, + { + "epoch": 2.489759572573464, + "grad_norm": 3.141608476638794, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9022437930107117, + "num_tokens": 759584015.0, + "step": 19572 + }, + { + "epoch": 2.4898867828520546, + "grad_norm": 3.048741340637207, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8941032290458679, + "num_tokens": 759622350.0, + "step": 19573 + }, + { + "epoch": 2.490013993130645, + "grad_norm": 2.8312721252441406, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.8990054130554199, + "num_tokens": 759661364.0, + "step": 19574 + }, + { + "epoch": 2.4901412034092356, + "grad_norm": 3.0985019207000732, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9072344303131104, + "num_tokens": 759694546.0, + "step": 19575 + }, + { + "epoch": 2.490268413687826, + "grad_norm": 3.025886058807373, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.902427077293396, + "num_tokens": 759733849.0, + "step": 19576 + }, + { + "epoch": 2.4903956239664167, + "grad_norm": 2.9077541828155518, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8997595310211182, + "num_tokens": 759773763.0, + "step": 19577 + }, + { + "epoch": 2.4905228342450068, + "grad_norm": 2.9021124839782715, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9071046113967896, + "num_tokens": 759811107.0, + "step": 19578 + }, + { + "epoch": 2.4906500445235977, + "grad_norm": 2.5620107650756836, + "learning_rate": 1e-06, + "loss": 0.2285, + "mean_token_accuracy": 0.91480553150177, + "num_tokens": 759852959.0, + "step": 19579 + }, + { + "epoch": 2.490777254802188, + "grad_norm": 2.876980781555176, + "learning_rate": 1e-06, + "loss": 0.3045, + "mean_token_accuracy": 0.8880659937858582, + "num_tokens": 759895852.0, + "step": 19580 + }, + { + "epoch": 2.4909044650807783, + "grad_norm": 2.8282151222229004, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9013216495513916, + "num_tokens": 759931019.0, + "step": 19581 + }, + { + "epoch": 2.491031675359369, + "grad_norm": 2.750304698944092, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.8996094465255737, + "num_tokens": 759972041.0, + "step": 19582 + }, + { + "epoch": 2.4911588856379594, + "grad_norm": 2.540527582168579, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.9147560596466064, + "num_tokens": 760015614.0, + "step": 19583 + }, + { + "epoch": 2.49128609591655, + "grad_norm": 2.915081024169922, + "learning_rate": 1e-06, + "loss": 0.3068, + "mean_token_accuracy": 0.8875821232795715, + "num_tokens": 760059740.0, + "step": 19584 + }, + { + "epoch": 2.4914133061951405, + "grad_norm": 2.8824121952056885, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8977503776550293, + "num_tokens": 760104122.0, + "step": 19585 + }, + { + "epoch": 2.491540516473731, + "grad_norm": 2.94205379486084, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8995361328125, + "num_tokens": 760142756.0, + "step": 19586 + }, + { + "epoch": 2.4916677267523215, + "grad_norm": 2.6423161029815674, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.9109153747558594, + "num_tokens": 760188192.0, + "step": 19587 + }, + { + "epoch": 2.491794937030912, + "grad_norm": 2.5971240997314453, + "learning_rate": 1e-06, + "loss": 0.2314, + "mean_token_accuracy": 0.9151451587677002, + "num_tokens": 760231910.0, + "step": 19588 + }, + { + "epoch": 2.4919221473095026, + "grad_norm": 2.875991106033325, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9117628335952759, + "num_tokens": 760268073.0, + "step": 19589 + }, + { + "epoch": 2.492049357588093, + "grad_norm": 2.859927177429199, + "learning_rate": 1e-06, + "loss": 0.2342, + "mean_token_accuracy": 0.9127532243728638, + "num_tokens": 760304908.0, + "step": 19590 + }, + { + "epoch": 2.4921765678666836, + "grad_norm": 2.975292444229126, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8997060060501099, + "num_tokens": 760342476.0, + "step": 19591 + }, + { + "epoch": 2.492303778145274, + "grad_norm": 2.834493398666382, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9089657664299011, + "num_tokens": 760379657.0, + "step": 19592 + }, + { + "epoch": 2.4924309884238647, + "grad_norm": 2.8384387493133545, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9050654768943787, + "num_tokens": 760419917.0, + "step": 19593 + }, + { + "epoch": 2.492558198702455, + "grad_norm": 2.9384772777557373, + "learning_rate": 1e-06, + "loss": 0.2422, + "mean_token_accuracy": 0.9085261821746826, + "num_tokens": 760458038.0, + "step": 19594 + }, + { + "epoch": 2.4926854089810457, + "grad_norm": 3.0430691242218018, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.9133235812187195, + "num_tokens": 760494731.0, + "step": 19595 + }, + { + "epoch": 2.4928126192596363, + "grad_norm": 2.7927591800689697, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9015439748764038, + "num_tokens": 760534769.0, + "step": 19596 + }, + { + "epoch": 2.492939829538227, + "grad_norm": 2.836416721343994, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9099841117858887, + "num_tokens": 760571127.0, + "step": 19597 + }, + { + "epoch": 2.4930670398168173, + "grad_norm": 2.7985029220581055, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9105421304702759, + "num_tokens": 760611446.0, + "step": 19598 + }, + { + "epoch": 2.493194250095408, + "grad_norm": 2.871894121170044, + "learning_rate": 1e-06, + "loss": 0.2275, + "mean_token_accuracy": 0.9145498275756836, + "num_tokens": 760647149.0, + "step": 19599 + }, + { + "epoch": 2.4933214603739984, + "grad_norm": 2.722553014755249, + "learning_rate": 1e-06, + "loss": 0.2307, + "mean_token_accuracy": 0.9134745597839355, + "num_tokens": 760683686.0, + "step": 19600 + }, + { + "epoch": 2.493448670652589, + "grad_norm": 2.7951948642730713, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9094719886779785, + "num_tokens": 760721336.0, + "step": 19601 + }, + { + "epoch": 2.4935758809311794, + "grad_norm": 2.5606367588043213, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9059386849403381, + "num_tokens": 760767587.0, + "step": 19602 + }, + { + "epoch": 2.4937030912097695, + "grad_norm": 3.01530122756958, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9035439491271973, + "num_tokens": 760804509.0, + "step": 19603 + }, + { + "epoch": 2.4938303014883605, + "grad_norm": 2.8899080753326416, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9090645909309387, + "num_tokens": 760843175.0, + "step": 19604 + }, + { + "epoch": 2.4939575117669506, + "grad_norm": 2.8808369636535645, + "learning_rate": 1e-06, + "loss": 0.2376, + "mean_token_accuracy": 0.9099664688110352, + "num_tokens": 760879865.0, + "step": 19605 + }, + { + "epoch": 2.494084722045541, + "grad_norm": 2.8399910926818848, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8997344374656677, + "num_tokens": 760921037.0, + "step": 19606 + }, + { + "epoch": 2.4942119323241316, + "grad_norm": 2.843923568725586, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9019975066184998, + "num_tokens": 760962361.0, + "step": 19607 + }, + { + "epoch": 2.494339142602722, + "grad_norm": 2.7670321464538574, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9068272709846497, + "num_tokens": 761002516.0, + "step": 19608 + }, + { + "epoch": 2.4944663528813127, + "grad_norm": 2.8506641387939453, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8977422714233398, + "num_tokens": 761042785.0, + "step": 19609 + }, + { + "epoch": 2.494593563159903, + "grad_norm": 2.5531392097473145, + "learning_rate": 1e-06, + "loss": 0.2134, + "mean_token_accuracy": 0.9219472408294678, + "num_tokens": 761085200.0, + "step": 19610 + }, + { + "epoch": 2.4947207734384937, + "grad_norm": 2.750239133834839, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9093143939971924, + "num_tokens": 761122701.0, + "step": 19611 + }, + { + "epoch": 2.4948479837170843, + "grad_norm": 3.151118516921997, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9071281552314758, + "num_tokens": 761153477.0, + "step": 19612 + }, + { + "epoch": 2.494975193995675, + "grad_norm": 2.7606070041656494, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9089833498001099, + "num_tokens": 761194353.0, + "step": 19613 + }, + { + "epoch": 2.4951024042742653, + "grad_norm": 3.1154286861419678, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8957380652427673, + "num_tokens": 761231559.0, + "step": 19614 + }, + { + "epoch": 2.495229614552856, + "grad_norm": 2.920907735824585, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9031943678855896, + "num_tokens": 761270706.0, + "step": 19615 + }, + { + "epoch": 2.4953568248314464, + "grad_norm": 2.4932758808135986, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9116233587265015, + "num_tokens": 761315944.0, + "step": 19616 + }, + { + "epoch": 2.495484035110037, + "grad_norm": 2.8751220703125, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9017075300216675, + "num_tokens": 761351464.0, + "step": 19617 + }, + { + "epoch": 2.4956112453886274, + "grad_norm": 3.1051251888275146, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9057095646858215, + "num_tokens": 761386750.0, + "step": 19618 + }, + { + "epoch": 2.495738455667218, + "grad_norm": 2.914381265640259, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.90228670835495, + "num_tokens": 761426682.0, + "step": 19619 + }, + { + "epoch": 2.4958656659458085, + "grad_norm": 2.671849489212036, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9112932682037354, + "num_tokens": 761471503.0, + "step": 19620 + }, + { + "epoch": 2.495992876224399, + "grad_norm": 2.973938465118408, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.908362627029419, + "num_tokens": 761509835.0, + "step": 19621 + }, + { + "epoch": 2.4961200865029896, + "grad_norm": 2.8147568702697754, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9042721390724182, + "num_tokens": 761554186.0, + "step": 19622 + }, + { + "epoch": 2.49624729678158, + "grad_norm": 2.986515998840332, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9036484956741333, + "num_tokens": 761593437.0, + "step": 19623 + }, + { + "epoch": 2.4963745070601706, + "grad_norm": 2.6468002796173096, + "learning_rate": 1e-06, + "loss": 0.2252, + "mean_token_accuracy": 0.9148560166358948, + "num_tokens": 761629040.0, + "step": 19624 + }, + { + "epoch": 2.496501717338761, + "grad_norm": 2.7929205894470215, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.9006938338279724, + "num_tokens": 761668926.0, + "step": 19625 + }, + { + "epoch": 2.4966289276173517, + "grad_norm": 2.686267137527466, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9026182889938354, + "num_tokens": 761711267.0, + "step": 19626 + }, + { + "epoch": 2.496756137895942, + "grad_norm": 2.67775559425354, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9069415926933289, + "num_tokens": 761752667.0, + "step": 19627 + }, + { + "epoch": 2.4968833481745323, + "grad_norm": 2.666193962097168, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9082812666893005, + "num_tokens": 761793374.0, + "step": 19628 + }, + { + "epoch": 2.4970105584531233, + "grad_norm": 2.9660861492156982, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9067515134811401, + "num_tokens": 761828571.0, + "step": 19629 + }, + { + "epoch": 2.4971377687317133, + "grad_norm": 2.794024705886841, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9025790095329285, + "num_tokens": 761866188.0, + "step": 19630 + }, + { + "epoch": 2.497264979010304, + "grad_norm": 2.8736424446105957, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8989160060882568, + "num_tokens": 761905197.0, + "step": 19631 + }, + { + "epoch": 2.4973921892888944, + "grad_norm": 2.782254695892334, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9060094356536865, + "num_tokens": 761945130.0, + "step": 19632 + }, + { + "epoch": 2.497519399567485, + "grad_norm": 2.974093198776245, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.907995879650116, + "num_tokens": 761979737.0, + "step": 19633 + }, + { + "epoch": 2.4976466098460754, + "grad_norm": 2.815297842025757, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9068411588668823, + "num_tokens": 762018240.0, + "step": 19634 + }, + { + "epoch": 2.497773820124666, + "grad_norm": 3.040860652923584, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9044689536094666, + "num_tokens": 762056196.0, + "step": 19635 + }, + { + "epoch": 2.4979010304032565, + "grad_norm": 3.4842171669006348, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9043286442756653, + "num_tokens": 762088897.0, + "step": 19636 + }, + { + "epoch": 2.498028240681847, + "grad_norm": 2.9615142345428467, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9045893549919128, + "num_tokens": 762125726.0, + "step": 19637 + }, + { + "epoch": 2.4981554509604376, + "grad_norm": 2.849600076675415, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8929373025894165, + "num_tokens": 762168676.0, + "step": 19638 + }, + { + "epoch": 2.498282661239028, + "grad_norm": 2.884674072265625, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9085592031478882, + "num_tokens": 762207890.0, + "step": 19639 + }, + { + "epoch": 2.4984098715176186, + "grad_norm": 2.654674530029297, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9101643562316895, + "num_tokens": 762248526.0, + "step": 19640 + }, + { + "epoch": 2.498537081796209, + "grad_norm": 2.735139846801758, + "learning_rate": 1e-06, + "loss": 0.2345, + "mean_token_accuracy": 0.9136427640914917, + "num_tokens": 762288658.0, + "step": 19641 + }, + { + "epoch": 2.4986642920747997, + "grad_norm": 3.0816168785095215, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8982715606689453, + "num_tokens": 762326254.0, + "step": 19642 + }, + { + "epoch": 2.49879150235339, + "grad_norm": 2.965607166290283, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9083095788955688, + "num_tokens": 762359899.0, + "step": 19643 + }, + { + "epoch": 2.4989187126319807, + "grad_norm": 2.9976487159729004, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8976390361785889, + "num_tokens": 762396362.0, + "step": 19644 + }, + { + "epoch": 2.4990459229105713, + "grad_norm": 2.558357000350952, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9099055528640747, + "num_tokens": 762441758.0, + "step": 19645 + }, + { + "epoch": 2.499173133189162, + "grad_norm": 2.8207695484161377, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.8998400568962097, + "num_tokens": 762480338.0, + "step": 19646 + }, + { + "epoch": 2.4993003434677523, + "grad_norm": 2.668409585952759, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9064831137657166, + "num_tokens": 762523134.0, + "step": 19647 + }, + { + "epoch": 2.499427553746343, + "grad_norm": 2.8559467792510986, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9079145193099976, + "num_tokens": 762557503.0, + "step": 19648 + }, + { + "epoch": 2.4995547640249334, + "grad_norm": 3.0069704055786133, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.896640419960022, + "num_tokens": 762595665.0, + "step": 19649 + }, + { + "epoch": 2.499681974303524, + "grad_norm": 2.8078293800354004, + "learning_rate": 1e-06, + "loss": 0.2292, + "mean_token_accuracy": 0.9154295921325684, + "num_tokens": 762634088.0, + "step": 19650 + }, + { + "epoch": 2.499809184582114, + "grad_norm": 2.7183802127838135, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9008898735046387, + "num_tokens": 762678155.0, + "step": 19651 + }, + { + "epoch": 2.499936394860705, + "grad_norm": 2.929290771484375, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9041910171508789, + "num_tokens": 762716013.0, + "step": 19652 + }, + { + "epoch": 2.500063605139295, + "grad_norm": 2.8692071437835693, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.9028477668762207, + "num_tokens": 762752884.0, + "step": 19653 + }, + { + "epoch": 2.500190815417886, + "grad_norm": 2.937232255935669, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8956940770149231, + "num_tokens": 762793223.0, + "step": 19654 + }, + { + "epoch": 2.500318025696476, + "grad_norm": 3.111940383911133, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9053994417190552, + "num_tokens": 762828063.0, + "step": 19655 + }, + { + "epoch": 2.5004452359750666, + "grad_norm": 3.113992214202881, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.902951717376709, + "num_tokens": 762862510.0, + "step": 19656 + }, + { + "epoch": 2.500572446253657, + "grad_norm": 3.029141664505005, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8928437829017639, + "num_tokens": 762902481.0, + "step": 19657 + }, + { + "epoch": 2.5006996565322477, + "grad_norm": 2.9404726028442383, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9072153568267822, + "num_tokens": 762937700.0, + "step": 19658 + }, + { + "epoch": 2.500826866810838, + "grad_norm": 3.194387435913086, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8930848836898804, + "num_tokens": 762975513.0, + "step": 19659 + }, + { + "epoch": 2.5009540770894287, + "grad_norm": 2.951556444168091, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9078762531280518, + "num_tokens": 763011634.0, + "step": 19660 + }, + { + "epoch": 2.5010812873680193, + "grad_norm": 2.808347702026367, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.908086359500885, + "num_tokens": 763049955.0, + "step": 19661 + }, + { + "epoch": 2.50120849764661, + "grad_norm": 2.9389538764953613, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9079671502113342, + "num_tokens": 763089324.0, + "step": 19662 + }, + { + "epoch": 2.5013357079252003, + "grad_norm": 3.5265326499938965, + "learning_rate": 1e-06, + "loss": 0.3148, + "mean_token_accuracy": 0.8875348567962646, + "num_tokens": 763123456.0, + "step": 19663 + }, + { + "epoch": 2.501462918203791, + "grad_norm": 2.8093013763427734, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.9002945423126221, + "num_tokens": 763163784.0, + "step": 19664 + }, + { + "epoch": 2.5015901284823814, + "grad_norm": 2.83782696723938, + "learning_rate": 1e-06, + "loss": 0.2127, + "mean_token_accuracy": 0.9184871912002563, + "num_tokens": 763199710.0, + "step": 19665 + }, + { + "epoch": 2.501717338760972, + "grad_norm": 3.2383065223693848, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9036192297935486, + "num_tokens": 763232780.0, + "step": 19666 + }, + { + "epoch": 2.5018445490395624, + "grad_norm": 2.753631591796875, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9061254262924194, + "num_tokens": 763272639.0, + "step": 19667 + }, + { + "epoch": 2.501971759318153, + "grad_norm": 3.1957623958587646, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9015331268310547, + "num_tokens": 763305966.0, + "step": 19668 + }, + { + "epoch": 2.5020989695967435, + "grad_norm": 2.623763084411621, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9126908779144287, + "num_tokens": 763346818.0, + "step": 19669 + }, + { + "epoch": 2.502226179875334, + "grad_norm": 2.644285202026367, + "learning_rate": 1e-06, + "loss": 0.2166, + "mean_token_accuracy": 0.9187102317810059, + "num_tokens": 763389662.0, + "step": 19670 + }, + { + "epoch": 2.5023533901539246, + "grad_norm": 2.79465389251709, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9039926528930664, + "num_tokens": 763428918.0, + "step": 19671 + }, + { + "epoch": 2.502480600432515, + "grad_norm": 3.22458553314209, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8943367004394531, + "num_tokens": 763463750.0, + "step": 19672 + }, + { + "epoch": 2.5026078107111056, + "grad_norm": 3.563119649887085, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.8967761397361755, + "num_tokens": 763494834.0, + "step": 19673 + }, + { + "epoch": 2.5027350209896957, + "grad_norm": 3.0386364459991455, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9031055569648743, + "num_tokens": 763531648.0, + "step": 19674 + }, + { + "epoch": 2.5028622312682867, + "grad_norm": 2.995875120162964, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9027779698371887, + "num_tokens": 763572040.0, + "step": 19675 + }, + { + "epoch": 2.5029894415468767, + "grad_norm": 5.374127388000488, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9051235914230347, + "num_tokens": 763610631.0, + "step": 19676 + }, + { + "epoch": 2.5031166518254677, + "grad_norm": 3.0986154079437256, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8987072706222534, + "num_tokens": 763646216.0, + "step": 19677 + }, + { + "epoch": 2.503243862104058, + "grad_norm": 2.97385311126709, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9007495641708374, + "num_tokens": 763686086.0, + "step": 19678 + }, + { + "epoch": 2.5033710723826488, + "grad_norm": 2.7545204162597656, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9068547487258911, + "num_tokens": 763727653.0, + "step": 19679 + }, + { + "epoch": 2.503498282661239, + "grad_norm": 2.943148374557495, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9046767950057983, + "num_tokens": 763766333.0, + "step": 19680 + }, + { + "epoch": 2.5036254929398294, + "grad_norm": 2.8375048637390137, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.8995721936225891, + "num_tokens": 763805367.0, + "step": 19681 + }, + { + "epoch": 2.50375270321842, + "grad_norm": 2.9301390647888184, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9012855887413025, + "num_tokens": 763844659.0, + "step": 19682 + }, + { + "epoch": 2.5038799134970104, + "grad_norm": 2.946336269378662, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8939517736434937, + "num_tokens": 763884506.0, + "step": 19683 + }, + { + "epoch": 2.504007123775601, + "grad_norm": 2.786285161972046, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.8977108001708984, + "num_tokens": 763930539.0, + "step": 19684 + }, + { + "epoch": 2.5041343340541915, + "grad_norm": 2.8490893840789795, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9057643413543701, + "num_tokens": 763970737.0, + "step": 19685 + }, + { + "epoch": 2.504261544332782, + "grad_norm": 2.8653552532196045, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9046899080276489, + "num_tokens": 764007958.0, + "step": 19686 + }, + { + "epoch": 2.5043887546113726, + "grad_norm": 2.631270170211792, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.906814455986023, + "num_tokens": 764046859.0, + "step": 19687 + }, + { + "epoch": 2.504515964889963, + "grad_norm": 3.0795130729675293, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9014331102371216, + "num_tokens": 764084548.0, + "step": 19688 + }, + { + "epoch": 2.5046431751685536, + "grad_norm": 2.9082932472229004, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9035753011703491, + "num_tokens": 764123455.0, + "step": 19689 + }, + { + "epoch": 2.504770385447144, + "grad_norm": 2.8878111839294434, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8968607187271118, + "num_tokens": 764164458.0, + "step": 19690 + }, + { + "epoch": 2.5048975957257347, + "grad_norm": 2.896960973739624, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9040299654006958, + "num_tokens": 764199046.0, + "step": 19691 + }, + { + "epoch": 2.505024806004325, + "grad_norm": 2.8208518028259277, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8993114233016968, + "num_tokens": 764241928.0, + "step": 19692 + }, + { + "epoch": 2.5051520162829157, + "grad_norm": 2.703853130340576, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.8974064588546753, + "num_tokens": 764288859.0, + "step": 19693 + }, + { + "epoch": 2.5052792265615063, + "grad_norm": 2.877840280532837, + "learning_rate": 1e-06, + "loss": 0.2338, + "mean_token_accuracy": 0.9119302034378052, + "num_tokens": 764325339.0, + "step": 19694 + }, + { + "epoch": 2.505406436840097, + "grad_norm": 3.0282819271087646, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9029097557067871, + "num_tokens": 764362270.0, + "step": 19695 + }, + { + "epoch": 2.5055336471186873, + "grad_norm": 4.50571870803833, + "learning_rate": 1e-06, + "loss": 0.2203, + "mean_token_accuracy": 0.9160774946212769, + "num_tokens": 764398877.0, + "step": 19696 + }, + { + "epoch": 2.505660857397278, + "grad_norm": 3.0856804847717285, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8993666172027588, + "num_tokens": 764434319.0, + "step": 19697 + }, + { + "epoch": 2.5057880676758684, + "grad_norm": 2.792449951171875, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9080883264541626, + "num_tokens": 764474508.0, + "step": 19698 + }, + { + "epoch": 2.5059152779544585, + "grad_norm": 2.855384588241577, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9018157720565796, + "num_tokens": 764514054.0, + "step": 19699 + }, + { + "epoch": 2.5060424882330494, + "grad_norm": 3.2574362754821777, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9034914970397949, + "num_tokens": 764546175.0, + "step": 19700 + }, + { + "epoch": 2.5061696985116395, + "grad_norm": 2.798495054244995, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9034466743469238, + "num_tokens": 764586668.0, + "step": 19701 + }, + { + "epoch": 2.5062969087902305, + "grad_norm": 2.779928684234619, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9013641476631165, + "num_tokens": 764628248.0, + "step": 19702 + }, + { + "epoch": 2.5064241190688206, + "grad_norm": 2.9695639610290527, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9039602875709534, + "num_tokens": 764668342.0, + "step": 19703 + }, + { + "epoch": 2.5065513293474115, + "grad_norm": 3.3525514602661133, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8835955262184143, + "num_tokens": 764703560.0, + "step": 19704 + }, + { + "epoch": 2.5066785396260016, + "grad_norm": 2.947025775909424, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.8985698223114014, + "num_tokens": 764741158.0, + "step": 19705 + }, + { + "epoch": 2.506805749904592, + "grad_norm": 2.881075143814087, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9117063283920288, + "num_tokens": 764779455.0, + "step": 19706 + }, + { + "epoch": 2.5069329601831827, + "grad_norm": 3.104179859161377, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9010472297668457, + "num_tokens": 764815450.0, + "step": 19707 + }, + { + "epoch": 2.507060170461773, + "grad_norm": 2.6854686737060547, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.9143342971801758, + "num_tokens": 764856163.0, + "step": 19708 + }, + { + "epoch": 2.5071873807403637, + "grad_norm": 3.1589322090148926, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.899850606918335, + "num_tokens": 764892244.0, + "step": 19709 + }, + { + "epoch": 2.5073145910189543, + "grad_norm": 3.071746826171875, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8879073262214661, + "num_tokens": 764933945.0, + "step": 19710 + }, + { + "epoch": 2.507441801297545, + "grad_norm": 2.933795213699341, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9047611355781555, + "num_tokens": 764975960.0, + "step": 19711 + }, + { + "epoch": 2.5075690115761353, + "grad_norm": 3.2126448154449463, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.8997839689254761, + "num_tokens": 765012765.0, + "step": 19712 + }, + { + "epoch": 2.507696221854726, + "grad_norm": 3.254905939102173, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9099003076553345, + "num_tokens": 765046068.0, + "step": 19713 + }, + { + "epoch": 2.5078234321333164, + "grad_norm": 2.875959634780884, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9004519581794739, + "num_tokens": 765089883.0, + "step": 19714 + }, + { + "epoch": 2.507950642411907, + "grad_norm": 3.0070900917053223, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.902028501033783, + "num_tokens": 765129622.0, + "step": 19715 + }, + { + "epoch": 2.5080778526904974, + "grad_norm": 2.7224318981170654, + "learning_rate": 1e-06, + "loss": 0.2349, + "mean_token_accuracy": 0.9131094217300415, + "num_tokens": 765168792.0, + "step": 19716 + }, + { + "epoch": 2.508205062969088, + "grad_norm": 3.0119597911834717, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8961235284805298, + "num_tokens": 765206194.0, + "step": 19717 + }, + { + "epoch": 2.5083322732476785, + "grad_norm": 2.58864426612854, + "learning_rate": 1e-06, + "loss": 0.2233, + "mean_token_accuracy": 0.917809247970581, + "num_tokens": 765247729.0, + "step": 19718 + }, + { + "epoch": 2.508459483526269, + "grad_norm": 3.1601617336273193, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8990321159362793, + "num_tokens": 765283372.0, + "step": 19719 + }, + { + "epoch": 2.5085866938048595, + "grad_norm": 2.9752278327941895, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9073004126548767, + "num_tokens": 765322584.0, + "step": 19720 + }, + { + "epoch": 2.50871390408345, + "grad_norm": 2.8817224502563477, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9088411927223206, + "num_tokens": 765359962.0, + "step": 19721 + }, + { + "epoch": 2.5088411143620406, + "grad_norm": 3.391094446182251, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8906397819519043, + "num_tokens": 765393916.0, + "step": 19722 + }, + { + "epoch": 2.508968324640631, + "grad_norm": 2.81622576713562, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.8997940421104431, + "num_tokens": 765437759.0, + "step": 19723 + }, + { + "epoch": 2.509095534919221, + "grad_norm": 2.8822033405303955, + "learning_rate": 1e-06, + "loss": 0.2158, + "mean_token_accuracy": 0.9190043807029724, + "num_tokens": 765470907.0, + "step": 19724 + }, + { + "epoch": 2.509222745197812, + "grad_norm": 3.170443058013916, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9034237861633301, + "num_tokens": 765511741.0, + "step": 19725 + }, + { + "epoch": 2.5093499554764023, + "grad_norm": 2.549710512161255, + "learning_rate": 1e-06, + "loss": 0.2256, + "mean_token_accuracy": 0.9160703420639038, + "num_tokens": 765554202.0, + "step": 19726 + }, + { + "epoch": 2.5094771657549932, + "grad_norm": 4.242883205413818, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9004351496696472, + "num_tokens": 765594886.0, + "step": 19727 + }, + { + "epoch": 2.5096043760335833, + "grad_norm": 2.873230457305908, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.898114800453186, + "num_tokens": 765635187.0, + "step": 19728 + }, + { + "epoch": 2.5097315863121743, + "grad_norm": 2.839836597442627, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9041991829872131, + "num_tokens": 765677814.0, + "step": 19729 + }, + { + "epoch": 2.5098587965907644, + "grad_norm": 3.042591094970703, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9009203910827637, + "num_tokens": 765713335.0, + "step": 19730 + }, + { + "epoch": 2.509986006869355, + "grad_norm": 2.9469525814056396, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.89753258228302, + "num_tokens": 765752360.0, + "step": 19731 + }, + { + "epoch": 2.5101132171479454, + "grad_norm": 2.804518938064575, + "learning_rate": 1e-06, + "loss": 0.2234, + "mean_token_accuracy": 0.9164007902145386, + "num_tokens": 765786518.0, + "step": 19732 + }, + { + "epoch": 2.510240427426536, + "grad_norm": 3.0757126808166504, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9032144546508789, + "num_tokens": 765823109.0, + "step": 19733 + }, + { + "epoch": 2.5103676377051265, + "grad_norm": 3.0147972106933594, + "learning_rate": 1e-06, + "loss": 0.2904, + "mean_token_accuracy": 0.8920193910598755, + "num_tokens": 765864035.0, + "step": 19734 + }, + { + "epoch": 2.510494847983717, + "grad_norm": 2.838855028152466, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9114910960197449, + "num_tokens": 765903230.0, + "step": 19735 + }, + { + "epoch": 2.5106220582623076, + "grad_norm": 3.0439231395721436, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9107239842414856, + "num_tokens": 765935682.0, + "step": 19736 + }, + { + "epoch": 2.510749268540898, + "grad_norm": 3.07686448097229, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9031136631965637, + "num_tokens": 765972270.0, + "step": 19737 + }, + { + "epoch": 2.5108764788194886, + "grad_norm": 2.7727084159851074, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8990045785903931, + "num_tokens": 766015376.0, + "step": 19738 + }, + { + "epoch": 2.511003689098079, + "grad_norm": 2.7090282440185547, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9135544896125793, + "num_tokens": 766054211.0, + "step": 19739 + }, + { + "epoch": 2.5111308993766697, + "grad_norm": 2.6218342781066895, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9094945192337036, + "num_tokens": 766093148.0, + "step": 19740 + }, + { + "epoch": 2.51125810965526, + "grad_norm": 2.683933973312378, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.9138584733009338, + "num_tokens": 766131820.0, + "step": 19741 + }, + { + "epoch": 2.5113853199338507, + "grad_norm": 2.8796093463897705, + "learning_rate": 1e-06, + "loss": 0.2292, + "mean_token_accuracy": 0.9149055480957031, + "num_tokens": 766168290.0, + "step": 19742 + }, + { + "epoch": 2.5115125302124413, + "grad_norm": 3.1960136890411377, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8852486610412598, + "num_tokens": 766205532.0, + "step": 19743 + }, + { + "epoch": 2.511639740491032, + "grad_norm": 3.2386763095855713, + "learning_rate": 1e-06, + "loss": 0.311, + "mean_token_accuracy": 0.8902767896652222, + "num_tokens": 766244960.0, + "step": 19744 + }, + { + "epoch": 2.5117669507696223, + "grad_norm": 2.8471381664276123, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9030487537384033, + "num_tokens": 766282454.0, + "step": 19745 + }, + { + "epoch": 2.511894161048213, + "grad_norm": 2.9968466758728027, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9135192036628723, + "num_tokens": 766319159.0, + "step": 19746 + }, + { + "epoch": 2.5120213713268034, + "grad_norm": 3.131497859954834, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9047666192054749, + "num_tokens": 766355178.0, + "step": 19747 + }, + { + "epoch": 2.512148581605394, + "grad_norm": 2.7526872158050537, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9107179641723633, + "num_tokens": 766394286.0, + "step": 19748 + }, + { + "epoch": 2.512275791883984, + "grad_norm": 2.804765462875366, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.8977440595626831, + "num_tokens": 766441242.0, + "step": 19749 + }, + { + "epoch": 2.512403002162575, + "grad_norm": 2.814027786254883, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.8978476524353027, + "num_tokens": 766483264.0, + "step": 19750 + }, + { + "epoch": 2.512530212441165, + "grad_norm": 3.208651542663574, + "learning_rate": 1e-06, + "loss": 0.2997, + "mean_token_accuracy": 0.8926002979278564, + "num_tokens": 766518631.0, + "step": 19751 + }, + { + "epoch": 2.512657422719756, + "grad_norm": 2.975130081176758, + "learning_rate": 1e-06, + "loss": 0.2219, + "mean_token_accuracy": 0.9159480333328247, + "num_tokens": 766551900.0, + "step": 19752 + }, + { + "epoch": 2.512784632998346, + "grad_norm": 2.732442617416382, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9059752225875854, + "num_tokens": 766596871.0, + "step": 19753 + }, + { + "epoch": 2.5129118432769366, + "grad_norm": 3.0271079540252686, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.904170036315918, + "num_tokens": 766633612.0, + "step": 19754 + }, + { + "epoch": 2.513039053555527, + "grad_norm": 3.049928665161133, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.895297646522522, + "num_tokens": 766668499.0, + "step": 19755 + }, + { + "epoch": 2.5131662638341177, + "grad_norm": 3.1537530422210693, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9036431312561035, + "num_tokens": 766701483.0, + "step": 19756 + }, + { + "epoch": 2.513293474112708, + "grad_norm": 2.984548568725586, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9001845717430115, + "num_tokens": 766740362.0, + "step": 19757 + }, + { + "epoch": 2.5134206843912987, + "grad_norm": 2.7923998832702637, + "learning_rate": 1e-06, + "loss": 0.2238, + "mean_token_accuracy": 0.9172164797782898, + "num_tokens": 766781970.0, + "step": 19758 + }, + { + "epoch": 2.5135478946698893, + "grad_norm": 2.7380638122558594, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9051333665847778, + "num_tokens": 766828859.0, + "step": 19759 + }, + { + "epoch": 2.51367510494848, + "grad_norm": 2.9675474166870117, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9104676246643066, + "num_tokens": 766863302.0, + "step": 19760 + }, + { + "epoch": 2.5138023152270703, + "grad_norm": 2.7439210414886475, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9078221321105957, + "num_tokens": 766904219.0, + "step": 19761 + }, + { + "epoch": 2.513929525505661, + "grad_norm": 2.995109796524048, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.8978040218353271, + "num_tokens": 766943745.0, + "step": 19762 + }, + { + "epoch": 2.5140567357842514, + "grad_norm": 3.0076727867126465, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8928223848342896, + "num_tokens": 766984601.0, + "step": 19763 + }, + { + "epoch": 2.514183946062842, + "grad_norm": 3.2249341011047363, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9001680016517639, + "num_tokens": 767017944.0, + "step": 19764 + }, + { + "epoch": 2.5143111563414324, + "grad_norm": 2.6424806118011475, + "learning_rate": 1e-06, + "loss": 0.2269, + "mean_token_accuracy": 0.9160064458847046, + "num_tokens": 767056445.0, + "step": 19765 + }, + { + "epoch": 2.514438366620023, + "grad_norm": 2.674224853515625, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9121541976928711, + "num_tokens": 767095579.0, + "step": 19766 + }, + { + "epoch": 2.5145655768986135, + "grad_norm": 3.1973085403442383, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.8992265462875366, + "num_tokens": 767131676.0, + "step": 19767 + }, + { + "epoch": 2.514692787177204, + "grad_norm": 2.74981951713562, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9073988199234009, + "num_tokens": 767176492.0, + "step": 19768 + }, + { + "epoch": 2.5148199974557945, + "grad_norm": 3.1427085399627686, + "learning_rate": 1e-06, + "loss": 0.2445, + "mean_token_accuracy": 0.9097667336463928, + "num_tokens": 767209781.0, + "step": 19769 + }, + { + "epoch": 2.514947207734385, + "grad_norm": 2.71282958984375, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9019166231155396, + "num_tokens": 767254338.0, + "step": 19770 + }, + { + "epoch": 2.5150744180129756, + "grad_norm": 2.9677093029022217, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9076310396194458, + "num_tokens": 767292228.0, + "step": 19771 + }, + { + "epoch": 2.5152016282915657, + "grad_norm": 2.5852439403533936, + "learning_rate": 1e-06, + "loss": 0.2143, + "mean_token_accuracy": 0.9203627705574036, + "num_tokens": 767335710.0, + "step": 19772 + }, + { + "epoch": 2.5153288385701567, + "grad_norm": 2.937438488006592, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8959762454032898, + "num_tokens": 767377836.0, + "step": 19773 + }, + { + "epoch": 2.5154560488487467, + "grad_norm": 3.20455265045166, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8939075469970703, + "num_tokens": 767412604.0, + "step": 19774 + }, + { + "epoch": 2.5155832591273377, + "grad_norm": 2.649440050125122, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9119434356689453, + "num_tokens": 767454887.0, + "step": 19775 + }, + { + "epoch": 2.515710469405928, + "grad_norm": 2.802783727645874, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9049651622772217, + "num_tokens": 767490736.0, + "step": 19776 + }, + { + "epoch": 2.5158376796845188, + "grad_norm": 2.8952577114105225, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.8982087969779968, + "num_tokens": 767531230.0, + "step": 19777 + }, + { + "epoch": 2.515964889963109, + "grad_norm": 2.7443549633026123, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9066288471221924, + "num_tokens": 767571573.0, + "step": 19778 + }, + { + "epoch": 2.5160921002416994, + "grad_norm": 2.8851258754730225, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.906631350517273, + "num_tokens": 767613362.0, + "step": 19779 + }, + { + "epoch": 2.51621931052029, + "grad_norm": 2.9283952713012695, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8942856788635254, + "num_tokens": 767653029.0, + "step": 19780 + }, + { + "epoch": 2.5163465207988804, + "grad_norm": 2.666861057281494, + "learning_rate": 1e-06, + "loss": 0.2296, + "mean_token_accuracy": 0.9153777360916138, + "num_tokens": 767694140.0, + "step": 19781 + }, + { + "epoch": 2.516473731077471, + "grad_norm": 2.786407709121704, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9079267382621765, + "num_tokens": 767735943.0, + "step": 19782 + }, + { + "epoch": 2.5166009413560615, + "grad_norm": 2.9221553802490234, + "learning_rate": 1e-06, + "loss": 0.2308, + "mean_token_accuracy": 0.9119666814804077, + "num_tokens": 767770763.0, + "step": 19783 + }, + { + "epoch": 2.516728151634652, + "grad_norm": 3.1290571689605713, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.899093747138977, + "num_tokens": 767807916.0, + "step": 19784 + }, + { + "epoch": 2.5168553619132426, + "grad_norm": 2.879019260406494, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9044076204299927, + "num_tokens": 767847464.0, + "step": 19785 + }, + { + "epoch": 2.516982572191833, + "grad_norm": 2.638598918914795, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9099221229553223, + "num_tokens": 767888535.0, + "step": 19786 + }, + { + "epoch": 2.5171097824704236, + "grad_norm": 2.722921371459961, + "learning_rate": 1e-06, + "loss": 0.2135, + "mean_token_accuracy": 0.9178300499916077, + "num_tokens": 767923094.0, + "step": 19787 + }, + { + "epoch": 2.517236992749014, + "grad_norm": 3.097900629043579, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9114540219306946, + "num_tokens": 767955174.0, + "step": 19788 + }, + { + "epoch": 2.5173642030276047, + "grad_norm": 3.011089324951172, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8983853459358215, + "num_tokens": 767993350.0, + "step": 19789 + }, + { + "epoch": 2.517491413306195, + "grad_norm": 3.0787904262542725, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9014825820922852, + "num_tokens": 768030022.0, + "step": 19790 + }, + { + "epoch": 2.5176186235847857, + "grad_norm": 2.7909886837005615, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.911886990070343, + "num_tokens": 768071212.0, + "step": 19791 + }, + { + "epoch": 2.5177458338633762, + "grad_norm": 3.029895067214966, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.899223268032074, + "num_tokens": 768106035.0, + "step": 19792 + }, + { + "epoch": 2.5178730441419668, + "grad_norm": 2.59372615814209, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9061962366104126, + "num_tokens": 768153509.0, + "step": 19793 + }, + { + "epoch": 2.5180002544205573, + "grad_norm": 2.68794322013855, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.907251238822937, + "num_tokens": 768200227.0, + "step": 19794 + }, + { + "epoch": 2.518127464699148, + "grad_norm": 3.207484483718872, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9008818864822388, + "num_tokens": 768232937.0, + "step": 19795 + }, + { + "epoch": 2.5182546749777384, + "grad_norm": 2.901750326156616, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9044589400291443, + "num_tokens": 768270582.0, + "step": 19796 + }, + { + "epoch": 2.5183818852563284, + "grad_norm": 2.8927247524261475, + "learning_rate": 1e-06, + "loss": 0.2334, + "mean_token_accuracy": 0.913330614566803, + "num_tokens": 768306593.0, + "step": 19797 + }, + { + "epoch": 2.5185090955349194, + "grad_norm": 2.918992757797241, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8938642144203186, + "num_tokens": 768348021.0, + "step": 19798 + }, + { + "epoch": 2.5186363058135095, + "grad_norm": 2.8577699661254883, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9009506106376648, + "num_tokens": 768388780.0, + "step": 19799 + }, + { + "epoch": 2.5187635160921005, + "grad_norm": 2.894576072692871, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9097349643707275, + "num_tokens": 768423601.0, + "step": 19800 + }, + { + "epoch": 2.5188907263706906, + "grad_norm": 2.769233465194702, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9105309247970581, + "num_tokens": 768461506.0, + "step": 19801 + }, + { + "epoch": 2.5190179366492815, + "grad_norm": 2.901059627532959, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9021225571632385, + "num_tokens": 768498930.0, + "step": 19802 + }, + { + "epoch": 2.5191451469278716, + "grad_norm": 2.6925160884857178, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9088783860206604, + "num_tokens": 768539851.0, + "step": 19803 + }, + { + "epoch": 2.519272357206462, + "grad_norm": 2.947089195251465, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9051054120063782, + "num_tokens": 768575801.0, + "step": 19804 + }, + { + "epoch": 2.5193995674850527, + "grad_norm": 2.613790512084961, + "learning_rate": 1e-06, + "loss": 0.2263, + "mean_token_accuracy": 0.9169884920120239, + "num_tokens": 768617129.0, + "step": 19805 + }, + { + "epoch": 2.519526777763643, + "grad_norm": 2.515716075897217, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9123536348342896, + "num_tokens": 768660229.0, + "step": 19806 + }, + { + "epoch": 2.5196539880422337, + "grad_norm": 2.825345516204834, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.904507577419281, + "num_tokens": 768698042.0, + "step": 19807 + }, + { + "epoch": 2.5197811983208243, + "grad_norm": 2.921661138534546, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8991796374320984, + "num_tokens": 768735306.0, + "step": 19808 + }, + { + "epoch": 2.519908408599415, + "grad_norm": 3.228503465652466, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8985613584518433, + "num_tokens": 768767334.0, + "step": 19809 + }, + { + "epoch": 2.5200356188780053, + "grad_norm": 2.6657495498657227, + "learning_rate": 1e-06, + "loss": 0.2303, + "mean_token_accuracy": 0.9140714406967163, + "num_tokens": 768809104.0, + "step": 19810 + }, + { + "epoch": 2.520162829156596, + "grad_norm": 2.638688087463379, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9045678377151489, + "num_tokens": 768855425.0, + "step": 19811 + }, + { + "epoch": 2.5202900394351864, + "grad_norm": 2.671295642852783, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.903421938419342, + "num_tokens": 768897507.0, + "step": 19812 + }, + { + "epoch": 2.520417249713777, + "grad_norm": 2.986147880554199, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9058592915534973, + "num_tokens": 768932767.0, + "step": 19813 + }, + { + "epoch": 2.5205444599923674, + "grad_norm": 2.969733238220215, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9086827635765076, + "num_tokens": 768968502.0, + "step": 19814 + }, + { + "epoch": 2.520671670270958, + "grad_norm": 3.0169360637664795, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9079880714416504, + "num_tokens": 769005294.0, + "step": 19815 + }, + { + "epoch": 2.5207988805495485, + "grad_norm": 3.180884838104248, + "learning_rate": 1e-06, + "loss": 0.315, + "mean_token_accuracy": 0.8851730823516846, + "num_tokens": 769044932.0, + "step": 19816 + }, + { + "epoch": 2.520926090828139, + "grad_norm": 2.8461461067199707, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8970203399658203, + "num_tokens": 769086366.0, + "step": 19817 + }, + { + "epoch": 2.5210533011067295, + "grad_norm": 2.9166362285614014, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8990716338157654, + "num_tokens": 769127997.0, + "step": 19818 + }, + { + "epoch": 2.52118051138532, + "grad_norm": 3.158872604370117, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8952800035476685, + "num_tokens": 769166015.0, + "step": 19819 + }, + { + "epoch": 2.5213077216639106, + "grad_norm": 2.9781289100646973, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8970402479171753, + "num_tokens": 769205311.0, + "step": 19820 + }, + { + "epoch": 2.521434931942501, + "grad_norm": 2.98301100730896, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.9004746079444885, + "num_tokens": 769240363.0, + "step": 19821 + }, + { + "epoch": 2.521562142221091, + "grad_norm": 2.7252674102783203, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.89638352394104, + "num_tokens": 769285451.0, + "step": 19822 + }, + { + "epoch": 2.521689352499682, + "grad_norm": 2.951504945755005, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9030841588973999, + "num_tokens": 769325254.0, + "step": 19823 + }, + { + "epoch": 2.5218165627782723, + "grad_norm": 2.904956579208374, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9069417119026184, + "num_tokens": 769367113.0, + "step": 19824 + }, + { + "epoch": 2.5219437730568632, + "grad_norm": 3.121830940246582, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.8976845741271973, + "num_tokens": 769404267.0, + "step": 19825 + }, + { + "epoch": 2.5220709833354533, + "grad_norm": 2.6877973079681396, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9078562259674072, + "num_tokens": 769447295.0, + "step": 19826 + }, + { + "epoch": 2.522198193614044, + "grad_norm": 2.712026834487915, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.8997511267662048, + "num_tokens": 769491834.0, + "step": 19827 + }, + { + "epoch": 2.5223254038926344, + "grad_norm": 3.075094699859619, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.890357494354248, + "num_tokens": 769535351.0, + "step": 19828 + }, + { + "epoch": 2.522452614171225, + "grad_norm": 2.773355722427368, + "learning_rate": 1e-06, + "loss": 0.2263, + "mean_token_accuracy": 0.9159206748008728, + "num_tokens": 769573916.0, + "step": 19829 + }, + { + "epoch": 2.5225798244498154, + "grad_norm": 2.789699077606201, + "learning_rate": 1e-06, + "loss": 0.2151, + "mean_token_accuracy": 0.9228472113609314, + "num_tokens": 769614973.0, + "step": 19830 + }, + { + "epoch": 2.522707034728406, + "grad_norm": 2.962242841720581, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9054244756698608, + "num_tokens": 769651255.0, + "step": 19831 + }, + { + "epoch": 2.5228342450069965, + "grad_norm": 2.536989688873291, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.9075330495834351, + "num_tokens": 769695478.0, + "step": 19832 + }, + { + "epoch": 2.522961455285587, + "grad_norm": 2.8692784309387207, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9062658548355103, + "num_tokens": 769730088.0, + "step": 19833 + }, + { + "epoch": 2.5230886655641775, + "grad_norm": 2.842660903930664, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9111461639404297, + "num_tokens": 769769535.0, + "step": 19834 + }, + { + "epoch": 2.523215875842768, + "grad_norm": 2.7027313709259033, + "learning_rate": 1e-06, + "loss": 0.2369, + "mean_token_accuracy": 0.9105466604232788, + "num_tokens": 769808604.0, + "step": 19835 + }, + { + "epoch": 2.5233430861213586, + "grad_norm": 2.9699113368988037, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.901309072971344, + "num_tokens": 769847885.0, + "step": 19836 + }, + { + "epoch": 2.523470296399949, + "grad_norm": 2.892390727996826, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9016444683074951, + "num_tokens": 769887719.0, + "step": 19837 + }, + { + "epoch": 2.5235975066785397, + "grad_norm": 2.8521714210510254, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.9136090874671936, + "num_tokens": 769924804.0, + "step": 19838 + }, + { + "epoch": 2.52372471695713, + "grad_norm": 2.8250832557678223, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9107406139373779, + "num_tokens": 769961957.0, + "step": 19839 + }, + { + "epoch": 2.5238519272357207, + "grad_norm": 2.999223232269287, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9075474739074707, + "num_tokens": 769998772.0, + "step": 19840 + }, + { + "epoch": 2.5239791375143112, + "grad_norm": 3.3675684928894043, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.8871123790740967, + "num_tokens": 770031788.0, + "step": 19841 + }, + { + "epoch": 2.5241063477929018, + "grad_norm": 2.913194417953491, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8937193751335144, + "num_tokens": 770073916.0, + "step": 19842 + }, + { + "epoch": 2.5242335580714923, + "grad_norm": 2.71993088722229, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9038166999816895, + "num_tokens": 770112984.0, + "step": 19843 + }, + { + "epoch": 2.524360768350083, + "grad_norm": 2.916012763977051, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.903456449508667, + "num_tokens": 770151702.0, + "step": 19844 + }, + { + "epoch": 2.5244879786286734, + "grad_norm": 2.690704345703125, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8987629413604736, + "num_tokens": 770200579.0, + "step": 19845 + }, + { + "epoch": 2.524615188907264, + "grad_norm": 2.9520821571350098, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9006794691085815, + "num_tokens": 770239072.0, + "step": 19846 + }, + { + "epoch": 2.524742399185854, + "grad_norm": 2.6746959686279297, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9133528470993042, + "num_tokens": 770277822.0, + "step": 19847 + }, + { + "epoch": 2.524869609464445, + "grad_norm": 3.008080244064331, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8911329507827759, + "num_tokens": 770320523.0, + "step": 19848 + }, + { + "epoch": 2.524996819743035, + "grad_norm": 2.843832015991211, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8993183374404907, + "num_tokens": 770364802.0, + "step": 19849 + }, + { + "epoch": 2.525124030021626, + "grad_norm": 2.6574788093566895, + "learning_rate": 1e-06, + "loss": 0.2158, + "mean_token_accuracy": 0.9184200167655945, + "num_tokens": 770404412.0, + "step": 19850 + }, + { + "epoch": 2.525251240300216, + "grad_norm": 2.954249143600464, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9036169648170471, + "num_tokens": 770443255.0, + "step": 19851 + }, + { + "epoch": 2.5253784505788066, + "grad_norm": 2.803938388824463, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9075930714607239, + "num_tokens": 770481495.0, + "step": 19852 + }, + { + "epoch": 2.525505660857397, + "grad_norm": 2.9368128776550293, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9020906090736389, + "num_tokens": 770519033.0, + "step": 19853 + }, + { + "epoch": 2.5256328711359877, + "grad_norm": 3.0619864463806152, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9041933417320251, + "num_tokens": 770554366.0, + "step": 19854 + }, + { + "epoch": 2.525760081414578, + "grad_norm": 3.188615560531616, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.896014928817749, + "num_tokens": 770590778.0, + "step": 19855 + }, + { + "epoch": 2.5258872916931687, + "grad_norm": 3.1467504501342773, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.8984689712524414, + "num_tokens": 770623964.0, + "step": 19856 + }, + { + "epoch": 2.5260145019717593, + "grad_norm": 2.786440134048462, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9112218022346497, + "num_tokens": 770661521.0, + "step": 19857 + }, + { + "epoch": 2.52614171225035, + "grad_norm": 2.859387159347534, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9097269177436829, + "num_tokens": 770698840.0, + "step": 19858 + }, + { + "epoch": 2.5262689225289403, + "grad_norm": 2.7700741291046143, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.904384970664978, + "num_tokens": 770738251.0, + "step": 19859 + }, + { + "epoch": 2.526396132807531, + "grad_norm": 2.844848871231079, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9008229374885559, + "num_tokens": 770780551.0, + "step": 19860 + }, + { + "epoch": 2.5265233430861214, + "grad_norm": 3.2996246814727783, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8904416561126709, + "num_tokens": 770814381.0, + "step": 19861 + }, + { + "epoch": 2.526650553364712, + "grad_norm": 2.8175292015075684, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9092541933059692, + "num_tokens": 770856122.0, + "step": 19862 + }, + { + "epoch": 2.5267777636433024, + "grad_norm": 2.921248197555542, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8988968133926392, + "num_tokens": 770898558.0, + "step": 19863 + }, + { + "epoch": 2.526904973921893, + "grad_norm": 2.6583285331726074, + "learning_rate": 1e-06, + "loss": 0.2249, + "mean_token_accuracy": 0.9177489280700684, + "num_tokens": 770937636.0, + "step": 19864 + }, + { + "epoch": 2.5270321842004835, + "grad_norm": 3.0307071208953857, + "learning_rate": 1e-06, + "loss": 0.233, + "mean_token_accuracy": 0.9115583896636963, + "num_tokens": 770969491.0, + "step": 19865 + }, + { + "epoch": 2.527159394479074, + "grad_norm": 3.051718235015869, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.8976255059242249, + "num_tokens": 771007640.0, + "step": 19866 + }, + { + "epoch": 2.5272866047576645, + "grad_norm": 3.000473976135254, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.907421350479126, + "num_tokens": 771041748.0, + "step": 19867 + }, + { + "epoch": 2.527413815036255, + "grad_norm": 3.351142406463623, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8940596580505371, + "num_tokens": 771077300.0, + "step": 19868 + }, + { + "epoch": 2.5275410253148456, + "grad_norm": 2.743464231491089, + "learning_rate": 1e-06, + "loss": 0.2254, + "mean_token_accuracy": 0.9173486232757568, + "num_tokens": 771115315.0, + "step": 19869 + }, + { + "epoch": 2.5276682355934357, + "grad_norm": 2.8738362789154053, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9078861474990845, + "num_tokens": 771154700.0, + "step": 19870 + }, + { + "epoch": 2.5277954458720266, + "grad_norm": 2.870821475982666, + "learning_rate": 1e-06, + "loss": 0.2394, + "mean_token_accuracy": 0.9080623388290405, + "num_tokens": 771191782.0, + "step": 19871 + }, + { + "epoch": 2.5279226561506167, + "grad_norm": 2.718231678009033, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9068930149078369, + "num_tokens": 771232010.0, + "step": 19872 + }, + { + "epoch": 2.5280498664292077, + "grad_norm": 3.221430778503418, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.895948588848114, + "num_tokens": 771267706.0, + "step": 19873 + }, + { + "epoch": 2.528177076707798, + "grad_norm": 2.755922317504883, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9063587188720703, + "num_tokens": 771306902.0, + "step": 19874 + }, + { + "epoch": 2.5283042869863888, + "grad_norm": 3.131596088409424, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8971781730651855, + "num_tokens": 771347445.0, + "step": 19875 + }, + { + "epoch": 2.528431497264979, + "grad_norm": 3.1228067874908447, + "learning_rate": 1e-06, + "loss": 0.2272, + "mean_token_accuracy": 0.9152100086212158, + "num_tokens": 771380476.0, + "step": 19876 + }, + { + "epoch": 2.5285587075435694, + "grad_norm": 3.2733283042907715, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9041092395782471, + "num_tokens": 771414102.0, + "step": 19877 + }, + { + "epoch": 2.52868591782216, + "grad_norm": 2.919297218322754, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9130334258079529, + "num_tokens": 771447753.0, + "step": 19878 + }, + { + "epoch": 2.5288131281007504, + "grad_norm": 3.091980457305908, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9112845659255981, + "num_tokens": 771478356.0, + "step": 19879 + }, + { + "epoch": 2.528940338379341, + "grad_norm": 3.03149151802063, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9081450700759888, + "num_tokens": 771512903.0, + "step": 19880 + }, + { + "epoch": 2.5290675486579315, + "grad_norm": 2.816692590713501, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9073041677474976, + "num_tokens": 771549657.0, + "step": 19881 + }, + { + "epoch": 2.529194758936522, + "grad_norm": 2.9727752208709717, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8949722647666931, + "num_tokens": 771589177.0, + "step": 19882 + }, + { + "epoch": 2.5293219692151125, + "grad_norm": 2.9302847385406494, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9029998183250427, + "num_tokens": 771630276.0, + "step": 19883 + }, + { + "epoch": 2.529449179493703, + "grad_norm": 2.94844651222229, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9066535234451294, + "num_tokens": 771664381.0, + "step": 19884 + }, + { + "epoch": 2.5295763897722936, + "grad_norm": 3.0376484394073486, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9056398868560791, + "num_tokens": 771700210.0, + "step": 19885 + }, + { + "epoch": 2.529703600050884, + "grad_norm": 2.8644261360168457, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9098218679428101, + "num_tokens": 771735620.0, + "step": 19886 + }, + { + "epoch": 2.5298308103294747, + "grad_norm": 2.762096643447876, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9016841650009155, + "num_tokens": 771780960.0, + "step": 19887 + }, + { + "epoch": 2.529958020608065, + "grad_norm": 2.879322052001953, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.9109675884246826, + "num_tokens": 771819968.0, + "step": 19888 + }, + { + "epoch": 2.5300852308866557, + "grad_norm": 3.0453054904937744, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9084604978561401, + "num_tokens": 771854877.0, + "step": 19889 + }, + { + "epoch": 2.5302124411652462, + "grad_norm": 2.947408676147461, + "learning_rate": 1e-06, + "loss": 0.2394, + "mean_token_accuracy": 0.9110866189002991, + "num_tokens": 771886914.0, + "step": 19890 + }, + { + "epoch": 2.5303396514438368, + "grad_norm": 2.9833080768585205, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9004295468330383, + "num_tokens": 771927539.0, + "step": 19891 + }, + { + "epoch": 2.5304668617224273, + "grad_norm": 3.2670695781707764, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9000442028045654, + "num_tokens": 771964853.0, + "step": 19892 + }, + { + "epoch": 2.530594072001018, + "grad_norm": 2.5214695930480957, + "learning_rate": 1e-06, + "loss": 0.2251, + "mean_token_accuracy": 0.914561927318573, + "num_tokens": 772011942.0, + "step": 19893 + }, + { + "epoch": 2.5307212822796084, + "grad_norm": 2.995638132095337, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.91037517786026, + "num_tokens": 772046931.0, + "step": 19894 + }, + { + "epoch": 2.5308484925581984, + "grad_norm": 3.150380849838257, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8913949131965637, + "num_tokens": 772082572.0, + "step": 19895 + }, + { + "epoch": 2.5309757028367894, + "grad_norm": 3.120777130126953, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9029566049575806, + "num_tokens": 772117209.0, + "step": 19896 + }, + { + "epoch": 2.5311029131153795, + "grad_norm": 2.624823808670044, + "learning_rate": 1e-06, + "loss": 0.2242, + "mean_token_accuracy": 0.9150300025939941, + "num_tokens": 772157247.0, + "step": 19897 + }, + { + "epoch": 2.5312301233939705, + "grad_norm": 2.6444144248962402, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9107644557952881, + "num_tokens": 772200147.0, + "step": 19898 + }, + { + "epoch": 2.5313573336725606, + "grad_norm": 2.9072344303131104, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8944013118743896, + "num_tokens": 772241016.0, + "step": 19899 + }, + { + "epoch": 2.5314845439511515, + "grad_norm": 2.7782013416290283, + "learning_rate": 1e-06, + "loss": 0.2392, + "mean_token_accuracy": 0.9120617508888245, + "num_tokens": 772279737.0, + "step": 19900 + }, + { + "epoch": 2.5316117542297416, + "grad_norm": 2.926276683807373, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9053152799606323, + "num_tokens": 772316640.0, + "step": 19901 + }, + { + "epoch": 2.531738964508332, + "grad_norm": 2.7702443599700928, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9030630588531494, + "num_tokens": 772357258.0, + "step": 19902 + }, + { + "epoch": 2.5318661747869227, + "grad_norm": 2.9647204875946045, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9001833200454712, + "num_tokens": 772399577.0, + "step": 19903 + }, + { + "epoch": 2.531993385065513, + "grad_norm": 3.4518959522247314, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8930652737617493, + "num_tokens": 772434458.0, + "step": 19904 + }, + { + "epoch": 2.5321205953441037, + "grad_norm": 3.1086106300354004, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8949829339981079, + "num_tokens": 772472407.0, + "step": 19905 + }, + { + "epoch": 2.5322478056226942, + "grad_norm": 2.7167723178863525, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.8999566435813904, + "num_tokens": 772519031.0, + "step": 19906 + }, + { + "epoch": 2.5323750159012848, + "grad_norm": 2.760204315185547, + "learning_rate": 1e-06, + "loss": 0.2301, + "mean_token_accuracy": 0.9147393107414246, + "num_tokens": 772552487.0, + "step": 19907 + }, + { + "epoch": 2.5325022261798753, + "grad_norm": 2.6965227127075195, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9096658229827881, + "num_tokens": 772593055.0, + "step": 19908 + }, + { + "epoch": 2.532629436458466, + "grad_norm": 2.7741386890411377, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9144469499588013, + "num_tokens": 772632058.0, + "step": 19909 + }, + { + "epoch": 2.5327566467370564, + "grad_norm": 2.7322206497192383, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.909576416015625, + "num_tokens": 772675326.0, + "step": 19910 + }, + { + "epoch": 2.532883857015647, + "grad_norm": 2.6799161434173584, + "learning_rate": 1e-06, + "loss": 0.2339, + "mean_token_accuracy": 0.9126210808753967, + "num_tokens": 772716141.0, + "step": 19911 + }, + { + "epoch": 2.5330110672942374, + "grad_norm": 2.8010494709014893, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8947114944458008, + "num_tokens": 772757212.0, + "step": 19912 + }, + { + "epoch": 2.533138277572828, + "grad_norm": 2.482055902481079, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9080947041511536, + "num_tokens": 772806374.0, + "step": 19913 + }, + { + "epoch": 2.5332654878514185, + "grad_norm": 2.7735118865966797, + "learning_rate": 1e-06, + "loss": 0.2209, + "mean_token_accuracy": 0.9157261848449707, + "num_tokens": 772841949.0, + "step": 19914 + }, + { + "epoch": 2.533392698130009, + "grad_norm": 2.7258522510528564, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8952345848083496, + "num_tokens": 772887353.0, + "step": 19915 + }, + { + "epoch": 2.5335199084085995, + "grad_norm": 2.9232213497161865, + "learning_rate": 1e-06, + "loss": 0.2656, + "mean_token_accuracy": 0.9026045799255371, + "num_tokens": 772925106.0, + "step": 19916 + }, + { + "epoch": 2.53364711868719, + "grad_norm": 2.7430622577667236, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9090083837509155, + "num_tokens": 772966668.0, + "step": 19917 + }, + { + "epoch": 2.5337743289657806, + "grad_norm": 3.5335116386413574, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.902247428894043, + "num_tokens": 772999643.0, + "step": 19918 + }, + { + "epoch": 2.533901539244371, + "grad_norm": 2.8055598735809326, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9110892415046692, + "num_tokens": 773037107.0, + "step": 19919 + }, + { + "epoch": 2.534028749522961, + "grad_norm": 3.254206418991089, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8953349590301514, + "num_tokens": 773075054.0, + "step": 19920 + }, + { + "epoch": 2.534155959801552, + "grad_norm": 3.061372756958008, + "learning_rate": 1e-06, + "loss": 0.2164, + "mean_token_accuracy": 0.9174016118049622, + "num_tokens": 773105543.0, + "step": 19921 + }, + { + "epoch": 2.5342831700801423, + "grad_norm": 2.7315514087677, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8980499505996704, + "num_tokens": 773146313.0, + "step": 19922 + }, + { + "epoch": 2.5344103803587332, + "grad_norm": 3.072570562362671, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.89864182472229, + "num_tokens": 773185429.0, + "step": 19923 + }, + { + "epoch": 2.5345375906373233, + "grad_norm": 3.1129205226898193, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9015123248100281, + "num_tokens": 773221518.0, + "step": 19924 + }, + { + "epoch": 2.534664800915914, + "grad_norm": 2.841531276702881, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9095792770385742, + "num_tokens": 773258311.0, + "step": 19925 + }, + { + "epoch": 2.5347920111945044, + "grad_norm": 2.9451565742492676, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8929799795150757, + "num_tokens": 773302786.0, + "step": 19926 + }, + { + "epoch": 2.534919221473095, + "grad_norm": 2.691162347793579, + "learning_rate": 1e-06, + "loss": 0.225, + "mean_token_accuracy": 0.9153913259506226, + "num_tokens": 773340636.0, + "step": 19927 + }, + { + "epoch": 2.5350464317516854, + "grad_norm": 2.666539430618286, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9077706336975098, + "num_tokens": 773384119.0, + "step": 19928 + }, + { + "epoch": 2.535173642030276, + "grad_norm": 2.932239532470703, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9121768474578857, + "num_tokens": 773422297.0, + "step": 19929 + }, + { + "epoch": 2.5353008523088665, + "grad_norm": 2.630141258239746, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9052836894989014, + "num_tokens": 773468313.0, + "step": 19930 + }, + { + "epoch": 2.535428062587457, + "grad_norm": 3.004528045654297, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8966259956359863, + "num_tokens": 773507789.0, + "step": 19931 + }, + { + "epoch": 2.5355552728660475, + "grad_norm": 2.875152587890625, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.8996546268463135, + "num_tokens": 773550873.0, + "step": 19932 + }, + { + "epoch": 2.535682483144638, + "grad_norm": 2.895217180252075, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.890900731086731, + "num_tokens": 773595391.0, + "step": 19933 + }, + { + "epoch": 2.5358096934232286, + "grad_norm": 2.939007520675659, + "learning_rate": 1e-06, + "loss": 0.2323, + "mean_token_accuracy": 0.9123518466949463, + "num_tokens": 773629458.0, + "step": 19934 + }, + { + "epoch": 2.535936903701819, + "grad_norm": 3.093515396118164, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9004728198051453, + "num_tokens": 773662065.0, + "step": 19935 + }, + { + "epoch": 2.5360641139804097, + "grad_norm": 2.9449310302734375, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9048203229904175, + "num_tokens": 773698675.0, + "step": 19936 + }, + { + "epoch": 2.536191324259, + "grad_norm": 3.0125696659088135, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9100266695022583, + "num_tokens": 773730718.0, + "step": 19937 + }, + { + "epoch": 2.5363185345375907, + "grad_norm": 2.978010654449463, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9002615213394165, + "num_tokens": 773767921.0, + "step": 19938 + }, + { + "epoch": 2.5364457448161812, + "grad_norm": 2.7764110565185547, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9108821153640747, + "num_tokens": 773803640.0, + "step": 19939 + }, + { + "epoch": 2.5365729550947718, + "grad_norm": 2.948232650756836, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.8999000787734985, + "num_tokens": 773841571.0, + "step": 19940 + }, + { + "epoch": 2.5367001653733623, + "grad_norm": 2.5765068531036377, + "learning_rate": 1e-06, + "loss": 0.231, + "mean_token_accuracy": 0.91371750831604, + "num_tokens": 773883882.0, + "step": 19941 + }, + { + "epoch": 2.536827375651953, + "grad_norm": 2.8803441524505615, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9029339551925659, + "num_tokens": 773923051.0, + "step": 19942 + }, + { + "epoch": 2.5369545859305433, + "grad_norm": 2.4773197174072266, + "learning_rate": 1e-06, + "loss": 0.2301, + "mean_token_accuracy": 0.9148862361907959, + "num_tokens": 773966459.0, + "step": 19943 + }, + { + "epoch": 2.537081796209134, + "grad_norm": 2.736203193664551, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9054872989654541, + "num_tokens": 774004622.0, + "step": 19944 + }, + { + "epoch": 2.537209006487724, + "grad_norm": 2.756436824798584, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.9088699817657471, + "num_tokens": 774041979.0, + "step": 19945 + }, + { + "epoch": 2.537336216766315, + "grad_norm": 2.6302037239074707, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9077314138412476, + "num_tokens": 774084317.0, + "step": 19946 + }, + { + "epoch": 2.537463427044905, + "grad_norm": 2.5816149711608887, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.913029670715332, + "num_tokens": 774128589.0, + "step": 19947 + }, + { + "epoch": 2.537590637323496, + "grad_norm": 3.054813861846924, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9005111455917358, + "num_tokens": 774164509.0, + "step": 19948 + }, + { + "epoch": 2.537717847602086, + "grad_norm": 2.826634645462036, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9035739898681641, + "num_tokens": 774203601.0, + "step": 19949 + }, + { + "epoch": 2.5378450578806766, + "grad_norm": 2.868995428085327, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9015411138534546, + "num_tokens": 774241724.0, + "step": 19950 + }, + { + "epoch": 2.537972268159267, + "grad_norm": 3.0856356620788574, + "learning_rate": 1e-06, + "loss": 0.2913, + "mean_token_accuracy": 0.8934367299079895, + "num_tokens": 774280982.0, + "step": 19951 + }, + { + "epoch": 2.5380994784378577, + "grad_norm": 2.8236382007598877, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9047760963439941, + "num_tokens": 774321811.0, + "step": 19952 + }, + { + "epoch": 2.538226688716448, + "grad_norm": 2.8180782794952393, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9111030101776123, + "num_tokens": 774358940.0, + "step": 19953 + }, + { + "epoch": 2.5383538989950387, + "grad_norm": 3.233704090118408, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9004818797111511, + "num_tokens": 774391834.0, + "step": 19954 + }, + { + "epoch": 2.5384811092736292, + "grad_norm": 3.3436508178710938, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8922625780105591, + "num_tokens": 774426579.0, + "step": 19955 + }, + { + "epoch": 2.5386083195522198, + "grad_norm": 2.8381950855255127, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9042390584945679, + "num_tokens": 774464985.0, + "step": 19956 + }, + { + "epoch": 2.5387355298308103, + "grad_norm": 3.0348851680755615, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9037227630615234, + "num_tokens": 774502269.0, + "step": 19957 + }, + { + "epoch": 2.538862740109401, + "grad_norm": 3.105747938156128, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8980624079704285, + "num_tokens": 774540263.0, + "step": 19958 + }, + { + "epoch": 2.5389899503879914, + "grad_norm": 3.095031976699829, + "learning_rate": 1e-06, + "loss": 0.2241, + "mean_token_accuracy": 0.9150426387786865, + "num_tokens": 774578009.0, + "step": 19959 + }, + { + "epoch": 2.539117160666582, + "grad_norm": 3.0823702812194824, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.904139518737793, + "num_tokens": 774613590.0, + "step": 19960 + }, + { + "epoch": 2.5392443709451724, + "grad_norm": 2.8246123790740967, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9046578407287598, + "num_tokens": 774654860.0, + "step": 19961 + }, + { + "epoch": 2.539371581223763, + "grad_norm": 2.911829948425293, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9060027599334717, + "num_tokens": 774696498.0, + "step": 19962 + }, + { + "epoch": 2.5394987915023535, + "grad_norm": 2.7070353031158447, + "learning_rate": 1e-06, + "loss": 0.2314, + "mean_token_accuracy": 0.913468599319458, + "num_tokens": 774736319.0, + "step": 19963 + }, + { + "epoch": 2.539626001780944, + "grad_norm": 2.943833589553833, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9069118499755859, + "num_tokens": 774773730.0, + "step": 19964 + }, + { + "epoch": 2.5397532120595345, + "grad_norm": 3.1058051586151123, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8917702436447144, + "num_tokens": 774808713.0, + "step": 19965 + }, + { + "epoch": 2.539880422338125, + "grad_norm": 2.7738380432128906, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.9102997779846191, + "num_tokens": 774844440.0, + "step": 19966 + }, + { + "epoch": 2.5400076326167156, + "grad_norm": 3.0332467555999756, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9062719941139221, + "num_tokens": 774878730.0, + "step": 19967 + }, + { + "epoch": 2.5401348428953057, + "grad_norm": 2.8147425651550293, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9068237543106079, + "num_tokens": 774919641.0, + "step": 19968 + }, + { + "epoch": 2.5402620531738966, + "grad_norm": 2.8915958404541016, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9041288495063782, + "num_tokens": 774960324.0, + "step": 19969 + }, + { + "epoch": 2.5403892634524867, + "grad_norm": 2.5697884559631348, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9070827960968018, + "num_tokens": 775005458.0, + "step": 19970 + }, + { + "epoch": 2.5405164737310777, + "grad_norm": 3.0582327842712402, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9020218253135681, + "num_tokens": 775041387.0, + "step": 19971 + }, + { + "epoch": 2.540643684009668, + "grad_norm": 3.0158426761627197, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.905346691608429, + "num_tokens": 775075626.0, + "step": 19972 + }, + { + "epoch": 2.5407708942882588, + "grad_norm": 2.8952479362487793, + "learning_rate": 1e-06, + "loss": 0.2765, + "mean_token_accuracy": 0.8995535969734192, + "num_tokens": 775119572.0, + "step": 19973 + }, + { + "epoch": 2.540898104566849, + "grad_norm": 2.6382694244384766, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9034358263015747, + "num_tokens": 775166513.0, + "step": 19974 + }, + { + "epoch": 2.5410253148454394, + "grad_norm": 2.894726514816284, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9062552452087402, + "num_tokens": 775202981.0, + "step": 19975 + }, + { + "epoch": 2.54115252512403, + "grad_norm": 2.7531845569610596, + "learning_rate": 1e-06, + "loss": 0.3088, + "mean_token_accuracy": 0.8871989250183105, + "num_tokens": 775249080.0, + "step": 19976 + }, + { + "epoch": 2.5412797354026204, + "grad_norm": 2.8142282962799072, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9072334170341492, + "num_tokens": 775288390.0, + "step": 19977 + }, + { + "epoch": 2.541406945681211, + "grad_norm": 2.9829976558685303, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8963624238967896, + "num_tokens": 775329221.0, + "step": 19978 + }, + { + "epoch": 2.5415341559598015, + "grad_norm": 3.0216586589813232, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8852255344390869, + "num_tokens": 775369429.0, + "step": 19979 + }, + { + "epoch": 2.541661366238392, + "grad_norm": 2.6882665157318115, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.909885048866272, + "num_tokens": 775409961.0, + "step": 19980 + }, + { + "epoch": 2.5417885765169825, + "grad_norm": 2.7755792140960693, + "learning_rate": 1e-06, + "loss": 0.2425, + "mean_token_accuracy": 0.907823920249939, + "num_tokens": 775449053.0, + "step": 19981 + }, + { + "epoch": 2.541915786795573, + "grad_norm": 2.9965715408325195, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9091996550559998, + "num_tokens": 775485561.0, + "step": 19982 + }, + { + "epoch": 2.5420429970741636, + "grad_norm": 3.1508615016937256, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8933966159820557, + "num_tokens": 775522917.0, + "step": 19983 + }, + { + "epoch": 2.542170207352754, + "grad_norm": 2.8010122776031494, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9057906866073608, + "num_tokens": 775563037.0, + "step": 19984 + }, + { + "epoch": 2.5422974176313446, + "grad_norm": 2.9202396869659424, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9050220251083374, + "num_tokens": 775600653.0, + "step": 19985 + }, + { + "epoch": 2.542424627909935, + "grad_norm": 3.093902349472046, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9023152589797974, + "num_tokens": 775635345.0, + "step": 19986 + }, + { + "epoch": 2.5425518381885257, + "grad_norm": 2.6892576217651367, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9057507514953613, + "num_tokens": 775678566.0, + "step": 19987 + }, + { + "epoch": 2.5426790484671162, + "grad_norm": 2.6090810298919678, + "learning_rate": 1e-06, + "loss": 0.226, + "mean_token_accuracy": 0.9168064594268799, + "num_tokens": 775721622.0, + "step": 19988 + }, + { + "epoch": 2.5428062587457068, + "grad_norm": 2.8643884658813477, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9101450443267822, + "num_tokens": 775757568.0, + "step": 19989 + }, + { + "epoch": 2.5429334690242973, + "grad_norm": 2.8170371055603027, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9069452285766602, + "num_tokens": 775794285.0, + "step": 19990 + }, + { + "epoch": 2.543060679302888, + "grad_norm": 2.8689112663269043, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8961350917816162, + "num_tokens": 775832623.0, + "step": 19991 + }, + { + "epoch": 2.5431878895814783, + "grad_norm": 2.608586072921753, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9040948748588562, + "num_tokens": 775879685.0, + "step": 19992 + }, + { + "epoch": 2.5433150998600684, + "grad_norm": 3.1689138412475586, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9036056995391846, + "num_tokens": 775913332.0, + "step": 19993 + }, + { + "epoch": 2.5434423101386594, + "grad_norm": 2.8801064491271973, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9055290818214417, + "num_tokens": 775952794.0, + "step": 19994 + }, + { + "epoch": 2.5435695204172495, + "grad_norm": 3.133347272872925, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9027904868125916, + "num_tokens": 775985063.0, + "step": 19995 + }, + { + "epoch": 2.5436967306958405, + "grad_norm": 2.781306266784668, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9010159969329834, + "num_tokens": 776025939.0, + "step": 19996 + }, + { + "epoch": 2.5438239409744305, + "grad_norm": 2.779265880584717, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9058776497840881, + "num_tokens": 776066333.0, + "step": 19997 + }, + { + "epoch": 2.5439511512530215, + "grad_norm": 3.0308094024658203, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.904055655002594, + "num_tokens": 776105441.0, + "step": 19998 + }, + { + "epoch": 2.5440783615316116, + "grad_norm": 2.980067253112793, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9071582555770874, + "num_tokens": 776141052.0, + "step": 19999 + }, + { + "epoch": 2.544205571810202, + "grad_norm": 3.0571048259735107, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8967914581298828, + "num_tokens": 776179491.0, + "step": 20000 + }, + { + "epoch": 2.5443327820887927, + "grad_norm": 2.9187510013580322, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8989534378051758, + "num_tokens": 776220122.0, + "step": 20001 + }, + { + "epoch": 2.544459992367383, + "grad_norm": 3.710123062133789, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8924427032470703, + "num_tokens": 776253908.0, + "step": 20002 + }, + { + "epoch": 2.5445872026459737, + "grad_norm": 2.7737152576446533, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8962637186050415, + "num_tokens": 776298372.0, + "step": 20003 + }, + { + "epoch": 2.5447144129245642, + "grad_norm": 2.724346399307251, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.9073120951652527, + "num_tokens": 776338385.0, + "step": 20004 + }, + { + "epoch": 2.5448416232031548, + "grad_norm": 2.8148062229156494, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9008485078811646, + "num_tokens": 776376829.0, + "step": 20005 + }, + { + "epoch": 2.5449688334817453, + "grad_norm": 2.9895927906036377, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.8973602056503296, + "num_tokens": 776413378.0, + "step": 20006 + }, + { + "epoch": 2.545096043760336, + "grad_norm": 2.792494058609009, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9067732095718384, + "num_tokens": 776451117.0, + "step": 20007 + }, + { + "epoch": 2.5452232540389264, + "grad_norm": 2.7790534496307373, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9048210382461548, + "num_tokens": 776490578.0, + "step": 20008 + }, + { + "epoch": 2.545350464317517, + "grad_norm": 3.1006948947906494, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9079306125640869, + "num_tokens": 776525755.0, + "step": 20009 + }, + { + "epoch": 2.5454776745961074, + "grad_norm": 2.7499794960021973, + "learning_rate": 1e-06, + "loss": 0.2234, + "mean_token_accuracy": 0.9158734083175659, + "num_tokens": 776562546.0, + "step": 20010 + }, + { + "epoch": 2.545604884874698, + "grad_norm": 3.08109188079834, + "learning_rate": 1e-06, + "loss": 0.2331, + "mean_token_accuracy": 0.9119731783866882, + "num_tokens": 776593483.0, + "step": 20011 + }, + { + "epoch": 2.5457320951532885, + "grad_norm": 2.8561654090881348, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8980298042297363, + "num_tokens": 776633277.0, + "step": 20012 + }, + { + "epoch": 2.545859305431879, + "grad_norm": 2.756650447845459, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9047443866729736, + "num_tokens": 776676989.0, + "step": 20013 + }, + { + "epoch": 2.5459865157104695, + "grad_norm": 2.7184572219848633, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9068155884742737, + "num_tokens": 776718682.0, + "step": 20014 + }, + { + "epoch": 2.54611372598906, + "grad_norm": 2.937037467956543, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8928233981132507, + "num_tokens": 776763770.0, + "step": 20015 + }, + { + "epoch": 2.5462409362676506, + "grad_norm": 2.9730334281921387, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8993147611618042, + "num_tokens": 776806021.0, + "step": 20016 + }, + { + "epoch": 2.546368146546241, + "grad_norm": 2.954841136932373, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8923097252845764, + "num_tokens": 776848187.0, + "step": 20017 + }, + { + "epoch": 2.546495356824831, + "grad_norm": 3.5140740871429443, + "learning_rate": 1e-06, + "loss": 0.3156, + "mean_token_accuracy": 0.8824958801269531, + "num_tokens": 776882230.0, + "step": 20018 + }, + { + "epoch": 2.546622567103422, + "grad_norm": 2.883958339691162, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9055620431900024, + "num_tokens": 776924151.0, + "step": 20019 + }, + { + "epoch": 2.5467497773820122, + "grad_norm": 3.116954803466797, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8948267698287964, + "num_tokens": 776962617.0, + "step": 20020 + }, + { + "epoch": 2.546876987660603, + "grad_norm": 2.855374336242676, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9015142321586609, + "num_tokens": 777002460.0, + "step": 20021 + }, + { + "epoch": 2.5470041979391933, + "grad_norm": 3.1089935302734375, + "learning_rate": 1e-06, + "loss": 0.301, + "mean_token_accuracy": 0.8925004601478577, + "num_tokens": 777039187.0, + "step": 20022 + }, + { + "epoch": 2.547131408217784, + "grad_norm": 2.7220356464385986, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.8894703984260559, + "num_tokens": 777080966.0, + "step": 20023 + }, + { + "epoch": 2.5472586184963744, + "grad_norm": 2.9704067707061768, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8900304436683655, + "num_tokens": 777122693.0, + "step": 20024 + }, + { + "epoch": 2.547385828774965, + "grad_norm": 3.2133235931396484, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9071110486984253, + "num_tokens": 777155024.0, + "step": 20025 + }, + { + "epoch": 2.5475130390535554, + "grad_norm": 2.794095993041992, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9024336338043213, + "num_tokens": 777199977.0, + "step": 20026 + }, + { + "epoch": 2.547640249332146, + "grad_norm": 2.9668965339660645, + "learning_rate": 1e-06, + "loss": 0.2438, + "mean_token_accuracy": 0.909311056137085, + "num_tokens": 777234206.0, + "step": 20027 + }, + { + "epoch": 2.5477674596107365, + "grad_norm": 2.8780765533447266, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.908305823802948, + "num_tokens": 777270484.0, + "step": 20028 + }, + { + "epoch": 2.547894669889327, + "grad_norm": 2.8411357402801514, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.910689651966095, + "num_tokens": 777306107.0, + "step": 20029 + }, + { + "epoch": 2.5480218801679175, + "grad_norm": 2.768296480178833, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9085326194763184, + "num_tokens": 777343519.0, + "step": 20030 + }, + { + "epoch": 2.548149090446508, + "grad_norm": 2.8791539669036865, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9102444648742676, + "num_tokens": 777379098.0, + "step": 20031 + }, + { + "epoch": 2.5482763007250986, + "grad_norm": 3.089696168899536, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8973963856697083, + "num_tokens": 777419167.0, + "step": 20032 + }, + { + "epoch": 2.548403511003689, + "grad_norm": 2.807868480682373, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.8996340036392212, + "num_tokens": 777456780.0, + "step": 20033 + }, + { + "epoch": 2.5485307212822796, + "grad_norm": 2.6713569164276123, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.908785879611969, + "num_tokens": 777502320.0, + "step": 20034 + }, + { + "epoch": 2.54865793156087, + "grad_norm": 2.8354146480560303, + "learning_rate": 1e-06, + "loss": 0.2337, + "mean_token_accuracy": 0.9098516702651978, + "num_tokens": 777538897.0, + "step": 20035 + }, + { + "epoch": 2.5487851418394607, + "grad_norm": 2.776033878326416, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9106442928314209, + "num_tokens": 777577024.0, + "step": 20036 + }, + { + "epoch": 2.5489123521180512, + "grad_norm": 2.974726915359497, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9040344953536987, + "num_tokens": 777613295.0, + "step": 20037 + }, + { + "epoch": 2.5490395623966418, + "grad_norm": 2.8937385082244873, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9097452759742737, + "num_tokens": 777650403.0, + "step": 20038 + }, + { + "epoch": 2.5491667726752323, + "grad_norm": 2.818843126296997, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8985946774482727, + "num_tokens": 777692581.0, + "step": 20039 + }, + { + "epoch": 2.549293982953823, + "grad_norm": 2.728466272354126, + "learning_rate": 1e-06, + "loss": 0.2432, + "mean_token_accuracy": 0.9092490673065186, + "num_tokens": 777733731.0, + "step": 20040 + }, + { + "epoch": 2.5494211932324133, + "grad_norm": 3.0068013668060303, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9091251492500305, + "num_tokens": 777770728.0, + "step": 20041 + }, + { + "epoch": 2.549548403511004, + "grad_norm": 2.8941242694854736, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9122763276100159, + "num_tokens": 777808181.0, + "step": 20042 + }, + { + "epoch": 2.549675613789594, + "grad_norm": 3.045442581176758, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9011828303337097, + "num_tokens": 777845380.0, + "step": 20043 + }, + { + "epoch": 2.549802824068185, + "grad_norm": 2.710312843322754, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.9116573333740234, + "num_tokens": 777885191.0, + "step": 20044 + }, + { + "epoch": 2.549930034346775, + "grad_norm": 2.742877721786499, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.9137155413627625, + "num_tokens": 777924231.0, + "step": 20045 + }, + { + "epoch": 2.550057244625366, + "grad_norm": 3.3866965770721436, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9014424085617065, + "num_tokens": 777954390.0, + "step": 20046 + }, + { + "epoch": 2.550184454903956, + "grad_norm": 2.96026873588562, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9064887762069702, + "num_tokens": 777988154.0, + "step": 20047 + }, + { + "epoch": 2.5503116651825466, + "grad_norm": 2.880523443222046, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.9151057004928589, + "num_tokens": 778020713.0, + "step": 20048 + }, + { + "epoch": 2.550438875461137, + "grad_norm": 2.930602788925171, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8984153866767883, + "num_tokens": 778056437.0, + "step": 20049 + }, + { + "epoch": 2.5505660857397277, + "grad_norm": 2.8311572074890137, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.8980714082717896, + "num_tokens": 778095090.0, + "step": 20050 + }, + { + "epoch": 2.550693296018318, + "grad_norm": 2.8778603076934814, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8998898267745972, + "num_tokens": 778136735.0, + "step": 20051 + }, + { + "epoch": 2.5508205062969087, + "grad_norm": 2.5722098350524902, + "learning_rate": 1e-06, + "loss": 0.2163, + "mean_token_accuracy": 0.9186999797821045, + "num_tokens": 778172844.0, + "step": 20052 + }, + { + "epoch": 2.5509477165754992, + "grad_norm": 3.0147807598114014, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9071332812309265, + "num_tokens": 778209156.0, + "step": 20053 + }, + { + "epoch": 2.5510749268540898, + "grad_norm": 2.8604464530944824, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9064053893089294, + "num_tokens": 778247219.0, + "step": 20054 + }, + { + "epoch": 2.5512021371326803, + "grad_norm": 2.8186771869659424, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8973069190979004, + "num_tokens": 778294522.0, + "step": 20055 + }, + { + "epoch": 2.551329347411271, + "grad_norm": 3.1270179748535156, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9103174209594727, + "num_tokens": 778328069.0, + "step": 20056 + }, + { + "epoch": 2.5514565576898613, + "grad_norm": 2.6916303634643555, + "learning_rate": 1e-06, + "loss": 0.2288, + "mean_token_accuracy": 0.9150878190994263, + "num_tokens": 778366272.0, + "step": 20057 + }, + { + "epoch": 2.551583767968452, + "grad_norm": 3.0483481884002686, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9106637239456177, + "num_tokens": 778398966.0, + "step": 20058 + }, + { + "epoch": 2.5517109782470424, + "grad_norm": 2.75069260597229, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9027747511863708, + "num_tokens": 778442048.0, + "step": 20059 + }, + { + "epoch": 2.551838188525633, + "grad_norm": 2.854578971862793, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9092235565185547, + "num_tokens": 778480084.0, + "step": 20060 + }, + { + "epoch": 2.5519653988042235, + "grad_norm": 3.0060248374938965, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.8998547792434692, + "num_tokens": 778517970.0, + "step": 20061 + }, + { + "epoch": 2.552092609082814, + "grad_norm": 2.822641372680664, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9044327735900879, + "num_tokens": 778557321.0, + "step": 20062 + }, + { + "epoch": 2.5522198193614045, + "grad_norm": 2.833667039871216, + "learning_rate": 1e-06, + "loss": 0.2308, + "mean_token_accuracy": 0.9135319590568542, + "num_tokens": 778595164.0, + "step": 20063 + }, + { + "epoch": 2.552347029639995, + "grad_norm": 2.9698665142059326, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9032130241394043, + "num_tokens": 778634458.0, + "step": 20064 + }, + { + "epoch": 2.5524742399185856, + "grad_norm": 2.832841396331787, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9037518501281738, + "num_tokens": 778675334.0, + "step": 20065 + }, + { + "epoch": 2.5526014501971757, + "grad_norm": 3.023038625717163, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9048165082931519, + "num_tokens": 778711282.0, + "step": 20066 + }, + { + "epoch": 2.5527286604757666, + "grad_norm": 2.9318737983703613, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.9045715928077698, + "num_tokens": 778751632.0, + "step": 20067 + }, + { + "epoch": 2.5528558707543567, + "grad_norm": 2.9381420612335205, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9014979600906372, + "num_tokens": 778791184.0, + "step": 20068 + }, + { + "epoch": 2.5529830810329477, + "grad_norm": 3.201692581176758, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8880658149719238, + "num_tokens": 778828743.0, + "step": 20069 + }, + { + "epoch": 2.5531102913115378, + "grad_norm": 3.021956443786621, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9005350470542908, + "num_tokens": 778867115.0, + "step": 20070 + }, + { + "epoch": 2.5532375015901287, + "grad_norm": 2.9004406929016113, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8921264410018921, + "num_tokens": 778912538.0, + "step": 20071 + }, + { + "epoch": 2.553364711868719, + "grad_norm": 3.012601852416992, + "learning_rate": 1e-06, + "loss": 0.3101, + "mean_token_accuracy": 0.885140061378479, + "num_tokens": 778952740.0, + "step": 20072 + }, + { + "epoch": 2.5534919221473094, + "grad_norm": 2.8750414848327637, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9065861701965332, + "num_tokens": 778991389.0, + "step": 20073 + }, + { + "epoch": 2.5536191324259, + "grad_norm": 2.7619102001190186, + "learning_rate": 1e-06, + "loss": 0.2176, + "mean_token_accuracy": 0.9167388677597046, + "num_tokens": 779027668.0, + "step": 20074 + }, + { + "epoch": 2.5537463427044904, + "grad_norm": 2.6786298751831055, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8971844911575317, + "num_tokens": 779072905.0, + "step": 20075 + }, + { + "epoch": 2.553873552983081, + "grad_norm": 3.03085994720459, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9060143828392029, + "num_tokens": 779112804.0, + "step": 20076 + }, + { + "epoch": 2.5540007632616715, + "grad_norm": 2.8112947940826416, + "learning_rate": 1e-06, + "loss": 0.2405, + "mean_token_accuracy": 0.9077835083007812, + "num_tokens": 779153159.0, + "step": 20077 + }, + { + "epoch": 2.554127973540262, + "grad_norm": 2.495933771133423, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9125343561172485, + "num_tokens": 779201301.0, + "step": 20078 + }, + { + "epoch": 2.5542551838188525, + "grad_norm": 2.7300937175750732, + "learning_rate": 1e-06, + "loss": 0.2297, + "mean_token_accuracy": 0.9149881601333618, + "num_tokens": 779239750.0, + "step": 20079 + }, + { + "epoch": 2.554382394097443, + "grad_norm": 2.8661768436431885, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9066120982170105, + "num_tokens": 779280494.0, + "step": 20080 + }, + { + "epoch": 2.5545096043760336, + "grad_norm": 2.7186989784240723, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9060759544372559, + "num_tokens": 779322610.0, + "step": 20081 + }, + { + "epoch": 2.554636814654624, + "grad_norm": 3.0804953575134277, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8946359157562256, + "num_tokens": 779362144.0, + "step": 20082 + }, + { + "epoch": 2.5547640249332146, + "grad_norm": 2.7550172805786133, + "learning_rate": 1e-06, + "loss": 0.2321, + "mean_token_accuracy": 0.9146398305892944, + "num_tokens": 779401450.0, + "step": 20083 + }, + { + "epoch": 2.554891235211805, + "grad_norm": 2.71662974357605, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9084941148757935, + "num_tokens": 779444111.0, + "step": 20084 + }, + { + "epoch": 2.5550184454903957, + "grad_norm": 2.5380725860595703, + "learning_rate": 1e-06, + "loss": 0.2043, + "mean_token_accuracy": 0.9210882782936096, + "num_tokens": 779484155.0, + "step": 20085 + }, + { + "epoch": 2.5551456557689862, + "grad_norm": 2.7953033447265625, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9134461879730225, + "num_tokens": 779525940.0, + "step": 20086 + }, + { + "epoch": 2.5552728660475768, + "grad_norm": 2.872799873352051, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9073741436004639, + "num_tokens": 779563634.0, + "step": 20087 + }, + { + "epoch": 2.5554000763261673, + "grad_norm": 3.0632247924804688, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9016082882881165, + "num_tokens": 779602707.0, + "step": 20088 + }, + { + "epoch": 2.555527286604758, + "grad_norm": 2.6416220664978027, + "learning_rate": 1e-06, + "loss": 0.2294, + "mean_token_accuracy": 0.9154466390609741, + "num_tokens": 779643748.0, + "step": 20089 + }, + { + "epoch": 2.5556544968833483, + "grad_norm": 2.758251905441284, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9030929207801819, + "num_tokens": 779685082.0, + "step": 20090 + }, + { + "epoch": 2.5557817071619384, + "grad_norm": 3.111693859100342, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9008735418319702, + "num_tokens": 779718292.0, + "step": 20091 + }, + { + "epoch": 2.5559089174405294, + "grad_norm": 3.2099828720092773, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8964636921882629, + "num_tokens": 779755098.0, + "step": 20092 + }, + { + "epoch": 2.5560361277191195, + "grad_norm": 2.9175448417663574, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9006541967391968, + "num_tokens": 779797248.0, + "step": 20093 + }, + { + "epoch": 2.5561633379977104, + "grad_norm": 3.2447669506073, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9014624357223511, + "num_tokens": 779832067.0, + "step": 20094 + }, + { + "epoch": 2.5562905482763005, + "grad_norm": 2.831697940826416, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9026126265525818, + "num_tokens": 779873509.0, + "step": 20095 + }, + { + "epoch": 2.5564177585548915, + "grad_norm": 3.0298542976379395, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9140851497650146, + "num_tokens": 779910075.0, + "step": 20096 + }, + { + "epoch": 2.5565449688334816, + "grad_norm": 3.196101188659668, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.8993152976036072, + "num_tokens": 779944419.0, + "step": 20097 + }, + { + "epoch": 2.556672179112072, + "grad_norm": 3.080094814300537, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9088302850723267, + "num_tokens": 779973786.0, + "step": 20098 + }, + { + "epoch": 2.5567993893906626, + "grad_norm": 2.826087713241577, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9016134738922119, + "num_tokens": 780013451.0, + "step": 20099 + }, + { + "epoch": 2.556926599669253, + "grad_norm": 2.7965805530548096, + "learning_rate": 1e-06, + "loss": 0.2378, + "mean_token_accuracy": 0.9120786190032959, + "num_tokens": 780051592.0, + "step": 20100 + }, + { + "epoch": 2.5570538099478437, + "grad_norm": 2.7098443508148193, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9054461121559143, + "num_tokens": 780094546.0, + "step": 20101 + }, + { + "epoch": 2.5571810202264342, + "grad_norm": 2.8762054443359375, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8978861570358276, + "num_tokens": 780131995.0, + "step": 20102 + }, + { + "epoch": 2.5573082305050248, + "grad_norm": 2.748946189880371, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8968493938446045, + "num_tokens": 780181218.0, + "step": 20103 + }, + { + "epoch": 2.5574354407836153, + "grad_norm": 3.018855333328247, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.9015893340110779, + "num_tokens": 780219090.0, + "step": 20104 + }, + { + "epoch": 2.557562651062206, + "grad_norm": 2.7938554286956787, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.8981534242630005, + "num_tokens": 780264537.0, + "step": 20105 + }, + { + "epoch": 2.5576898613407963, + "grad_norm": 2.960350513458252, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9012219905853271, + "num_tokens": 780301270.0, + "step": 20106 + }, + { + "epoch": 2.557817071619387, + "grad_norm": 2.807295083999634, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.90694659948349, + "num_tokens": 780341281.0, + "step": 20107 + }, + { + "epoch": 2.5579442818979774, + "grad_norm": 2.945073366165161, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9085825085639954, + "num_tokens": 780380231.0, + "step": 20108 + }, + { + "epoch": 2.558071492176568, + "grad_norm": 3.0231411457061768, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9003677368164062, + "num_tokens": 780418166.0, + "step": 20109 + }, + { + "epoch": 2.5581987024551585, + "grad_norm": 3.163597822189331, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8776686787605286, + "num_tokens": 780458190.0, + "step": 20110 + }, + { + "epoch": 2.558325912733749, + "grad_norm": 2.7599544525146484, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9067760109901428, + "num_tokens": 780497005.0, + "step": 20111 + }, + { + "epoch": 2.5584531230123395, + "grad_norm": 2.9229345321655273, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9021750092506409, + "num_tokens": 780536923.0, + "step": 20112 + }, + { + "epoch": 2.55858033329093, + "grad_norm": 2.7140796184539795, + "learning_rate": 1e-06, + "loss": 0.2206, + "mean_token_accuracy": 0.9195005297660828, + "num_tokens": 780576667.0, + "step": 20113 + }, + { + "epoch": 2.5587075435695206, + "grad_norm": 3.158705234527588, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9046627283096313, + "num_tokens": 780609501.0, + "step": 20114 + }, + { + "epoch": 2.558834753848111, + "grad_norm": 2.6398637294769287, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9132432341575623, + "num_tokens": 780649443.0, + "step": 20115 + }, + { + "epoch": 2.558961964126701, + "grad_norm": 2.740877866744995, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9138500094413757, + "num_tokens": 780688329.0, + "step": 20116 + }, + { + "epoch": 2.559089174405292, + "grad_norm": 3.118807315826416, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9047217965126038, + "num_tokens": 780721195.0, + "step": 20117 + }, + { + "epoch": 2.5592163846838822, + "grad_norm": 2.6392033100128174, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9029161334037781, + "num_tokens": 780765842.0, + "step": 20118 + }, + { + "epoch": 2.559343594962473, + "grad_norm": 2.6899914741516113, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9030971527099609, + "num_tokens": 780811544.0, + "step": 20119 + }, + { + "epoch": 2.5594708052410633, + "grad_norm": 2.687607526779175, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9066938757896423, + "num_tokens": 780853898.0, + "step": 20120 + }, + { + "epoch": 2.559598015519654, + "grad_norm": 2.7649497985839844, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.904938280582428, + "num_tokens": 780894148.0, + "step": 20121 + }, + { + "epoch": 2.5597252257982444, + "grad_norm": 2.9828546047210693, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9033356308937073, + "num_tokens": 780931427.0, + "step": 20122 + }, + { + "epoch": 2.559852436076835, + "grad_norm": 2.8362855911254883, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.8984949588775635, + "num_tokens": 780973254.0, + "step": 20123 + }, + { + "epoch": 2.5599796463554254, + "grad_norm": 2.765584945678711, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9042321443557739, + "num_tokens": 781016609.0, + "step": 20124 + }, + { + "epoch": 2.560106856634016, + "grad_norm": 2.903900623321533, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9025102853775024, + "num_tokens": 781058577.0, + "step": 20125 + }, + { + "epoch": 2.5602340669126065, + "grad_norm": 3.071073293685913, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9075446128845215, + "num_tokens": 781092145.0, + "step": 20126 + }, + { + "epoch": 2.560361277191197, + "grad_norm": 2.7437539100646973, + "learning_rate": 1e-06, + "loss": 0.2316, + "mean_token_accuracy": 0.9146816730499268, + "num_tokens": 781133058.0, + "step": 20127 + }, + { + "epoch": 2.5604884874697875, + "grad_norm": 2.942357063293457, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9082449674606323, + "num_tokens": 781167054.0, + "step": 20128 + }, + { + "epoch": 2.560615697748378, + "grad_norm": 2.7900009155273438, + "learning_rate": 1e-06, + "loss": 0.2266, + "mean_token_accuracy": 0.9135585427284241, + "num_tokens": 781201758.0, + "step": 20129 + }, + { + "epoch": 2.5607429080269686, + "grad_norm": 2.9570767879486084, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9047752022743225, + "num_tokens": 781241806.0, + "step": 20130 + }, + { + "epoch": 2.560870118305559, + "grad_norm": 2.924931049346924, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9046359062194824, + "num_tokens": 781283184.0, + "step": 20131 + }, + { + "epoch": 2.5609973285841496, + "grad_norm": 3.0381600856781006, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9032646417617798, + "num_tokens": 781318125.0, + "step": 20132 + }, + { + "epoch": 2.56112453886274, + "grad_norm": 2.758913516998291, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8936523795127869, + "num_tokens": 781366819.0, + "step": 20133 + }, + { + "epoch": 2.5612517491413307, + "grad_norm": 3.039583206176758, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9023653268814087, + "num_tokens": 781402018.0, + "step": 20134 + }, + { + "epoch": 2.561378959419921, + "grad_norm": 3.240281105041504, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8954688310623169, + "num_tokens": 781437516.0, + "step": 20135 + }, + { + "epoch": 2.5615061696985117, + "grad_norm": 2.8749732971191406, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9037470817565918, + "num_tokens": 781478000.0, + "step": 20136 + }, + { + "epoch": 2.5616333799771023, + "grad_norm": 2.850270986557007, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9084123373031616, + "num_tokens": 781516940.0, + "step": 20137 + }, + { + "epoch": 2.561760590255693, + "grad_norm": 2.9599955081939697, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8983389139175415, + "num_tokens": 781555068.0, + "step": 20138 + }, + { + "epoch": 2.5618878005342833, + "grad_norm": 2.9255316257476807, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.91011643409729, + "num_tokens": 781590617.0, + "step": 20139 + }, + { + "epoch": 2.562015010812874, + "grad_norm": 2.761664390563965, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9124203324317932, + "num_tokens": 781626914.0, + "step": 20140 + }, + { + "epoch": 2.562142221091464, + "grad_norm": 3.0202748775482178, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9041029214859009, + "num_tokens": 781664579.0, + "step": 20141 + }, + { + "epoch": 2.562269431370055, + "grad_norm": 2.9323222637176514, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9077692031860352, + "num_tokens": 781702134.0, + "step": 20142 + }, + { + "epoch": 2.562396641648645, + "grad_norm": 3.2060999870300293, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.89610755443573, + "num_tokens": 781735087.0, + "step": 20143 + }, + { + "epoch": 2.562523851927236, + "grad_norm": 2.5978503227233887, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9070644378662109, + "num_tokens": 781779607.0, + "step": 20144 + }, + { + "epoch": 2.562651062205826, + "grad_norm": 2.7810094356536865, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9063317775726318, + "num_tokens": 781822132.0, + "step": 20145 + }, + { + "epoch": 2.5627782724844166, + "grad_norm": 2.78877329826355, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9020857214927673, + "num_tokens": 781865513.0, + "step": 20146 + }, + { + "epoch": 2.562905482763007, + "grad_norm": 3.1544342041015625, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9059784412384033, + "num_tokens": 781899727.0, + "step": 20147 + }, + { + "epoch": 2.5630326930415976, + "grad_norm": 2.666724443435669, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.900528073310852, + "num_tokens": 781944750.0, + "step": 20148 + }, + { + "epoch": 2.563159903320188, + "grad_norm": 3.050297975540161, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8946613669395447, + "num_tokens": 781981860.0, + "step": 20149 + }, + { + "epoch": 2.5632871135987787, + "grad_norm": 2.951957941055298, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9117988348007202, + "num_tokens": 782020292.0, + "step": 20150 + }, + { + "epoch": 2.5634143238773692, + "grad_norm": 2.9727425575256348, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9064626693725586, + "num_tokens": 782055740.0, + "step": 20151 + }, + { + "epoch": 2.5635415341559598, + "grad_norm": 2.802013635635376, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9073297381401062, + "num_tokens": 782092813.0, + "step": 20152 + }, + { + "epoch": 2.5636687444345503, + "grad_norm": 2.7583119869232178, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9058637022972107, + "num_tokens": 782133272.0, + "step": 20153 + }, + { + "epoch": 2.563795954713141, + "grad_norm": 3.2812726497650146, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9109545946121216, + "num_tokens": 782162272.0, + "step": 20154 + }, + { + "epoch": 2.5639231649917313, + "grad_norm": 2.9965662956237793, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9028459787368774, + "num_tokens": 782198953.0, + "step": 20155 + }, + { + "epoch": 2.564050375270322, + "grad_norm": 2.8269541263580322, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8963667154312134, + "num_tokens": 782242834.0, + "step": 20156 + }, + { + "epoch": 2.5641775855489124, + "grad_norm": 2.923680067062378, + "learning_rate": 1e-06, + "loss": 0.2348, + "mean_token_accuracy": 0.9141595363616943, + "num_tokens": 782277403.0, + "step": 20157 + }, + { + "epoch": 2.564304795827503, + "grad_norm": 2.6604182720184326, + "learning_rate": 1e-06, + "loss": 0.2671, + "mean_token_accuracy": 0.9037609696388245, + "num_tokens": 782325158.0, + "step": 20158 + }, + { + "epoch": 2.5644320061060935, + "grad_norm": 3.0280065536499023, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.8991398215293884, + "num_tokens": 782365031.0, + "step": 20159 + }, + { + "epoch": 2.564559216384684, + "grad_norm": 2.837172031402588, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.907066285610199, + "num_tokens": 782403783.0, + "step": 20160 + }, + { + "epoch": 2.5646864266632745, + "grad_norm": 2.7348146438598633, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9118392467498779, + "num_tokens": 782440368.0, + "step": 20161 + }, + { + "epoch": 2.564813636941865, + "grad_norm": 2.7832746505737305, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9065670967102051, + "num_tokens": 782482383.0, + "step": 20162 + }, + { + "epoch": 2.5649408472204556, + "grad_norm": 2.76149845123291, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9042867422103882, + "num_tokens": 782530100.0, + "step": 20163 + }, + { + "epoch": 2.5650680574990457, + "grad_norm": 2.9816253185272217, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8990426063537598, + "num_tokens": 782570446.0, + "step": 20164 + }, + { + "epoch": 2.5651952677776366, + "grad_norm": 3.030458688735962, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9087333679199219, + "num_tokens": 782604710.0, + "step": 20165 + }, + { + "epoch": 2.5653224780562267, + "grad_norm": 2.866178035736084, + "learning_rate": 1e-06, + "loss": 0.2223, + "mean_token_accuracy": 0.9193536043167114, + "num_tokens": 782639518.0, + "step": 20166 + }, + { + "epoch": 2.5654496883348177, + "grad_norm": 3.0345170497894287, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9109092354774475, + "num_tokens": 782672026.0, + "step": 20167 + }, + { + "epoch": 2.5655768986134078, + "grad_norm": 2.9183719158172607, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9069361090660095, + "num_tokens": 782708729.0, + "step": 20168 + }, + { + "epoch": 2.5657041088919987, + "grad_norm": 3.07299542427063, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9067815542221069, + "num_tokens": 782739282.0, + "step": 20169 + }, + { + "epoch": 2.565831319170589, + "grad_norm": 2.947153091430664, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.9103937149047852, + "num_tokens": 782772874.0, + "step": 20170 + }, + { + "epoch": 2.5659585294491793, + "grad_norm": 3.17634916305542, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9084529280662537, + "num_tokens": 782805736.0, + "step": 20171 + }, + { + "epoch": 2.56608573972777, + "grad_norm": 2.8924448490142822, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9039520025253296, + "num_tokens": 782846682.0, + "step": 20172 + }, + { + "epoch": 2.5662129500063604, + "grad_norm": 2.8245368003845215, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.900860071182251, + "num_tokens": 782886728.0, + "step": 20173 + }, + { + "epoch": 2.566340160284951, + "grad_norm": 2.641934394836426, + "learning_rate": 1e-06, + "loss": 0.2253, + "mean_token_accuracy": 0.9142715930938721, + "num_tokens": 782928071.0, + "step": 20174 + }, + { + "epoch": 2.5664673705635415, + "grad_norm": 2.9944207668304443, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.905407190322876, + "num_tokens": 782965466.0, + "step": 20175 + }, + { + "epoch": 2.566594580842132, + "grad_norm": 3.325890064239502, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9087554216384888, + "num_tokens": 782996147.0, + "step": 20176 + }, + { + "epoch": 2.5667217911207225, + "grad_norm": 3.171881675720215, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9134605526924133, + "num_tokens": 783028765.0, + "step": 20177 + }, + { + "epoch": 2.566849001399313, + "grad_norm": 2.8860957622528076, + "learning_rate": 1e-06, + "loss": 0.223, + "mean_token_accuracy": 0.9170737266540527, + "num_tokens": 783064032.0, + "step": 20178 + }, + { + "epoch": 2.5669762116779036, + "grad_norm": 2.5832440853118896, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9065779447555542, + "num_tokens": 783107762.0, + "step": 20179 + }, + { + "epoch": 2.567103421956494, + "grad_norm": 2.8065409660339355, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9108874201774597, + "num_tokens": 783147774.0, + "step": 20180 + }, + { + "epoch": 2.5672306322350846, + "grad_norm": 2.752017021179199, + "learning_rate": 1e-06, + "loss": 0.226, + "mean_token_accuracy": 0.9138827323913574, + "num_tokens": 783184179.0, + "step": 20181 + }, + { + "epoch": 2.567357842513675, + "grad_norm": 2.6004061698913574, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9111965298652649, + "num_tokens": 783226324.0, + "step": 20182 + }, + { + "epoch": 2.5674850527922657, + "grad_norm": 2.928926706314087, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9047947525978088, + "num_tokens": 783264846.0, + "step": 20183 + }, + { + "epoch": 2.567612263070856, + "grad_norm": 2.78348970413208, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9102796316146851, + "num_tokens": 783306560.0, + "step": 20184 + }, + { + "epoch": 2.5677394733494467, + "grad_norm": 3.1868157386779785, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8966902494430542, + "num_tokens": 783342942.0, + "step": 20185 + }, + { + "epoch": 2.5678666836280373, + "grad_norm": 2.5783069133758545, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.905145525932312, + "num_tokens": 783392172.0, + "step": 20186 + }, + { + "epoch": 2.567993893906628, + "grad_norm": 2.6732184886932373, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9104939699172974, + "num_tokens": 783436872.0, + "step": 20187 + }, + { + "epoch": 2.5681211041852183, + "grad_norm": 2.870150089263916, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9100475311279297, + "num_tokens": 783473612.0, + "step": 20188 + }, + { + "epoch": 2.5682483144638084, + "grad_norm": 2.748853921890259, + "learning_rate": 1e-06, + "loss": 0.2321, + "mean_token_accuracy": 0.912493884563446, + "num_tokens": 783510346.0, + "step": 20189 + }, + { + "epoch": 2.5683755247423994, + "grad_norm": 2.8671531677246094, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9012441635131836, + "num_tokens": 783545026.0, + "step": 20190 + }, + { + "epoch": 2.5685027350209895, + "grad_norm": 2.816270351409912, + "learning_rate": 1e-06, + "loss": 0.2248, + "mean_token_accuracy": 0.9135148525238037, + "num_tokens": 783580960.0, + "step": 20191 + }, + { + "epoch": 2.5686299452995804, + "grad_norm": 3.177993059158325, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9051393270492554, + "num_tokens": 783618552.0, + "step": 20192 + }, + { + "epoch": 2.5687571555781705, + "grad_norm": 2.529680013656616, + "learning_rate": 1e-06, + "loss": 0.2204, + "mean_token_accuracy": 0.918005645275116, + "num_tokens": 783661676.0, + "step": 20193 + }, + { + "epoch": 2.5688843658567615, + "grad_norm": 2.8011281490325928, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9097272157669067, + "num_tokens": 783697666.0, + "step": 20194 + }, + { + "epoch": 2.5690115761353516, + "grad_norm": 3.2152860164642334, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8924498558044434, + "num_tokens": 783733936.0, + "step": 20195 + }, + { + "epoch": 2.569138786413942, + "grad_norm": 2.898516893386841, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9102720022201538, + "num_tokens": 783767975.0, + "step": 20196 + }, + { + "epoch": 2.5692659966925326, + "grad_norm": 2.9100239276885986, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9017134308815002, + "num_tokens": 783807297.0, + "step": 20197 + }, + { + "epoch": 2.569393206971123, + "grad_norm": 3.0371484756469727, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9033442735671997, + "num_tokens": 783845418.0, + "step": 20198 + }, + { + "epoch": 2.5695204172497137, + "grad_norm": 2.8323936462402344, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.8983992338180542, + "num_tokens": 783887022.0, + "step": 20199 + }, + { + "epoch": 2.5696476275283042, + "grad_norm": 3.1693472862243652, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8950827717781067, + "num_tokens": 783919953.0, + "step": 20200 + }, + { + "epoch": 2.5697748378068948, + "grad_norm": 2.685487985610962, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9074814915657043, + "num_tokens": 783962708.0, + "step": 20201 + }, + { + "epoch": 2.5699020480854853, + "grad_norm": 2.755232572555542, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9063737988471985, + "num_tokens": 784003187.0, + "step": 20202 + }, + { + "epoch": 2.570029258364076, + "grad_norm": 2.7739880084991455, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9121146202087402, + "num_tokens": 784041440.0, + "step": 20203 + }, + { + "epoch": 2.5701564686426663, + "grad_norm": 2.733102560043335, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9088895320892334, + "num_tokens": 784079576.0, + "step": 20204 + }, + { + "epoch": 2.570283678921257, + "grad_norm": 2.6416492462158203, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9130834341049194, + "num_tokens": 784122355.0, + "step": 20205 + }, + { + "epoch": 2.5704108891998474, + "grad_norm": 3.087817668914795, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.8999781608581543, + "num_tokens": 784155416.0, + "step": 20206 + }, + { + "epoch": 2.570538099478438, + "grad_norm": 2.7939858436584473, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8970361351966858, + "num_tokens": 784196674.0, + "step": 20207 + }, + { + "epoch": 2.5706653097570284, + "grad_norm": 2.809813976287842, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9056562185287476, + "num_tokens": 784234697.0, + "step": 20208 + }, + { + "epoch": 2.570792520035619, + "grad_norm": 2.8835582733154297, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9029269218444824, + "num_tokens": 784274154.0, + "step": 20209 + }, + { + "epoch": 2.5709197303142095, + "grad_norm": 2.8252110481262207, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9003411531448364, + "num_tokens": 784315467.0, + "step": 20210 + }, + { + "epoch": 2.5710469405928, + "grad_norm": 2.883033037185669, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9002642631530762, + "num_tokens": 784352053.0, + "step": 20211 + }, + { + "epoch": 2.5711741508713906, + "grad_norm": 2.746049642562866, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9000117778778076, + "num_tokens": 784395117.0, + "step": 20212 + }, + { + "epoch": 2.571301361149981, + "grad_norm": 2.7512238025665283, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9106391668319702, + "num_tokens": 784434433.0, + "step": 20213 + }, + { + "epoch": 2.571428571428571, + "grad_norm": 2.8589930534362793, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9098808765411377, + "num_tokens": 784471168.0, + "step": 20214 + }, + { + "epoch": 2.571555781707162, + "grad_norm": 2.915787935256958, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.913525402545929, + "num_tokens": 784503265.0, + "step": 20215 + }, + { + "epoch": 2.5716829919857522, + "grad_norm": 2.645535945892334, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9075968861579895, + "num_tokens": 784546747.0, + "step": 20216 + }, + { + "epoch": 2.571810202264343, + "grad_norm": 2.8353664875030518, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9070985913276672, + "num_tokens": 784589863.0, + "step": 20217 + }, + { + "epoch": 2.5719374125429333, + "grad_norm": 2.982266426086426, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.900066077709198, + "num_tokens": 784630883.0, + "step": 20218 + }, + { + "epoch": 2.572064622821524, + "grad_norm": 2.9257171154022217, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9060269594192505, + "num_tokens": 784667058.0, + "step": 20219 + }, + { + "epoch": 2.5721918331001143, + "grad_norm": 3.0214922428131104, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9045188426971436, + "num_tokens": 784702438.0, + "step": 20220 + }, + { + "epoch": 2.572319043378705, + "grad_norm": 2.801515817642212, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8964999914169312, + "num_tokens": 784743102.0, + "step": 20221 + }, + { + "epoch": 2.5724462536572954, + "grad_norm": 2.9778003692626953, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8946823477745056, + "num_tokens": 784783700.0, + "step": 20222 + }, + { + "epoch": 2.572573463935886, + "grad_norm": 2.7498960494995117, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8964525461196899, + "num_tokens": 784831166.0, + "step": 20223 + }, + { + "epoch": 2.5727006742144765, + "grad_norm": 2.970686912536621, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9047055244445801, + "num_tokens": 784866621.0, + "step": 20224 + }, + { + "epoch": 2.572827884493067, + "grad_norm": 2.8916778564453125, + "learning_rate": 1e-06, + "loss": 0.2422, + "mean_token_accuracy": 0.9101338386535645, + "num_tokens": 784904169.0, + "step": 20225 + }, + { + "epoch": 2.5729550947716575, + "grad_norm": 2.6529698371887207, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9011306762695312, + "num_tokens": 784951623.0, + "step": 20226 + }, + { + "epoch": 2.573082305050248, + "grad_norm": 2.9181466102600098, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9090911149978638, + "num_tokens": 784989649.0, + "step": 20227 + }, + { + "epoch": 2.5732095153288386, + "grad_norm": 2.932581901550293, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9082773923873901, + "num_tokens": 785026481.0, + "step": 20228 + }, + { + "epoch": 2.573336725607429, + "grad_norm": 2.9081661701202393, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.896120548248291, + "num_tokens": 785072732.0, + "step": 20229 + }, + { + "epoch": 2.5734639358860196, + "grad_norm": 2.9690420627593994, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9080625772476196, + "num_tokens": 785106667.0, + "step": 20230 + }, + { + "epoch": 2.57359114616461, + "grad_norm": 2.833538055419922, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.8969626426696777, + "num_tokens": 785149524.0, + "step": 20231 + }, + { + "epoch": 2.5737183564432007, + "grad_norm": 2.8982701301574707, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9059537649154663, + "num_tokens": 785190163.0, + "step": 20232 + }, + { + "epoch": 2.573845566721791, + "grad_norm": 3.241748571395874, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8993096947669983, + "num_tokens": 785221907.0, + "step": 20233 + }, + { + "epoch": 2.5739727770003817, + "grad_norm": 2.8367321491241455, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.9087332487106323, + "num_tokens": 785256573.0, + "step": 20234 + }, + { + "epoch": 2.5740999872789723, + "grad_norm": 2.750943183898926, + "learning_rate": 1e-06, + "loss": 0.2254, + "mean_token_accuracy": 0.9177989959716797, + "num_tokens": 785294752.0, + "step": 20235 + }, + { + "epoch": 2.574227197557563, + "grad_norm": 3.117222785949707, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9083462357521057, + "num_tokens": 785331723.0, + "step": 20236 + }, + { + "epoch": 2.5743544078361533, + "grad_norm": 3.4697976112365723, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8956136703491211, + "num_tokens": 785361846.0, + "step": 20237 + }, + { + "epoch": 2.574481618114744, + "grad_norm": 2.8561532497406006, + "learning_rate": 1e-06, + "loss": 0.2267, + "mean_token_accuracy": 0.9168446063995361, + "num_tokens": 785397913.0, + "step": 20238 + }, + { + "epoch": 2.574608828393334, + "grad_norm": 2.782021999359131, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.8995624780654907, + "num_tokens": 785438456.0, + "step": 20239 + }, + { + "epoch": 2.574736038671925, + "grad_norm": 2.7183871269226074, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9060071706771851, + "num_tokens": 785479672.0, + "step": 20240 + }, + { + "epoch": 2.574863248950515, + "grad_norm": 2.8424205780029297, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9038126468658447, + "num_tokens": 785516909.0, + "step": 20241 + }, + { + "epoch": 2.574990459229106, + "grad_norm": 3.0945091247558594, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8993157744407654, + "num_tokens": 785552182.0, + "step": 20242 + }, + { + "epoch": 2.575117669507696, + "grad_norm": 3.238241195678711, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9026060700416565, + "num_tokens": 785587462.0, + "step": 20243 + }, + { + "epoch": 2.5752448797862866, + "grad_norm": 3.0808374881744385, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.897524356842041, + "num_tokens": 785625326.0, + "step": 20244 + }, + { + "epoch": 2.575372090064877, + "grad_norm": 2.7354612350463867, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9100897312164307, + "num_tokens": 785664489.0, + "step": 20245 + }, + { + "epoch": 2.5754993003434676, + "grad_norm": 2.9317820072174072, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9033551812171936, + "num_tokens": 785702631.0, + "step": 20246 + }, + { + "epoch": 2.575626510622058, + "grad_norm": 3.2110679149627686, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.896557092666626, + "num_tokens": 785740161.0, + "step": 20247 + }, + { + "epoch": 2.5757537209006487, + "grad_norm": 2.5266332626342773, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.9118435382843018, + "num_tokens": 785785718.0, + "step": 20248 + }, + { + "epoch": 2.575880931179239, + "grad_norm": 3.1319026947021484, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9005087018013, + "num_tokens": 785820412.0, + "step": 20249 + }, + { + "epoch": 2.5760081414578297, + "grad_norm": 2.948720693588257, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.8992800712585449, + "num_tokens": 785857094.0, + "step": 20250 + }, + { + "epoch": 2.5761353517364203, + "grad_norm": 2.987041711807251, + "learning_rate": 1e-06, + "loss": 0.2422, + "mean_token_accuracy": 0.9088786244392395, + "num_tokens": 785890635.0, + "step": 20251 + }, + { + "epoch": 2.576262562015011, + "grad_norm": 2.8076045513153076, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9060125350952148, + "num_tokens": 785930652.0, + "step": 20252 + }, + { + "epoch": 2.5763897722936013, + "grad_norm": 2.8873093128204346, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9001240730285645, + "num_tokens": 785970592.0, + "step": 20253 + }, + { + "epoch": 2.576516982572192, + "grad_norm": 3.0844924449920654, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9100492000579834, + "num_tokens": 786007177.0, + "step": 20254 + }, + { + "epoch": 2.5766441928507824, + "grad_norm": 3.1151347160339355, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8919042944908142, + "num_tokens": 786046352.0, + "step": 20255 + }, + { + "epoch": 2.576771403129373, + "grad_norm": 3.1005666255950928, + "learning_rate": 1e-06, + "loss": 0.2983, + "mean_token_accuracy": 0.8908452987670898, + "num_tokens": 786083256.0, + "step": 20256 + }, + { + "epoch": 2.5768986134079634, + "grad_norm": 2.760023832321167, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9098212718963623, + "num_tokens": 786123109.0, + "step": 20257 + }, + { + "epoch": 2.577025823686554, + "grad_norm": 2.7661540508270264, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9048324823379517, + "num_tokens": 786165119.0, + "step": 20258 + }, + { + "epoch": 2.5771530339651445, + "grad_norm": 2.920851945877075, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.890381932258606, + "num_tokens": 786209608.0, + "step": 20259 + }, + { + "epoch": 2.577280244243735, + "grad_norm": 2.8535141944885254, + "learning_rate": 1e-06, + "loss": 0.2159, + "mean_token_accuracy": 0.9188571572303772, + "num_tokens": 786245522.0, + "step": 20260 + }, + { + "epoch": 2.5774074545223256, + "grad_norm": 3.045578718185425, + "learning_rate": 1e-06, + "loss": 0.2826, + "mean_token_accuracy": 0.8956345319747925, + "num_tokens": 786284429.0, + "step": 20261 + }, + { + "epoch": 2.5775346648009156, + "grad_norm": 2.9710683822631836, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.907401978969574, + "num_tokens": 786323178.0, + "step": 20262 + }, + { + "epoch": 2.5776618750795066, + "grad_norm": 2.8156399726867676, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9093478918075562, + "num_tokens": 786359042.0, + "step": 20263 + }, + { + "epoch": 2.5777890853580967, + "grad_norm": 3.0862629413604736, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8958079814910889, + "num_tokens": 786399423.0, + "step": 20264 + }, + { + "epoch": 2.5779162956366877, + "grad_norm": 3.028869867324829, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9086260795593262, + "num_tokens": 786433443.0, + "step": 20265 + }, + { + "epoch": 2.5780435059152778, + "grad_norm": 2.9634556770324707, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9078840613365173, + "num_tokens": 786468310.0, + "step": 20266 + }, + { + "epoch": 2.5781707161938687, + "grad_norm": 2.7660272121429443, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8964372277259827, + "num_tokens": 786513135.0, + "step": 20267 + }, + { + "epoch": 2.578297926472459, + "grad_norm": 2.9368886947631836, + "learning_rate": 1e-06, + "loss": 0.3054, + "mean_token_accuracy": 0.8924857378005981, + "num_tokens": 786554011.0, + "step": 20268 + }, + { + "epoch": 2.5784251367510493, + "grad_norm": 2.845012903213501, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8994619250297546, + "num_tokens": 786594744.0, + "step": 20269 + }, + { + "epoch": 2.57855234702964, + "grad_norm": 2.867196559906006, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9010917544364929, + "num_tokens": 786635828.0, + "step": 20270 + }, + { + "epoch": 2.5786795573082304, + "grad_norm": 2.9158236980438232, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9029248952865601, + "num_tokens": 786676462.0, + "step": 20271 + }, + { + "epoch": 2.578806767586821, + "grad_norm": 2.8183188438415527, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9093499183654785, + "num_tokens": 786716924.0, + "step": 20272 + }, + { + "epoch": 2.5789339778654115, + "grad_norm": 2.828439712524414, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8960541486740112, + "num_tokens": 786760137.0, + "step": 20273 + }, + { + "epoch": 2.579061188144002, + "grad_norm": 2.8679494857788086, + "learning_rate": 1e-06, + "loss": 0.2295, + "mean_token_accuracy": 0.9134343862533569, + "num_tokens": 786797193.0, + "step": 20274 + }, + { + "epoch": 2.5791883984225925, + "grad_norm": 2.7351953983306885, + "learning_rate": 1e-06, + "loss": 0.234, + "mean_token_accuracy": 0.9102749824523926, + "num_tokens": 786835199.0, + "step": 20275 + }, + { + "epoch": 2.579315608701183, + "grad_norm": 2.8553738594055176, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9041136503219604, + "num_tokens": 786875931.0, + "step": 20276 + }, + { + "epoch": 2.5794428189797736, + "grad_norm": 2.9663093090057373, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9043856859207153, + "num_tokens": 786915104.0, + "step": 20277 + }, + { + "epoch": 2.579570029258364, + "grad_norm": 2.9724795818328857, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9077885150909424, + "num_tokens": 786951265.0, + "step": 20278 + }, + { + "epoch": 2.5796972395369546, + "grad_norm": 2.7995359897613525, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9082251787185669, + "num_tokens": 786992587.0, + "step": 20279 + }, + { + "epoch": 2.579824449815545, + "grad_norm": 2.8576292991638184, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9076470732688904, + "num_tokens": 787029509.0, + "step": 20280 + }, + { + "epoch": 2.5799516600941357, + "grad_norm": 2.883113384246826, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.906693160533905, + "num_tokens": 787067454.0, + "step": 20281 + }, + { + "epoch": 2.580078870372726, + "grad_norm": 3.338610887527466, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.9014136791229248, + "num_tokens": 787102339.0, + "step": 20282 + }, + { + "epoch": 2.5802060806513167, + "grad_norm": 3.1562018394470215, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9105640053749084, + "num_tokens": 787138257.0, + "step": 20283 + }, + { + "epoch": 2.5803332909299073, + "grad_norm": 3.1391141414642334, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9001694321632385, + "num_tokens": 787172677.0, + "step": 20284 + }, + { + "epoch": 2.580460501208498, + "grad_norm": 2.960951089859009, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.8972700238227844, + "num_tokens": 787210615.0, + "step": 20285 + }, + { + "epoch": 2.5805877114870883, + "grad_norm": 2.807337760925293, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9085526466369629, + "num_tokens": 787248829.0, + "step": 20286 + }, + { + "epoch": 2.5807149217656784, + "grad_norm": 3.026470899581909, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8868861794471741, + "num_tokens": 787288395.0, + "step": 20287 + }, + { + "epoch": 2.5808421320442694, + "grad_norm": 3.119110345840454, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9097421765327454, + "num_tokens": 787322960.0, + "step": 20288 + }, + { + "epoch": 2.5809693423228595, + "grad_norm": 2.960435152053833, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.908034086227417, + "num_tokens": 787361322.0, + "step": 20289 + }, + { + "epoch": 2.5810965526014504, + "grad_norm": 3.0729615688323975, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.9006863832473755, + "num_tokens": 787395535.0, + "step": 20290 + }, + { + "epoch": 2.5812237628800405, + "grad_norm": 2.675191879272461, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9068928956985474, + "num_tokens": 787434650.0, + "step": 20291 + }, + { + "epoch": 2.5813509731586315, + "grad_norm": 3.3020143508911133, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8965072631835938, + "num_tokens": 787470856.0, + "step": 20292 + }, + { + "epoch": 2.5814781834372216, + "grad_norm": 2.907850503921509, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8987502455711365, + "num_tokens": 787511123.0, + "step": 20293 + }, + { + "epoch": 2.581605393715812, + "grad_norm": 2.8997080326080322, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9005550146102905, + "num_tokens": 787551396.0, + "step": 20294 + }, + { + "epoch": 2.5817326039944026, + "grad_norm": 2.4262983798980713, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9062983393669128, + "num_tokens": 787599051.0, + "step": 20295 + }, + { + "epoch": 2.581859814272993, + "grad_norm": 2.796635389328003, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9065797924995422, + "num_tokens": 787638757.0, + "step": 20296 + }, + { + "epoch": 2.5819870245515837, + "grad_norm": 2.6525373458862305, + "learning_rate": 1e-06, + "loss": 0.2195, + "mean_token_accuracy": 0.9154635667800903, + "num_tokens": 787677741.0, + "step": 20297 + }, + { + "epoch": 2.582114234830174, + "grad_norm": 2.77233624458313, + "learning_rate": 1e-06, + "loss": 0.2393, + "mean_token_accuracy": 0.9142604470252991, + "num_tokens": 787717747.0, + "step": 20298 + }, + { + "epoch": 2.5822414451087647, + "grad_norm": 2.8566787242889404, + "learning_rate": 1e-06, + "loss": 0.2435, + "mean_token_accuracy": 0.9067210555076599, + "num_tokens": 787753095.0, + "step": 20299 + }, + { + "epoch": 2.5823686553873553, + "grad_norm": 2.7484190464019775, + "learning_rate": 1e-06, + "loss": 0.2247, + "mean_token_accuracy": 0.9166971445083618, + "num_tokens": 787791214.0, + "step": 20300 + }, + { + "epoch": 2.582495865665946, + "grad_norm": 2.7366530895233154, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9052913188934326, + "num_tokens": 787831427.0, + "step": 20301 + }, + { + "epoch": 2.5826230759445363, + "grad_norm": 2.769442558288574, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9044855833053589, + "num_tokens": 787868515.0, + "step": 20302 + }, + { + "epoch": 2.582750286223127, + "grad_norm": 2.908559799194336, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9002450108528137, + "num_tokens": 787907925.0, + "step": 20303 + }, + { + "epoch": 2.5828774965017174, + "grad_norm": 2.5898401737213135, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9079384207725525, + "num_tokens": 787952251.0, + "step": 20304 + }, + { + "epoch": 2.583004706780308, + "grad_norm": 2.8698997497558594, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9064103364944458, + "num_tokens": 787992838.0, + "step": 20305 + }, + { + "epoch": 2.5831319170588984, + "grad_norm": 2.77664852142334, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9068253040313721, + "num_tokens": 788034150.0, + "step": 20306 + }, + { + "epoch": 2.583259127337489, + "grad_norm": 3.2388575077056885, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8869572877883911, + "num_tokens": 788066381.0, + "step": 20307 + }, + { + "epoch": 2.5833863376160795, + "grad_norm": 2.8191561698913574, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9065120220184326, + "num_tokens": 788103107.0, + "step": 20308 + }, + { + "epoch": 2.58351354789467, + "grad_norm": 2.887051820755005, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8998599052429199, + "num_tokens": 788142414.0, + "step": 20309 + }, + { + "epoch": 2.5836407581732606, + "grad_norm": 2.7732999324798584, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9053242206573486, + "num_tokens": 788183926.0, + "step": 20310 + }, + { + "epoch": 2.583767968451851, + "grad_norm": 2.947195291519165, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9043779373168945, + "num_tokens": 788221897.0, + "step": 20311 + }, + { + "epoch": 2.583895178730441, + "grad_norm": 2.845033884048462, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9044616222381592, + "num_tokens": 788262151.0, + "step": 20312 + }, + { + "epoch": 2.584022389009032, + "grad_norm": 2.8269433975219727, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8988698124885559, + "num_tokens": 788307179.0, + "step": 20313 + }, + { + "epoch": 2.5841495992876222, + "grad_norm": 2.860107421875, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.899789035320282, + "num_tokens": 788349785.0, + "step": 20314 + }, + { + "epoch": 2.584276809566213, + "grad_norm": 2.8865206241607666, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8933735489845276, + "num_tokens": 788388952.0, + "step": 20315 + }, + { + "epoch": 2.5844040198448033, + "grad_norm": 3.013707160949707, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9066030979156494, + "num_tokens": 788423914.0, + "step": 20316 + }, + { + "epoch": 2.584531230123394, + "grad_norm": 3.0210330486297607, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8963123559951782, + "num_tokens": 788464073.0, + "step": 20317 + }, + { + "epoch": 2.5846584404019843, + "grad_norm": 2.9515833854675293, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9075967073440552, + "num_tokens": 788499678.0, + "step": 20318 + }, + { + "epoch": 2.584785650680575, + "grad_norm": 2.9180362224578857, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9111752510070801, + "num_tokens": 788535312.0, + "step": 20319 + }, + { + "epoch": 2.5849128609591654, + "grad_norm": 3.075411558151245, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8985416293144226, + "num_tokens": 788572354.0, + "step": 20320 + }, + { + "epoch": 2.585040071237756, + "grad_norm": 2.723385810852051, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9083781838417053, + "num_tokens": 788618314.0, + "step": 20321 + }, + { + "epoch": 2.5851672815163464, + "grad_norm": 3.270486354827881, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9024221897125244, + "num_tokens": 788651800.0, + "step": 20322 + }, + { + "epoch": 2.585294491794937, + "grad_norm": 2.843620777130127, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8906608819961548, + "num_tokens": 788692991.0, + "step": 20323 + }, + { + "epoch": 2.5854217020735275, + "grad_norm": 2.6933772563934326, + "learning_rate": 1e-06, + "loss": 0.2273, + "mean_token_accuracy": 0.9145826101303101, + "num_tokens": 788733089.0, + "step": 20324 + }, + { + "epoch": 2.585548912352118, + "grad_norm": 3.0216944217681885, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9042986035346985, + "num_tokens": 788768203.0, + "step": 20325 + }, + { + "epoch": 2.5856761226307086, + "grad_norm": 2.9666595458984375, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9012452363967896, + "num_tokens": 788805636.0, + "step": 20326 + }, + { + "epoch": 2.585803332909299, + "grad_norm": 3.2131545543670654, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.8969785571098328, + "num_tokens": 788839662.0, + "step": 20327 + }, + { + "epoch": 2.5859305431878896, + "grad_norm": 2.615029811859131, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.9031078815460205, + "num_tokens": 788885296.0, + "step": 20328 + }, + { + "epoch": 2.58605775346648, + "grad_norm": 2.962336778640747, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.902313232421875, + "num_tokens": 788924145.0, + "step": 20329 + }, + { + "epoch": 2.5861849637450707, + "grad_norm": 2.683797836303711, + "learning_rate": 1e-06, + "loss": 0.2151, + "mean_token_accuracy": 0.9176607131958008, + "num_tokens": 788961124.0, + "step": 20330 + }, + { + "epoch": 2.586312174023661, + "grad_norm": 2.9544763565063477, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9071192145347595, + "num_tokens": 789000197.0, + "step": 20331 + }, + { + "epoch": 2.5864393843022517, + "grad_norm": 2.7010934352874756, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9062237739562988, + "num_tokens": 789044224.0, + "step": 20332 + }, + { + "epoch": 2.5865665945808423, + "grad_norm": 2.550516128540039, + "learning_rate": 1e-06, + "loss": 0.2247, + "mean_token_accuracy": 0.9154669642448425, + "num_tokens": 789084237.0, + "step": 20333 + }, + { + "epoch": 2.586693804859433, + "grad_norm": 2.895698070526123, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8963729739189148, + "num_tokens": 789129168.0, + "step": 20334 + }, + { + "epoch": 2.5868210151380233, + "grad_norm": 2.9011247158050537, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9053122997283936, + "num_tokens": 789172103.0, + "step": 20335 + }, + { + "epoch": 2.586948225416614, + "grad_norm": 2.976269483566284, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.902389407157898, + "num_tokens": 789207273.0, + "step": 20336 + }, + { + "epoch": 2.587075435695204, + "grad_norm": 2.622004747390747, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9051576256752014, + "num_tokens": 789252608.0, + "step": 20337 + }, + { + "epoch": 2.587202645973795, + "grad_norm": 2.618647813796997, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9079077243804932, + "num_tokens": 789296302.0, + "step": 20338 + }, + { + "epoch": 2.587329856252385, + "grad_norm": 3.00189208984375, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9132879972457886, + "num_tokens": 789330978.0, + "step": 20339 + }, + { + "epoch": 2.587457066530976, + "grad_norm": 2.7410569190979004, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9048179388046265, + "num_tokens": 789373176.0, + "step": 20340 + }, + { + "epoch": 2.587584276809566, + "grad_norm": 2.6784679889678955, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.908213198184967, + "num_tokens": 789412468.0, + "step": 20341 + }, + { + "epoch": 2.5877114870881566, + "grad_norm": 2.788149356842041, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.9088373184204102, + "num_tokens": 789450749.0, + "step": 20342 + }, + { + "epoch": 2.587838697366747, + "grad_norm": 2.8643836975097656, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.8844130635261536, + "num_tokens": 789497574.0, + "step": 20343 + }, + { + "epoch": 2.5879659076453376, + "grad_norm": 2.9428818225860596, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9075242280960083, + "num_tokens": 789536378.0, + "step": 20344 + }, + { + "epoch": 2.588093117923928, + "grad_norm": 3.002967596054077, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9099904298782349, + "num_tokens": 789569885.0, + "step": 20345 + }, + { + "epoch": 2.5882203282025187, + "grad_norm": 2.8821280002593994, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9025363326072693, + "num_tokens": 789609718.0, + "step": 20346 + }, + { + "epoch": 2.588347538481109, + "grad_norm": 2.650897741317749, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9042781591415405, + "num_tokens": 789652964.0, + "step": 20347 + }, + { + "epoch": 2.5884747487596997, + "grad_norm": 2.7465226650238037, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9059515595436096, + "num_tokens": 789692566.0, + "step": 20348 + }, + { + "epoch": 2.5886019590382903, + "grad_norm": 2.7200467586517334, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9163328409194946, + "num_tokens": 789730264.0, + "step": 20349 + }, + { + "epoch": 2.588729169316881, + "grad_norm": 2.9600400924682617, + "learning_rate": 1e-06, + "loss": 0.2254, + "mean_token_accuracy": 0.9158975481987, + "num_tokens": 789763964.0, + "step": 20350 + }, + { + "epoch": 2.5888563795954713, + "grad_norm": 2.9405269622802734, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8944363594055176, + "num_tokens": 789805461.0, + "step": 20351 + }, + { + "epoch": 2.588983589874062, + "grad_norm": 2.5624639987945557, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8952059149742126, + "num_tokens": 789851737.0, + "step": 20352 + }, + { + "epoch": 2.5891108001526524, + "grad_norm": 2.586313247680664, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9095423221588135, + "num_tokens": 789894724.0, + "step": 20353 + }, + { + "epoch": 2.589238010431243, + "grad_norm": 2.6929726600646973, + "learning_rate": 1e-06, + "loss": 0.2182, + "mean_token_accuracy": 0.9179619550704956, + "num_tokens": 789932017.0, + "step": 20354 + }, + { + "epoch": 2.5893652207098334, + "grad_norm": 2.570636034011841, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9081950187683105, + "num_tokens": 789977253.0, + "step": 20355 + }, + { + "epoch": 2.589492430988424, + "grad_norm": 2.5929911136627197, + "learning_rate": 1e-06, + "loss": 0.2682, + "mean_token_accuracy": 0.9011520743370056, + "num_tokens": 790025175.0, + "step": 20356 + }, + { + "epoch": 2.5896196412670145, + "grad_norm": 2.791391372680664, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9065244793891907, + "num_tokens": 790065206.0, + "step": 20357 + }, + { + "epoch": 2.589746851545605, + "grad_norm": 2.8794562816619873, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9111849069595337, + "num_tokens": 790102760.0, + "step": 20358 + }, + { + "epoch": 2.5898740618241956, + "grad_norm": 2.996659278869629, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9035730957984924, + "num_tokens": 790139222.0, + "step": 20359 + }, + { + "epoch": 2.5900012721027856, + "grad_norm": 2.783923625946045, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9091376662254333, + "num_tokens": 790181921.0, + "step": 20360 + }, + { + "epoch": 2.5901284823813766, + "grad_norm": 2.8516745567321777, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9080076813697815, + "num_tokens": 790222507.0, + "step": 20361 + }, + { + "epoch": 2.5902556926599667, + "grad_norm": 2.6950533390045166, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9071590900421143, + "num_tokens": 790265425.0, + "step": 20362 + }, + { + "epoch": 2.5903829029385577, + "grad_norm": 2.7161879539489746, + "learning_rate": 1e-06, + "loss": 0.2612, + "mean_token_accuracy": 0.903121829032898, + "num_tokens": 790307145.0, + "step": 20363 + }, + { + "epoch": 2.5905101132171477, + "grad_norm": 3.1398661136627197, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9040278196334839, + "num_tokens": 790340790.0, + "step": 20364 + }, + { + "epoch": 2.5906373234957387, + "grad_norm": 2.74706768989563, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9066081047058105, + "num_tokens": 790381030.0, + "step": 20365 + }, + { + "epoch": 2.590764533774329, + "grad_norm": 2.888740301132202, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9088858962059021, + "num_tokens": 790419771.0, + "step": 20366 + }, + { + "epoch": 2.5908917440529193, + "grad_norm": 3.0506699085235596, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9026203155517578, + "num_tokens": 790452806.0, + "step": 20367 + }, + { + "epoch": 2.59101895433151, + "grad_norm": 2.9440019130706787, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.890842854976654, + "num_tokens": 790496248.0, + "step": 20368 + }, + { + "epoch": 2.5911461646101004, + "grad_norm": 3.0217843055725098, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.900368332862854, + "num_tokens": 790530161.0, + "step": 20369 + }, + { + "epoch": 2.591273374888691, + "grad_norm": 2.783602714538574, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9040650129318237, + "num_tokens": 790569692.0, + "step": 20370 + }, + { + "epoch": 2.5914005851672814, + "grad_norm": 2.424947738647461, + "learning_rate": 1e-06, + "loss": 0.2156, + "mean_token_accuracy": 0.9195559024810791, + "num_tokens": 790611906.0, + "step": 20371 + }, + { + "epoch": 2.591527795445872, + "grad_norm": 2.9198756217956543, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8913378715515137, + "num_tokens": 790652831.0, + "step": 20372 + }, + { + "epoch": 2.5916550057244625, + "grad_norm": 2.6850550174713135, + "learning_rate": 1e-06, + "loss": 0.2206, + "mean_token_accuracy": 0.92002272605896, + "num_tokens": 790689906.0, + "step": 20373 + }, + { + "epoch": 2.591782216003053, + "grad_norm": 2.993549346923828, + "learning_rate": 1e-06, + "loss": 0.3112, + "mean_token_accuracy": 0.886332631111145, + "num_tokens": 790731377.0, + "step": 20374 + }, + { + "epoch": 2.5919094262816436, + "grad_norm": 2.857884407043457, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9094719886779785, + "num_tokens": 790768568.0, + "step": 20375 + }, + { + "epoch": 2.592036636560234, + "grad_norm": 2.742330551147461, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9026353359222412, + "num_tokens": 790808057.0, + "step": 20376 + }, + { + "epoch": 2.5921638468388246, + "grad_norm": 3.2540745735168457, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8961887955665588, + "num_tokens": 790842290.0, + "step": 20377 + }, + { + "epoch": 2.592291057117415, + "grad_norm": 2.9968762397766113, + "learning_rate": 1e-06, + "loss": 0.2822, + "mean_token_accuracy": 0.8982900381088257, + "num_tokens": 790884044.0, + "step": 20378 + }, + { + "epoch": 2.5924182673960057, + "grad_norm": 2.676927089691162, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9104470610618591, + "num_tokens": 790923865.0, + "step": 20379 + }, + { + "epoch": 2.592545477674596, + "grad_norm": 2.9625465869903564, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9082930088043213, + "num_tokens": 790959902.0, + "step": 20380 + }, + { + "epoch": 2.5926726879531867, + "grad_norm": 2.6522068977355957, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9023019075393677, + "num_tokens": 791005247.0, + "step": 20381 + }, + { + "epoch": 2.5927998982317773, + "grad_norm": 2.717604160308838, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.9088493585586548, + "num_tokens": 791048700.0, + "step": 20382 + }, + { + "epoch": 2.592927108510368, + "grad_norm": 3.2761712074279785, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9018300771713257, + "num_tokens": 791083037.0, + "step": 20383 + }, + { + "epoch": 2.5930543187889583, + "grad_norm": 2.9603147506713867, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9016469120979309, + "num_tokens": 791119060.0, + "step": 20384 + }, + { + "epoch": 2.5931815290675484, + "grad_norm": 2.9316494464874268, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9043363928794861, + "num_tokens": 791160625.0, + "step": 20385 + }, + { + "epoch": 2.5933087393461394, + "grad_norm": 3.103156805038452, + "learning_rate": 1e-06, + "loss": 0.2967, + "mean_token_accuracy": 0.8896667957305908, + "num_tokens": 791199221.0, + "step": 20386 + }, + { + "epoch": 2.5934359496247295, + "grad_norm": 2.8868911266326904, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9009720683097839, + "num_tokens": 791238014.0, + "step": 20387 + }, + { + "epoch": 2.5935631599033204, + "grad_norm": 2.9365572929382324, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.902680516242981, + "num_tokens": 791275193.0, + "step": 20388 + }, + { + "epoch": 2.5936903701819105, + "grad_norm": 2.6916449069976807, + "learning_rate": 1e-06, + "loss": 0.2234, + "mean_token_accuracy": 0.9172934889793396, + "num_tokens": 791314307.0, + "step": 20389 + }, + { + "epoch": 2.5938175804605015, + "grad_norm": 3.0138583183288574, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.909152626991272, + "num_tokens": 791347968.0, + "step": 20390 + }, + { + "epoch": 2.5939447907390916, + "grad_norm": 3.175258159637451, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.885940432548523, + "num_tokens": 791386685.0, + "step": 20391 + }, + { + "epoch": 2.594072001017682, + "grad_norm": 2.815812826156616, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9036821126937866, + "num_tokens": 791425753.0, + "step": 20392 + }, + { + "epoch": 2.5941992112962726, + "grad_norm": 3.0902395248413086, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9039703607559204, + "num_tokens": 791457910.0, + "step": 20393 + }, + { + "epoch": 2.594326421574863, + "grad_norm": 3.234588384628296, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8896083831787109, + "num_tokens": 791492112.0, + "step": 20394 + }, + { + "epoch": 2.5944536318534537, + "grad_norm": 2.962993860244751, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9076165556907654, + "num_tokens": 791526473.0, + "step": 20395 + }, + { + "epoch": 2.594580842132044, + "grad_norm": 2.826080322265625, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.905451238155365, + "num_tokens": 791564801.0, + "step": 20396 + }, + { + "epoch": 2.5947080524106347, + "grad_norm": 2.864917755126953, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9101899266242981, + "num_tokens": 791600554.0, + "step": 20397 + }, + { + "epoch": 2.5948352626892253, + "grad_norm": 2.671201467514038, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9092640280723572, + "num_tokens": 791645419.0, + "step": 20398 + }, + { + "epoch": 2.594962472967816, + "grad_norm": 2.9039578437805176, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8959497809410095, + "num_tokens": 791683430.0, + "step": 20399 + }, + { + "epoch": 2.5950896832464063, + "grad_norm": 2.7250592708587646, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9085866212844849, + "num_tokens": 791725238.0, + "step": 20400 + }, + { + "epoch": 2.595216893524997, + "grad_norm": 2.8301384449005127, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9058827757835388, + "num_tokens": 791766630.0, + "step": 20401 + }, + { + "epoch": 2.5953441038035874, + "grad_norm": 2.672658681869507, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9075853824615479, + "num_tokens": 791808902.0, + "step": 20402 + }, + { + "epoch": 2.595471314082178, + "grad_norm": 2.6739466190338135, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.9041739702224731, + "num_tokens": 791850332.0, + "step": 20403 + }, + { + "epoch": 2.5955985243607684, + "grad_norm": 3.2257704734802246, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9060423374176025, + "num_tokens": 791883938.0, + "step": 20404 + }, + { + "epoch": 2.595725734639359, + "grad_norm": 2.890080690383911, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9020792245864868, + "num_tokens": 791921359.0, + "step": 20405 + }, + { + "epoch": 2.5958529449179495, + "grad_norm": 2.880558967590332, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8939071893692017, + "num_tokens": 791962490.0, + "step": 20406 + }, + { + "epoch": 2.59598015519654, + "grad_norm": 3.052333116531372, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8995905518531799, + "num_tokens": 791999244.0, + "step": 20407 + }, + { + "epoch": 2.5961073654751305, + "grad_norm": 2.607976198196411, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9063096642494202, + "num_tokens": 792047550.0, + "step": 20408 + }, + { + "epoch": 2.596234575753721, + "grad_norm": 2.8177602291107178, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9073411226272583, + "num_tokens": 792081672.0, + "step": 20409 + }, + { + "epoch": 2.596361786032311, + "grad_norm": 2.921570062637329, + "learning_rate": 1e-06, + "loss": 0.3185, + "mean_token_accuracy": 0.8841071128845215, + "num_tokens": 792125703.0, + "step": 20410 + }, + { + "epoch": 2.596488996310902, + "grad_norm": 2.905184745788574, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9018117785453796, + "num_tokens": 792160992.0, + "step": 20411 + }, + { + "epoch": 2.596616206589492, + "grad_norm": 2.9204905033111572, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8924990892410278, + "num_tokens": 792203561.0, + "step": 20412 + }, + { + "epoch": 2.596743416868083, + "grad_norm": 2.848578929901123, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.8957338333129883, + "num_tokens": 792244287.0, + "step": 20413 + }, + { + "epoch": 2.5968706271466733, + "grad_norm": 2.8580689430236816, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9049676060676575, + "num_tokens": 792283086.0, + "step": 20414 + }, + { + "epoch": 2.596997837425264, + "grad_norm": 2.6548922061920166, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9105448126792908, + "num_tokens": 792325405.0, + "step": 20415 + }, + { + "epoch": 2.5971250477038543, + "grad_norm": 2.5985703468322754, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9048280715942383, + "num_tokens": 792373281.0, + "step": 20416 + }, + { + "epoch": 2.597252257982445, + "grad_norm": 2.938568115234375, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.8975260257720947, + "num_tokens": 792412497.0, + "step": 20417 + }, + { + "epoch": 2.5973794682610354, + "grad_norm": 2.532047748565674, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9088765382766724, + "num_tokens": 792459827.0, + "step": 20418 + }, + { + "epoch": 2.597506678539626, + "grad_norm": 2.8399767875671387, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.9021627902984619, + "num_tokens": 792501744.0, + "step": 20419 + }, + { + "epoch": 2.5976338888182164, + "grad_norm": 2.6934165954589844, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9087158441543579, + "num_tokens": 792541898.0, + "step": 20420 + }, + { + "epoch": 2.597761099096807, + "grad_norm": 2.865683078765869, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.8969879746437073, + "num_tokens": 792581455.0, + "step": 20421 + }, + { + "epoch": 2.5978883093753975, + "grad_norm": 2.8729543685913086, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8973488211631775, + "num_tokens": 792619588.0, + "step": 20422 + }, + { + "epoch": 2.598015519653988, + "grad_norm": 2.8075873851776123, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9080933332443237, + "num_tokens": 792661199.0, + "step": 20423 + }, + { + "epoch": 2.5981427299325786, + "grad_norm": 2.770195960998535, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9042124152183533, + "num_tokens": 792700367.0, + "step": 20424 + }, + { + "epoch": 2.598269940211169, + "grad_norm": 2.7588491439819336, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9019430875778198, + "num_tokens": 792742384.0, + "step": 20425 + }, + { + "epoch": 2.5983971504897596, + "grad_norm": 2.8677446842193604, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9084030985832214, + "num_tokens": 792778648.0, + "step": 20426 + }, + { + "epoch": 2.59852436076835, + "grad_norm": 2.5222833156585693, + "learning_rate": 1e-06, + "loss": 0.2235, + "mean_token_accuracy": 0.915702760219574, + "num_tokens": 792821870.0, + "step": 20427 + }, + { + "epoch": 2.5986515710469407, + "grad_norm": 2.8393445014953613, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.904813289642334, + "num_tokens": 792863765.0, + "step": 20428 + }, + { + "epoch": 2.598778781325531, + "grad_norm": 2.8493006229400635, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9030641317367554, + "num_tokens": 792904835.0, + "step": 20429 + }, + { + "epoch": 2.5989059916041217, + "grad_norm": 2.79687762260437, + "learning_rate": 1e-06, + "loss": 0.2242, + "mean_token_accuracy": 0.9151392579078674, + "num_tokens": 792943381.0, + "step": 20430 + }, + { + "epoch": 2.5990332018827123, + "grad_norm": 2.632274866104126, + "learning_rate": 1e-06, + "loss": 0.2301, + "mean_token_accuracy": 0.9132772088050842, + "num_tokens": 792986383.0, + "step": 20431 + }, + { + "epoch": 2.599160412161303, + "grad_norm": 2.9696176052093506, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.906602680683136, + "num_tokens": 793025118.0, + "step": 20432 + }, + { + "epoch": 2.5992876224398933, + "grad_norm": 2.8905718326568604, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9008709788322449, + "num_tokens": 793065656.0, + "step": 20433 + }, + { + "epoch": 2.599414832718484, + "grad_norm": 3.032381534576416, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.8998744487762451, + "num_tokens": 793104517.0, + "step": 20434 + }, + { + "epoch": 2.599542042997074, + "grad_norm": 2.8697190284729004, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9070174098014832, + "num_tokens": 793142764.0, + "step": 20435 + }, + { + "epoch": 2.599669253275665, + "grad_norm": 3.1487441062927246, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.899552583694458, + "num_tokens": 793180951.0, + "step": 20436 + }, + { + "epoch": 2.599796463554255, + "grad_norm": 3.2520158290863037, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8916572332382202, + "num_tokens": 793215563.0, + "step": 20437 + }, + { + "epoch": 2.599923673832846, + "grad_norm": 2.4735324382781982, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9124934673309326, + "num_tokens": 793260053.0, + "step": 20438 + }, + { + "epoch": 2.600050884111436, + "grad_norm": 2.8112473487854004, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9102851748466492, + "num_tokens": 793301163.0, + "step": 20439 + }, + { + "epoch": 2.6001780943900266, + "grad_norm": 3.0919694900512695, + "learning_rate": 1e-06, + "loss": 0.2351, + "mean_token_accuracy": 0.91408371925354, + "num_tokens": 793335779.0, + "step": 20440 + }, + { + "epoch": 2.600305304668617, + "grad_norm": 2.8320200443267822, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.9012815952301025, + "num_tokens": 793372238.0, + "step": 20441 + }, + { + "epoch": 2.6004325149472076, + "grad_norm": 3.1445295810699463, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9040499925613403, + "num_tokens": 793407642.0, + "step": 20442 + }, + { + "epoch": 2.600559725225798, + "grad_norm": 2.7041804790496826, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9084524512290955, + "num_tokens": 793449807.0, + "step": 20443 + }, + { + "epoch": 2.6006869355043887, + "grad_norm": 3.0288350582122803, + "learning_rate": 1e-06, + "loss": 0.2976, + "mean_token_accuracy": 0.8919187784194946, + "num_tokens": 793488622.0, + "step": 20444 + }, + { + "epoch": 2.600814145782979, + "grad_norm": 2.9729316234588623, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9079582691192627, + "num_tokens": 793524617.0, + "step": 20445 + }, + { + "epoch": 2.6009413560615697, + "grad_norm": 2.651916980743408, + "learning_rate": 1e-06, + "loss": 0.2382, + "mean_token_accuracy": 0.9111961126327515, + "num_tokens": 793565634.0, + "step": 20446 + }, + { + "epoch": 2.6010685663401603, + "grad_norm": 2.8567073345184326, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9042418003082275, + "num_tokens": 793603775.0, + "step": 20447 + }, + { + "epoch": 2.601195776618751, + "grad_norm": 2.8293697834014893, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9077445268630981, + "num_tokens": 793640781.0, + "step": 20448 + }, + { + "epoch": 2.6013229868973413, + "grad_norm": 2.7291598320007324, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9119024276733398, + "num_tokens": 793678557.0, + "step": 20449 + }, + { + "epoch": 2.601450197175932, + "grad_norm": 2.5205612182617188, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9092028737068176, + "num_tokens": 793724244.0, + "step": 20450 + }, + { + "epoch": 2.6015774074545224, + "grad_norm": 2.994900703430176, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8925966024398804, + "num_tokens": 793764134.0, + "step": 20451 + }, + { + "epoch": 2.601704617733113, + "grad_norm": 2.809523582458496, + "learning_rate": 1e-06, + "loss": 0.2364, + "mean_token_accuracy": 0.9135095477104187, + "num_tokens": 793801825.0, + "step": 20452 + }, + { + "epoch": 2.6018318280117034, + "grad_norm": 2.967043161392212, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8920274376869202, + "num_tokens": 793842799.0, + "step": 20453 + }, + { + "epoch": 2.601959038290294, + "grad_norm": 2.5551247596740723, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.9132185578346252, + "num_tokens": 793887341.0, + "step": 20454 + }, + { + "epoch": 2.6020862485688845, + "grad_norm": 3.026974678039551, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9077197313308716, + "num_tokens": 793921676.0, + "step": 20455 + }, + { + "epoch": 2.602213458847475, + "grad_norm": 2.9857089519500732, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8981918692588806, + "num_tokens": 793963467.0, + "step": 20456 + }, + { + "epoch": 2.6023406691260655, + "grad_norm": 2.8603017330169678, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8971270322799683, + "num_tokens": 794006331.0, + "step": 20457 + }, + { + "epoch": 2.6024678794046556, + "grad_norm": 2.9827589988708496, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8990888595581055, + "num_tokens": 794043987.0, + "step": 20458 + }, + { + "epoch": 2.6025950896832466, + "grad_norm": 2.9607927799224854, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9003804922103882, + "num_tokens": 794077606.0, + "step": 20459 + }, + { + "epoch": 2.6027222999618367, + "grad_norm": 2.7524259090423584, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.8997601866722107, + "num_tokens": 794118243.0, + "step": 20460 + }, + { + "epoch": 2.6028495102404277, + "grad_norm": 2.8458025455474854, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9079131484031677, + "num_tokens": 794156652.0, + "step": 20461 + }, + { + "epoch": 2.6029767205190177, + "grad_norm": 2.76462721824646, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.8993246555328369, + "num_tokens": 794199317.0, + "step": 20462 + }, + { + "epoch": 2.6031039307976087, + "grad_norm": 2.5968916416168213, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9050279855728149, + "num_tokens": 794243741.0, + "step": 20463 + }, + { + "epoch": 2.603231141076199, + "grad_norm": 2.915001630783081, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9006485342979431, + "num_tokens": 794282748.0, + "step": 20464 + }, + { + "epoch": 2.6033583513547893, + "grad_norm": 2.952122688293457, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9064213037490845, + "num_tokens": 794319351.0, + "step": 20465 + }, + { + "epoch": 2.60348556163338, + "grad_norm": 2.924924373626709, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9056769609451294, + "num_tokens": 794356698.0, + "step": 20466 + }, + { + "epoch": 2.6036127719119704, + "grad_norm": 2.770514488220215, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.9150575399398804, + "num_tokens": 794392082.0, + "step": 20467 + }, + { + "epoch": 2.603739982190561, + "grad_norm": 2.7520999908447266, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9072299003601074, + "num_tokens": 794434230.0, + "step": 20468 + }, + { + "epoch": 2.6038671924691514, + "grad_norm": 2.83319354057312, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9109437465667725, + "num_tokens": 794472155.0, + "step": 20469 + }, + { + "epoch": 2.603994402747742, + "grad_norm": 2.9891488552093506, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9027414321899414, + "num_tokens": 794513730.0, + "step": 20470 + }, + { + "epoch": 2.6041216130263325, + "grad_norm": 2.9497413635253906, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8991802334785461, + "num_tokens": 794554195.0, + "step": 20471 + }, + { + "epoch": 2.604248823304923, + "grad_norm": 3.117731809616089, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9052057862281799, + "num_tokens": 794588520.0, + "step": 20472 + }, + { + "epoch": 2.6043760335835135, + "grad_norm": 2.8064188957214355, + "learning_rate": 1e-06, + "loss": 0.232, + "mean_token_accuracy": 0.9131515026092529, + "num_tokens": 794630196.0, + "step": 20473 + }, + { + "epoch": 2.604503243862104, + "grad_norm": 2.8214657306671143, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9067551493644714, + "num_tokens": 794670303.0, + "step": 20474 + }, + { + "epoch": 2.6046304541406946, + "grad_norm": 2.799576997756958, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.903318464756012, + "num_tokens": 794707114.0, + "step": 20475 + }, + { + "epoch": 2.604757664419285, + "grad_norm": 3.089662551879883, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.897705078125, + "num_tokens": 794743712.0, + "step": 20476 + }, + { + "epoch": 2.6048848746978757, + "grad_norm": 2.7210466861724854, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9029955863952637, + "num_tokens": 794784811.0, + "step": 20477 + }, + { + "epoch": 2.605012084976466, + "grad_norm": 2.9775097370147705, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.885977029800415, + "num_tokens": 794826217.0, + "step": 20478 + }, + { + "epoch": 2.6051392952550567, + "grad_norm": 2.6566689014434814, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9108818769454956, + "num_tokens": 794868088.0, + "step": 20479 + }, + { + "epoch": 2.6052665055336472, + "grad_norm": 2.9839940071105957, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.905644953250885, + "num_tokens": 794904965.0, + "step": 20480 + }, + { + "epoch": 2.6053937158122378, + "grad_norm": 2.7854959964752197, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9105373024940491, + "num_tokens": 794942061.0, + "step": 20481 + }, + { + "epoch": 2.6055209260908283, + "grad_norm": 2.7286267280578613, + "learning_rate": 1e-06, + "loss": 0.2273, + "mean_token_accuracy": 0.9149249792098999, + "num_tokens": 794980252.0, + "step": 20482 + }, + { + "epoch": 2.6056481363694184, + "grad_norm": 3.204841136932373, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.889854907989502, + "num_tokens": 795015546.0, + "step": 20483 + }, + { + "epoch": 2.6057753466480094, + "grad_norm": 2.9826269149780273, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9066564440727234, + "num_tokens": 795051348.0, + "step": 20484 + }, + { + "epoch": 2.6059025569265994, + "grad_norm": 2.95805287361145, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9058471918106079, + "num_tokens": 795089871.0, + "step": 20485 + }, + { + "epoch": 2.6060297672051904, + "grad_norm": 3.037132740020752, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.8997416496276855, + "num_tokens": 795124094.0, + "step": 20486 + }, + { + "epoch": 2.6061569774837805, + "grad_norm": 2.911834716796875, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9004734754562378, + "num_tokens": 795164151.0, + "step": 20487 + }, + { + "epoch": 2.6062841877623715, + "grad_norm": 3.000974416732788, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9020529985427856, + "num_tokens": 795203422.0, + "step": 20488 + }, + { + "epoch": 2.6064113980409616, + "grad_norm": 2.8810648918151855, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9020570516586304, + "num_tokens": 795241968.0, + "step": 20489 + }, + { + "epoch": 2.606538608319552, + "grad_norm": 2.8475852012634277, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9026967287063599, + "num_tokens": 795284231.0, + "step": 20490 + }, + { + "epoch": 2.6066658185981426, + "grad_norm": 3.0459911823272705, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9052910804748535, + "num_tokens": 795317639.0, + "step": 20491 + }, + { + "epoch": 2.606793028876733, + "grad_norm": 2.839813470840454, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9129334688186646, + "num_tokens": 795355823.0, + "step": 20492 + }, + { + "epoch": 2.6069202391553237, + "grad_norm": 3.091872453689575, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9080566167831421, + "num_tokens": 795386449.0, + "step": 20493 + }, + { + "epoch": 2.607047449433914, + "grad_norm": 2.57287859916687, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.9118319153785706, + "num_tokens": 795431433.0, + "step": 20494 + }, + { + "epoch": 2.6071746597125047, + "grad_norm": 3.1456215381622314, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.8971767425537109, + "num_tokens": 795465680.0, + "step": 20495 + }, + { + "epoch": 2.6073018699910953, + "grad_norm": 2.8203346729278564, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9026167392730713, + "num_tokens": 795511986.0, + "step": 20496 + }, + { + "epoch": 2.607429080269686, + "grad_norm": 3.112058162689209, + "learning_rate": 1e-06, + "loss": 0.2925, + "mean_token_accuracy": 0.8957581520080566, + "num_tokens": 795550465.0, + "step": 20497 + }, + { + "epoch": 2.6075562905482763, + "grad_norm": 2.6421475410461426, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.8987026810646057, + "num_tokens": 795595914.0, + "step": 20498 + }, + { + "epoch": 2.607683500826867, + "grad_norm": 2.8158910274505615, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9036017656326294, + "num_tokens": 795635070.0, + "step": 20499 + }, + { + "epoch": 2.6078107111054574, + "grad_norm": 2.8034844398498535, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.9010673761367798, + "num_tokens": 795677288.0, + "step": 20500 + }, + { + "epoch": 2.607937921384048, + "grad_norm": 2.498744010925293, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9064019322395325, + "num_tokens": 795724029.0, + "step": 20501 + }, + { + "epoch": 2.6080651316626384, + "grad_norm": 3.0471811294555664, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.8966164588928223, + "num_tokens": 795762288.0, + "step": 20502 + }, + { + "epoch": 2.608192341941229, + "grad_norm": 2.911421298980713, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8961467742919922, + "num_tokens": 795804206.0, + "step": 20503 + }, + { + "epoch": 2.6083195522198195, + "grad_norm": 2.832608461380005, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.8988702893257141, + "num_tokens": 795845596.0, + "step": 20504 + }, + { + "epoch": 2.60844676249841, + "grad_norm": 2.954533576965332, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9039884805679321, + "num_tokens": 795883979.0, + "step": 20505 + }, + { + "epoch": 2.6085739727770005, + "grad_norm": 2.6604795455932617, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9058971405029297, + "num_tokens": 795928946.0, + "step": 20506 + }, + { + "epoch": 2.608701183055591, + "grad_norm": 2.8407769203186035, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9046138525009155, + "num_tokens": 795969133.0, + "step": 20507 + }, + { + "epoch": 2.608828393334181, + "grad_norm": 3.3817288875579834, + "learning_rate": 1e-06, + "loss": 0.3007, + "mean_token_accuracy": 0.8925087451934814, + "num_tokens": 796007081.0, + "step": 20508 + }, + { + "epoch": 2.608955603612772, + "grad_norm": 2.789853096008301, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9004378914833069, + "num_tokens": 796046140.0, + "step": 20509 + }, + { + "epoch": 2.609082813891362, + "grad_norm": 2.997011661529541, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8940739035606384, + "num_tokens": 796087225.0, + "step": 20510 + }, + { + "epoch": 2.609210024169953, + "grad_norm": 2.6325201988220215, + "learning_rate": 1e-06, + "loss": 0.2301, + "mean_token_accuracy": 0.9156938791275024, + "num_tokens": 796125557.0, + "step": 20511 + }, + { + "epoch": 2.6093372344485433, + "grad_norm": 2.926435708999634, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.904836118221283, + "num_tokens": 796161536.0, + "step": 20512 + }, + { + "epoch": 2.609464444727134, + "grad_norm": 2.985513210296631, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9052349328994751, + "num_tokens": 796198731.0, + "step": 20513 + }, + { + "epoch": 2.6095916550057243, + "grad_norm": 2.8027772903442383, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9028125405311584, + "num_tokens": 796239153.0, + "step": 20514 + }, + { + "epoch": 2.609718865284315, + "grad_norm": 2.984522581100464, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9106207489967346, + "num_tokens": 796274423.0, + "step": 20515 + }, + { + "epoch": 2.6098460755629054, + "grad_norm": 2.845060348510742, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.902928352355957, + "num_tokens": 796312782.0, + "step": 20516 + }, + { + "epoch": 2.609973285841496, + "grad_norm": 2.810563087463379, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9076730012893677, + "num_tokens": 796352678.0, + "step": 20517 + }, + { + "epoch": 2.6101004961200864, + "grad_norm": 2.8019769191741943, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.903214693069458, + "num_tokens": 796392287.0, + "step": 20518 + }, + { + "epoch": 2.610227706398677, + "grad_norm": 2.759868860244751, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9148465991020203, + "num_tokens": 796431895.0, + "step": 20519 + }, + { + "epoch": 2.6103549166772675, + "grad_norm": 2.92561411857605, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9020734429359436, + "num_tokens": 796468462.0, + "step": 20520 + }, + { + "epoch": 2.610482126955858, + "grad_norm": 2.7913150787353516, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9062739014625549, + "num_tokens": 796510849.0, + "step": 20521 + }, + { + "epoch": 2.6106093372344485, + "grad_norm": 2.664483070373535, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9028490781784058, + "num_tokens": 796554615.0, + "step": 20522 + }, + { + "epoch": 2.610736547513039, + "grad_norm": 2.7400667667388916, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.8993054032325745, + "num_tokens": 796596019.0, + "step": 20523 + }, + { + "epoch": 2.6108637577916296, + "grad_norm": 3.1474967002868652, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9076917171478271, + "num_tokens": 796628266.0, + "step": 20524 + }, + { + "epoch": 2.61099096807022, + "grad_norm": 2.9143340587615967, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9010487794876099, + "num_tokens": 796668985.0, + "step": 20525 + }, + { + "epoch": 2.6111181783488107, + "grad_norm": 3.227548122406006, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8957414627075195, + "num_tokens": 796702955.0, + "step": 20526 + }, + { + "epoch": 2.611245388627401, + "grad_norm": 2.9122884273529053, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9096935987472534, + "num_tokens": 796742649.0, + "step": 20527 + }, + { + "epoch": 2.6113725989059917, + "grad_norm": 2.8468151092529297, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9087419509887695, + "num_tokens": 796778936.0, + "step": 20528 + }, + { + "epoch": 2.6114998091845822, + "grad_norm": 3.09189772605896, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.9008970260620117, + "num_tokens": 796813501.0, + "step": 20529 + }, + { + "epoch": 2.6116270194631728, + "grad_norm": 2.809776782989502, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9015024900436401, + "num_tokens": 796855227.0, + "step": 20530 + }, + { + "epoch": 2.6117542297417633, + "grad_norm": 3.0611138343811035, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9040412306785583, + "num_tokens": 796889123.0, + "step": 20531 + }, + { + "epoch": 2.611881440020354, + "grad_norm": 2.77594256401062, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9062547087669373, + "num_tokens": 796928573.0, + "step": 20532 + }, + { + "epoch": 2.612008650298944, + "grad_norm": 2.834235906600952, + "learning_rate": 1e-06, + "loss": 0.2331, + "mean_token_accuracy": 0.9130861163139343, + "num_tokens": 796964042.0, + "step": 20533 + }, + { + "epoch": 2.612135860577535, + "grad_norm": 2.9132392406463623, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.9134241342544556, + "num_tokens": 796998108.0, + "step": 20534 + }, + { + "epoch": 2.612263070856125, + "grad_norm": 2.949373483657837, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9092075824737549, + "num_tokens": 797033720.0, + "step": 20535 + }, + { + "epoch": 2.612390281134716, + "grad_norm": 2.9678637981414795, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9037848711013794, + "num_tokens": 797076477.0, + "step": 20536 + }, + { + "epoch": 2.612517491413306, + "grad_norm": 3.162123441696167, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9014825224876404, + "num_tokens": 797108590.0, + "step": 20537 + }, + { + "epoch": 2.6126447016918966, + "grad_norm": 3.087636709213257, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9057286381721497, + "num_tokens": 797141840.0, + "step": 20538 + }, + { + "epoch": 2.612771911970487, + "grad_norm": 3.3041787147521973, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8950290679931641, + "num_tokens": 797173997.0, + "step": 20539 + }, + { + "epoch": 2.6128991222490776, + "grad_norm": 2.699709177017212, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9112311601638794, + "num_tokens": 797213888.0, + "step": 20540 + }, + { + "epoch": 2.613026332527668, + "grad_norm": 2.6558709144592285, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9142928123474121, + "num_tokens": 797255045.0, + "step": 20541 + }, + { + "epoch": 2.6131535428062587, + "grad_norm": 2.850060224533081, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8955650329589844, + "num_tokens": 797299706.0, + "step": 20542 + }, + { + "epoch": 2.613280753084849, + "grad_norm": 2.78910231590271, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9021192193031311, + "num_tokens": 797343006.0, + "step": 20543 + }, + { + "epoch": 2.6134079633634397, + "grad_norm": 2.859225273132324, + "learning_rate": 1e-06, + "loss": 0.2987, + "mean_token_accuracy": 0.8895633220672607, + "num_tokens": 797387529.0, + "step": 20544 + }, + { + "epoch": 2.6135351736420303, + "grad_norm": 2.847316026687622, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9032211899757385, + "num_tokens": 797425550.0, + "step": 20545 + }, + { + "epoch": 2.613662383920621, + "grad_norm": 2.7949554920196533, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.9086235761642456, + "num_tokens": 797463726.0, + "step": 20546 + }, + { + "epoch": 2.6137895941992113, + "grad_norm": 2.752873420715332, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.9175491333007812, + "num_tokens": 797500002.0, + "step": 20547 + }, + { + "epoch": 2.613916804477802, + "grad_norm": 2.7102839946746826, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9007395505905151, + "num_tokens": 797541936.0, + "step": 20548 + }, + { + "epoch": 2.6140440147563924, + "grad_norm": 2.782602548599243, + "learning_rate": 1e-06, + "loss": 0.2284, + "mean_token_accuracy": 0.9155552387237549, + "num_tokens": 797577896.0, + "step": 20549 + }, + { + "epoch": 2.614171225034983, + "grad_norm": 2.8549587726593018, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9084506034851074, + "num_tokens": 797617340.0, + "step": 20550 + }, + { + "epoch": 2.6142984353135734, + "grad_norm": 2.9808342456817627, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9049859046936035, + "num_tokens": 797653822.0, + "step": 20551 + }, + { + "epoch": 2.614425645592164, + "grad_norm": 2.7111976146698, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.9108935594558716, + "num_tokens": 797690709.0, + "step": 20552 + }, + { + "epoch": 2.6145528558707545, + "grad_norm": 2.614611864089966, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.8997195959091187, + "num_tokens": 797738075.0, + "step": 20553 + }, + { + "epoch": 2.614680066149345, + "grad_norm": 2.8193440437316895, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9103735685348511, + "num_tokens": 797774102.0, + "step": 20554 + }, + { + "epoch": 2.6148072764279355, + "grad_norm": 2.940044641494751, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8961151838302612, + "num_tokens": 797811179.0, + "step": 20555 + }, + { + "epoch": 2.6149344867065256, + "grad_norm": 2.974930763244629, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9005711078643799, + "num_tokens": 797848369.0, + "step": 20556 + }, + { + "epoch": 2.6150616969851166, + "grad_norm": 2.8195314407348633, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9057338237762451, + "num_tokens": 797887296.0, + "step": 20557 + }, + { + "epoch": 2.6151889072637067, + "grad_norm": 2.971477746963501, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8966277837753296, + "num_tokens": 797927125.0, + "step": 20558 + }, + { + "epoch": 2.6153161175422976, + "grad_norm": 2.958801746368408, + "learning_rate": 1e-06, + "loss": 0.2416, + "mean_token_accuracy": 0.9095232486724854, + "num_tokens": 797961318.0, + "step": 20559 + }, + { + "epoch": 2.6154433278208877, + "grad_norm": 2.745897054672241, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9126873016357422, + "num_tokens": 797999775.0, + "step": 20560 + }, + { + "epoch": 2.6155705380994787, + "grad_norm": 3.187223196029663, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9066731333732605, + "num_tokens": 798035052.0, + "step": 20561 + }, + { + "epoch": 2.615697748378069, + "grad_norm": 2.8710215091705322, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9056183695793152, + "num_tokens": 798075372.0, + "step": 20562 + }, + { + "epoch": 2.6158249586566593, + "grad_norm": 2.845857858657837, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.89662766456604, + "num_tokens": 798120073.0, + "step": 20563 + }, + { + "epoch": 2.61595216893525, + "grad_norm": 2.8721771240234375, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9008668661117554, + "num_tokens": 798165586.0, + "step": 20564 + }, + { + "epoch": 2.6160793792138404, + "grad_norm": 2.855985164642334, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.905616283416748, + "num_tokens": 798203279.0, + "step": 20565 + }, + { + "epoch": 2.616206589492431, + "grad_norm": 2.8350107669830322, + "learning_rate": 1e-06, + "loss": 0.2393, + "mean_token_accuracy": 0.9100738763809204, + "num_tokens": 798244095.0, + "step": 20566 + }, + { + "epoch": 2.6163337997710214, + "grad_norm": 2.7355587482452393, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9043092131614685, + "num_tokens": 798284877.0, + "step": 20567 + }, + { + "epoch": 2.616461010049612, + "grad_norm": 2.9690730571746826, + "learning_rate": 1e-06, + "loss": 0.2443, + "mean_token_accuracy": 0.908685564994812, + "num_tokens": 798319886.0, + "step": 20568 + }, + { + "epoch": 2.6165882203282025, + "grad_norm": 2.7406039237976074, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9035016298294067, + "num_tokens": 798361616.0, + "step": 20569 + }, + { + "epoch": 2.616715430606793, + "grad_norm": 2.830983877182007, + "learning_rate": 1e-06, + "loss": 0.2348, + "mean_token_accuracy": 0.9100728034973145, + "num_tokens": 798400163.0, + "step": 20570 + }, + { + "epoch": 2.6168426408853835, + "grad_norm": 2.7865824699401855, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.898745059967041, + "num_tokens": 798443958.0, + "step": 20571 + }, + { + "epoch": 2.616969851163974, + "grad_norm": 2.743257522583008, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9103974103927612, + "num_tokens": 798485376.0, + "step": 20572 + }, + { + "epoch": 2.6170970614425646, + "grad_norm": 2.8462822437286377, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9036803245544434, + "num_tokens": 798525451.0, + "step": 20573 + }, + { + "epoch": 2.617224271721155, + "grad_norm": 3.014380931854248, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9029779434204102, + "num_tokens": 798562274.0, + "step": 20574 + }, + { + "epoch": 2.6173514819997457, + "grad_norm": 2.847302198410034, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9033434391021729, + "num_tokens": 798602991.0, + "step": 20575 + }, + { + "epoch": 2.617478692278336, + "grad_norm": 3.0402748584747314, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8966504335403442, + "num_tokens": 798639102.0, + "step": 20576 + }, + { + "epoch": 2.6176059025569267, + "grad_norm": 2.9864532947540283, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9135551452636719, + "num_tokens": 798672195.0, + "step": 20577 + }, + { + "epoch": 2.6177331128355172, + "grad_norm": 2.9898221492767334, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8981583118438721, + "num_tokens": 798713513.0, + "step": 20578 + }, + { + "epoch": 2.6178603231141078, + "grad_norm": 2.5301156044006348, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9085801243782043, + "num_tokens": 798757410.0, + "step": 20579 + }, + { + "epoch": 2.6179875333926983, + "grad_norm": 2.99975323677063, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8962217569351196, + "num_tokens": 798796981.0, + "step": 20580 + }, + { + "epoch": 2.6181147436712884, + "grad_norm": 2.7161495685577393, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9041820764541626, + "num_tokens": 798838530.0, + "step": 20581 + }, + { + "epoch": 2.6182419539498794, + "grad_norm": 3.038336753845215, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.899992823600769, + "num_tokens": 798875225.0, + "step": 20582 + }, + { + "epoch": 2.6183691642284694, + "grad_norm": 3.1320037841796875, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9024877548217773, + "num_tokens": 798906661.0, + "step": 20583 + }, + { + "epoch": 2.6184963745070604, + "grad_norm": 2.688837766647339, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9077776670455933, + "num_tokens": 798950542.0, + "step": 20584 + }, + { + "epoch": 2.6186235847856505, + "grad_norm": 2.8294553756713867, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9031544923782349, + "num_tokens": 798990546.0, + "step": 20585 + }, + { + "epoch": 2.6187507950642415, + "grad_norm": 2.808293581008911, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9028478860855103, + "num_tokens": 799028790.0, + "step": 20586 + }, + { + "epoch": 2.6188780053428315, + "grad_norm": 2.888810396194458, + "learning_rate": 1e-06, + "loss": 0.2438, + "mean_token_accuracy": 0.9104048609733582, + "num_tokens": 799064733.0, + "step": 20587 + }, + { + "epoch": 2.619005215621422, + "grad_norm": 3.1058197021484375, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9031731486320496, + "num_tokens": 799100901.0, + "step": 20588 + }, + { + "epoch": 2.6191324259000126, + "grad_norm": 2.802371025085449, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9033218026161194, + "num_tokens": 799143384.0, + "step": 20589 + }, + { + "epoch": 2.619259636178603, + "grad_norm": 2.9489147663116455, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8914276361465454, + "num_tokens": 799182724.0, + "step": 20590 + }, + { + "epoch": 2.6193868464571937, + "grad_norm": 2.723400831222534, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9026801586151123, + "num_tokens": 799225701.0, + "step": 20591 + }, + { + "epoch": 2.619514056735784, + "grad_norm": 2.641525983810425, + "learning_rate": 1e-06, + "loss": 0.2398, + "mean_token_accuracy": 0.9101453423500061, + "num_tokens": 799267993.0, + "step": 20592 + }, + { + "epoch": 2.6196412670143747, + "grad_norm": 3.0282065868377686, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8909185528755188, + "num_tokens": 799305158.0, + "step": 20593 + }, + { + "epoch": 2.6197684772929652, + "grad_norm": 3.1927433013916016, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8869481682777405, + "num_tokens": 799345231.0, + "step": 20594 + }, + { + "epoch": 2.6198956875715558, + "grad_norm": 3.05741548538208, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9087543487548828, + "num_tokens": 799379117.0, + "step": 20595 + }, + { + "epoch": 2.6200228978501463, + "grad_norm": 3.1337168216705322, + "learning_rate": 1e-06, + "loss": 0.3096, + "mean_token_accuracy": 0.8862388134002686, + "num_tokens": 799417462.0, + "step": 20596 + }, + { + "epoch": 2.620150108128737, + "grad_norm": 3.319551944732666, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8945759534835815, + "num_tokens": 799448847.0, + "step": 20597 + }, + { + "epoch": 2.6202773184073274, + "grad_norm": 3.173860549926758, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9011787176132202, + "num_tokens": 799479063.0, + "step": 20598 + }, + { + "epoch": 2.620404528685918, + "grad_norm": 2.8201310634613037, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9043253660202026, + "num_tokens": 799519441.0, + "step": 20599 + }, + { + "epoch": 2.6205317389645084, + "grad_norm": 2.9102210998535156, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9089287519454956, + "num_tokens": 799554717.0, + "step": 20600 + }, + { + "epoch": 2.620658949243099, + "grad_norm": 3.3517022132873535, + "learning_rate": 1e-06, + "loss": 0.3014, + "mean_token_accuracy": 0.8889252543449402, + "num_tokens": 799588720.0, + "step": 20601 + }, + { + "epoch": 2.6207861595216895, + "grad_norm": 2.8538384437561035, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9112468957901001, + "num_tokens": 799627395.0, + "step": 20602 + }, + { + "epoch": 2.62091336980028, + "grad_norm": 2.8332040309906006, + "learning_rate": 1e-06, + "loss": 0.2287, + "mean_token_accuracy": 0.9132432341575623, + "num_tokens": 799665909.0, + "step": 20603 + }, + { + "epoch": 2.6210405800788705, + "grad_norm": 2.747260332107544, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9109750986099243, + "num_tokens": 799702666.0, + "step": 20604 + }, + { + "epoch": 2.621167790357461, + "grad_norm": 2.8325724601745605, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.893051028251648, + "num_tokens": 799747988.0, + "step": 20605 + }, + { + "epoch": 2.621295000636051, + "grad_norm": 2.8312361240386963, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9105379581451416, + "num_tokens": 799784159.0, + "step": 20606 + }, + { + "epoch": 2.621422210914642, + "grad_norm": 2.873762845993042, + "learning_rate": 1e-06, + "loss": 0.2755, + "mean_token_accuracy": 0.9004137516021729, + "num_tokens": 799827624.0, + "step": 20607 + }, + { + "epoch": 2.621549421193232, + "grad_norm": 3.037160873413086, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.895399808883667, + "num_tokens": 799866878.0, + "step": 20608 + }, + { + "epoch": 2.621676631471823, + "grad_norm": 2.892348527908325, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.907752275466919, + "num_tokens": 799905753.0, + "step": 20609 + }, + { + "epoch": 2.6218038417504133, + "grad_norm": 2.958732843399048, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.8986367583274841, + "num_tokens": 799945866.0, + "step": 20610 + }, + { + "epoch": 2.621931052029004, + "grad_norm": 2.6424295902252197, + "learning_rate": 1e-06, + "loss": 0.2268, + "mean_token_accuracy": 0.9159790277481079, + "num_tokens": 799984931.0, + "step": 20611 + }, + { + "epoch": 2.6220582623075943, + "grad_norm": 2.6992874145507812, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.909299373626709, + "num_tokens": 800027989.0, + "step": 20612 + }, + { + "epoch": 2.622185472586185, + "grad_norm": 2.8308379650115967, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9052462577819824, + "num_tokens": 800063378.0, + "step": 20613 + }, + { + "epoch": 2.6223126828647754, + "grad_norm": 2.897081136703491, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9038453102111816, + "num_tokens": 800103358.0, + "step": 20614 + }, + { + "epoch": 2.622439893143366, + "grad_norm": 3.228789806365967, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8865338563919067, + "num_tokens": 800137859.0, + "step": 20615 + }, + { + "epoch": 2.6225671034219564, + "grad_norm": 2.9688527584075928, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8939145803451538, + "num_tokens": 800178308.0, + "step": 20616 + }, + { + "epoch": 2.622694313700547, + "grad_norm": 2.644894599914551, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9077359437942505, + "num_tokens": 800220290.0, + "step": 20617 + }, + { + "epoch": 2.6228215239791375, + "grad_norm": 2.782003164291382, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9089113473892212, + "num_tokens": 800259116.0, + "step": 20618 + }, + { + "epoch": 2.622948734257728, + "grad_norm": 2.9414706230163574, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9009580612182617, + "num_tokens": 800299818.0, + "step": 20619 + }, + { + "epoch": 2.6230759445363185, + "grad_norm": 3.2022831439971924, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9062361717224121, + "num_tokens": 800332814.0, + "step": 20620 + }, + { + "epoch": 2.623203154814909, + "grad_norm": 2.6770026683807373, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9081927537918091, + "num_tokens": 800374901.0, + "step": 20621 + }, + { + "epoch": 2.6233303650934996, + "grad_norm": 2.861384391784668, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9116610884666443, + "num_tokens": 800411630.0, + "step": 20622 + }, + { + "epoch": 2.62345757537209, + "grad_norm": 2.854562282562256, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9079214334487915, + "num_tokens": 800449717.0, + "step": 20623 + }, + { + "epoch": 2.6235847856506807, + "grad_norm": 2.9829163551330566, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.906901478767395, + "num_tokens": 800483491.0, + "step": 20624 + }, + { + "epoch": 2.623711995929271, + "grad_norm": 2.859847068786621, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9083284735679626, + "num_tokens": 800520805.0, + "step": 20625 + }, + { + "epoch": 2.6238392062078617, + "grad_norm": 3.122497081756592, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9062426090240479, + "num_tokens": 800556209.0, + "step": 20626 + }, + { + "epoch": 2.6239664164864522, + "grad_norm": 2.7359988689422607, + "learning_rate": 1e-06, + "loss": 0.2203, + "mean_token_accuracy": 0.9190092086791992, + "num_tokens": 800595212.0, + "step": 20627 + }, + { + "epoch": 2.6240936267650428, + "grad_norm": 2.7076478004455566, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9131864905357361, + "num_tokens": 800635364.0, + "step": 20628 + }, + { + "epoch": 2.6242208370436333, + "grad_norm": 2.738821268081665, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9139212965965271, + "num_tokens": 800674042.0, + "step": 20629 + }, + { + "epoch": 2.624348047322224, + "grad_norm": 3.1131327152252197, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9051811695098877, + "num_tokens": 800711452.0, + "step": 20630 + }, + { + "epoch": 2.624475257600814, + "grad_norm": 2.5313148498535156, + "learning_rate": 1e-06, + "loss": 0.218, + "mean_token_accuracy": 0.9180073738098145, + "num_tokens": 800753049.0, + "step": 20631 + }, + { + "epoch": 2.624602467879405, + "grad_norm": 2.990178346633911, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9040688276290894, + "num_tokens": 800784961.0, + "step": 20632 + }, + { + "epoch": 2.624729678157995, + "grad_norm": 2.934633493423462, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9012739658355713, + "num_tokens": 800822539.0, + "step": 20633 + }, + { + "epoch": 2.624856888436586, + "grad_norm": 2.6559481620788574, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.8978214263916016, + "num_tokens": 800871081.0, + "step": 20634 + }, + { + "epoch": 2.624984098715176, + "grad_norm": 2.807643175125122, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9047520160675049, + "num_tokens": 800911307.0, + "step": 20635 + }, + { + "epoch": 2.6251113089937665, + "grad_norm": 3.058608055114746, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9011071920394897, + "num_tokens": 800946743.0, + "step": 20636 + }, + { + "epoch": 2.625238519272357, + "grad_norm": 2.8800976276397705, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9123754501342773, + "num_tokens": 800984487.0, + "step": 20637 + }, + { + "epoch": 2.6253657295509476, + "grad_norm": 2.844512939453125, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9114312529563904, + "num_tokens": 801022840.0, + "step": 20638 + }, + { + "epoch": 2.625492939829538, + "grad_norm": 2.893404483795166, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9060924053192139, + "num_tokens": 801061070.0, + "step": 20639 + }, + { + "epoch": 2.6256201501081287, + "grad_norm": 2.8704891204833984, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9009121656417847, + "num_tokens": 801105558.0, + "step": 20640 + }, + { + "epoch": 2.625747360386719, + "grad_norm": 2.716862440109253, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9077996015548706, + "num_tokens": 801149222.0, + "step": 20641 + }, + { + "epoch": 2.6258745706653097, + "grad_norm": 2.971217393875122, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9076181054115295, + "num_tokens": 801187140.0, + "step": 20642 + }, + { + "epoch": 2.6260017809439002, + "grad_norm": 2.7678215503692627, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8959858417510986, + "num_tokens": 801232458.0, + "step": 20643 + }, + { + "epoch": 2.6261289912224908, + "grad_norm": 2.7635691165924072, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9016333818435669, + "num_tokens": 801272166.0, + "step": 20644 + }, + { + "epoch": 2.6262562015010813, + "grad_norm": 2.88450288772583, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.8994332551956177, + "num_tokens": 801312337.0, + "step": 20645 + }, + { + "epoch": 2.626383411779672, + "grad_norm": 2.947695732116699, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8956820368766785, + "num_tokens": 801353506.0, + "step": 20646 + }, + { + "epoch": 2.6265106220582624, + "grad_norm": 2.868335485458374, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9003037214279175, + "num_tokens": 801395179.0, + "step": 20647 + }, + { + "epoch": 2.626637832336853, + "grad_norm": 2.71269154548645, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9055526852607727, + "num_tokens": 801433550.0, + "step": 20648 + }, + { + "epoch": 2.6267650426154434, + "grad_norm": 2.7011377811431885, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9074626564979553, + "num_tokens": 801472704.0, + "step": 20649 + }, + { + "epoch": 2.626892252894034, + "grad_norm": 3.22546124458313, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9008662700653076, + "num_tokens": 801508374.0, + "step": 20650 + }, + { + "epoch": 2.6270194631726245, + "grad_norm": 2.8288915157318115, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9048337936401367, + "num_tokens": 801548307.0, + "step": 20651 + }, + { + "epoch": 2.627146673451215, + "grad_norm": 2.9483494758605957, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9091250896453857, + "num_tokens": 801584187.0, + "step": 20652 + }, + { + "epoch": 2.6272738837298055, + "grad_norm": 3.4143292903900146, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8998085260391235, + "num_tokens": 801616466.0, + "step": 20653 + }, + { + "epoch": 2.6274010940083956, + "grad_norm": 2.7493467330932617, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9064004421234131, + "num_tokens": 801660645.0, + "step": 20654 + }, + { + "epoch": 2.6275283042869866, + "grad_norm": 3.1187777519226074, + "learning_rate": 1e-06, + "loss": 0.299, + "mean_token_accuracy": 0.8937849998474121, + "num_tokens": 801699381.0, + "step": 20655 + }, + { + "epoch": 2.6276555145655767, + "grad_norm": 2.732860565185547, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9117817878723145, + "num_tokens": 801736851.0, + "step": 20656 + }, + { + "epoch": 2.6277827248441676, + "grad_norm": 2.7074854373931885, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9002045392990112, + "num_tokens": 801780646.0, + "step": 20657 + }, + { + "epoch": 2.6279099351227577, + "grad_norm": 3.200162887573242, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8998550772666931, + "num_tokens": 801818248.0, + "step": 20658 + }, + { + "epoch": 2.6280371454013487, + "grad_norm": 2.817361354827881, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9072731733322144, + "num_tokens": 801857828.0, + "step": 20659 + }, + { + "epoch": 2.628164355679939, + "grad_norm": 2.854623317718506, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9104399681091309, + "num_tokens": 801893140.0, + "step": 20660 + }, + { + "epoch": 2.6282915659585293, + "grad_norm": 3.218520164489746, + "learning_rate": 1e-06, + "loss": 0.294, + "mean_token_accuracy": 0.8974686861038208, + "num_tokens": 801933141.0, + "step": 20661 + }, + { + "epoch": 2.62841877623712, + "grad_norm": 2.9192941188812256, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9081500768661499, + "num_tokens": 801971444.0, + "step": 20662 + }, + { + "epoch": 2.6285459865157104, + "grad_norm": 3.014343738555908, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9076338410377502, + "num_tokens": 802004258.0, + "step": 20663 + }, + { + "epoch": 2.628673196794301, + "grad_norm": 2.962509870529175, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.896353006362915, + "num_tokens": 802038220.0, + "step": 20664 + }, + { + "epoch": 2.6288004070728914, + "grad_norm": 2.903834581375122, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9062862396240234, + "num_tokens": 802074693.0, + "step": 20665 + }, + { + "epoch": 2.628927617351482, + "grad_norm": 2.811211585998535, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9042887687683105, + "num_tokens": 802116483.0, + "step": 20666 + }, + { + "epoch": 2.6290548276300725, + "grad_norm": 3.1691086292266846, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9033963084220886, + "num_tokens": 802147128.0, + "step": 20667 + }, + { + "epoch": 2.629182037908663, + "grad_norm": 2.8990163803100586, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9047800302505493, + "num_tokens": 802184144.0, + "step": 20668 + }, + { + "epoch": 2.6293092481872535, + "grad_norm": 2.8695287704467773, + "learning_rate": 1e-06, + "loss": 0.2363, + "mean_token_accuracy": 0.9137021899223328, + "num_tokens": 802220155.0, + "step": 20669 + }, + { + "epoch": 2.629436458465844, + "grad_norm": 2.946143627166748, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9024561643600464, + "num_tokens": 802260049.0, + "step": 20670 + }, + { + "epoch": 2.6295636687444346, + "grad_norm": 2.7104721069335938, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.9007341861724854, + "num_tokens": 802302476.0, + "step": 20671 + }, + { + "epoch": 2.629690879023025, + "grad_norm": 2.784276247024536, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9014972448348999, + "num_tokens": 802342297.0, + "step": 20672 + }, + { + "epoch": 2.6298180893016156, + "grad_norm": 2.7452847957611084, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9095000624656677, + "num_tokens": 802379927.0, + "step": 20673 + }, + { + "epoch": 2.629945299580206, + "grad_norm": 4.108516216278076, + "learning_rate": 1e-06, + "loss": 0.2995, + "mean_token_accuracy": 0.8901969790458679, + "num_tokens": 802413151.0, + "step": 20674 + }, + { + "epoch": 2.6300725098587967, + "grad_norm": 3.0679738521575928, + "learning_rate": 1e-06, + "loss": 0.3057, + "mean_token_accuracy": 0.8913329839706421, + "num_tokens": 802452498.0, + "step": 20675 + }, + { + "epoch": 2.6301997201373872, + "grad_norm": 2.770341634750366, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9069631099700928, + "num_tokens": 802490982.0, + "step": 20676 + }, + { + "epoch": 2.6303269304159778, + "grad_norm": 2.858970880508423, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9018394351005554, + "num_tokens": 802528710.0, + "step": 20677 + }, + { + "epoch": 2.6304541406945683, + "grad_norm": 2.9327335357666016, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8968805074691772, + "num_tokens": 802570722.0, + "step": 20678 + }, + { + "epoch": 2.6305813509731584, + "grad_norm": 2.89825439453125, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.9107030034065247, + "num_tokens": 802607370.0, + "step": 20679 + }, + { + "epoch": 2.6307085612517493, + "grad_norm": 3.030518054962158, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8953855037689209, + "num_tokens": 802649295.0, + "step": 20680 + }, + { + "epoch": 2.6308357715303394, + "grad_norm": 3.0604121685028076, + "learning_rate": 1e-06, + "loss": 0.2275, + "mean_token_accuracy": 0.9145241975784302, + "num_tokens": 802679239.0, + "step": 20681 + }, + { + "epoch": 2.6309629818089304, + "grad_norm": 2.8683652877807617, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.908844530582428, + "num_tokens": 802719126.0, + "step": 20682 + }, + { + "epoch": 2.6310901920875205, + "grad_norm": 3.0851192474365234, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9087295532226562, + "num_tokens": 802751795.0, + "step": 20683 + }, + { + "epoch": 2.6312174023661115, + "grad_norm": 3.166409969329834, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8988843560218811, + "num_tokens": 802786064.0, + "step": 20684 + }, + { + "epoch": 2.6313446126447015, + "grad_norm": 2.665262460708618, + "learning_rate": 1e-06, + "loss": 0.2206, + "mean_token_accuracy": 0.9179145097732544, + "num_tokens": 802825421.0, + "step": 20685 + }, + { + "epoch": 2.631471822923292, + "grad_norm": 2.7821900844573975, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9040085077285767, + "num_tokens": 802867727.0, + "step": 20686 + }, + { + "epoch": 2.6315990332018826, + "grad_norm": 2.4724576473236084, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9114439487457275, + "num_tokens": 802911970.0, + "step": 20687 + }, + { + "epoch": 2.631726243480473, + "grad_norm": 2.7857882976531982, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9103550910949707, + "num_tokens": 802949558.0, + "step": 20688 + }, + { + "epoch": 2.6318534537590637, + "grad_norm": 2.9854819774627686, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.913165807723999, + "num_tokens": 802983416.0, + "step": 20689 + }, + { + "epoch": 2.631980664037654, + "grad_norm": 2.775174379348755, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9042171239852905, + "num_tokens": 803024214.0, + "step": 20690 + }, + { + "epoch": 2.6321078743162447, + "grad_norm": 2.855414867401123, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9051917791366577, + "num_tokens": 803062431.0, + "step": 20691 + }, + { + "epoch": 2.6322350845948352, + "grad_norm": 2.809788227081299, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9064023494720459, + "num_tokens": 803100568.0, + "step": 20692 + }, + { + "epoch": 2.6323622948734258, + "grad_norm": 2.7954204082489014, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9034131765365601, + "num_tokens": 803140975.0, + "step": 20693 + }, + { + "epoch": 2.6324895051520163, + "grad_norm": 2.8036537170410156, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9132122993469238, + "num_tokens": 803177726.0, + "step": 20694 + }, + { + "epoch": 2.632616715430607, + "grad_norm": 2.988520622253418, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9124642610549927, + "num_tokens": 803216296.0, + "step": 20695 + }, + { + "epoch": 2.6327439257091974, + "grad_norm": 2.682776927947998, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9026923775672913, + "num_tokens": 803264707.0, + "step": 20696 + }, + { + "epoch": 2.632871135987788, + "grad_norm": 2.850389003753662, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9085969924926758, + "num_tokens": 803306299.0, + "step": 20697 + }, + { + "epoch": 2.6329983462663784, + "grad_norm": 2.7715983390808105, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9111227989196777, + "num_tokens": 803344073.0, + "step": 20698 + }, + { + "epoch": 2.633125556544969, + "grad_norm": 3.1165857315063477, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.894268274307251, + "num_tokens": 803384744.0, + "step": 20699 + }, + { + "epoch": 2.6332527668235595, + "grad_norm": 3.330721855163574, + "learning_rate": 1e-06, + "loss": 0.312, + "mean_token_accuracy": 0.8865421414375305, + "num_tokens": 803423210.0, + "step": 20700 + }, + { + "epoch": 2.63337997710215, + "grad_norm": 2.919611692428589, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9040833115577698, + "num_tokens": 803459560.0, + "step": 20701 + }, + { + "epoch": 2.6335071873807405, + "grad_norm": 2.9136688709259033, + "learning_rate": 1e-06, + "loss": 0.2234, + "mean_token_accuracy": 0.9171299934387207, + "num_tokens": 803494698.0, + "step": 20702 + }, + { + "epoch": 2.633634397659331, + "grad_norm": 2.9797818660736084, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.9072075486183167, + "num_tokens": 803532804.0, + "step": 20703 + }, + { + "epoch": 2.633761607937921, + "grad_norm": 3.1175785064697266, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.9019119739532471, + "num_tokens": 803571038.0, + "step": 20704 + }, + { + "epoch": 2.633888818216512, + "grad_norm": 2.907202959060669, + "learning_rate": 1e-06, + "loss": 0.3171, + "mean_token_accuracy": 0.884748637676239, + "num_tokens": 803620393.0, + "step": 20705 + }, + { + "epoch": 2.634016028495102, + "grad_norm": 2.775498628616333, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.905190110206604, + "num_tokens": 803659246.0, + "step": 20706 + }, + { + "epoch": 2.634143238773693, + "grad_norm": 2.850076913833618, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.9002776145935059, + "num_tokens": 803706176.0, + "step": 20707 + }, + { + "epoch": 2.6342704490522832, + "grad_norm": 2.9846959114074707, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9035292267799377, + "num_tokens": 803746024.0, + "step": 20708 + }, + { + "epoch": 2.6343976593308738, + "grad_norm": 2.916764736175537, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9052056670188904, + "num_tokens": 803783043.0, + "step": 20709 + }, + { + "epoch": 2.6345248696094643, + "grad_norm": 3.1941030025482178, + "learning_rate": 1e-06, + "loss": 0.3105, + "mean_token_accuracy": 0.8868743181228638, + "num_tokens": 803819920.0, + "step": 20710 + }, + { + "epoch": 2.634652079888055, + "grad_norm": 3.031748056411743, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9035983085632324, + "num_tokens": 803854435.0, + "step": 20711 + }, + { + "epoch": 2.6347792901666454, + "grad_norm": 2.958148717880249, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8990586996078491, + "num_tokens": 803892443.0, + "step": 20712 + }, + { + "epoch": 2.634906500445236, + "grad_norm": 2.8699591159820557, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.906059205532074, + "num_tokens": 803927475.0, + "step": 20713 + }, + { + "epoch": 2.6350337107238264, + "grad_norm": 2.9227561950683594, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9079309105873108, + "num_tokens": 803962469.0, + "step": 20714 + }, + { + "epoch": 2.635160921002417, + "grad_norm": 3.06630277633667, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9018401503562927, + "num_tokens": 804000498.0, + "step": 20715 + }, + { + "epoch": 2.6352881312810075, + "grad_norm": 3.114506721496582, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8894630670547485, + "num_tokens": 804039204.0, + "step": 20716 + }, + { + "epoch": 2.635415341559598, + "grad_norm": 2.8635525703430176, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9021533727645874, + "num_tokens": 804081394.0, + "step": 20717 + }, + { + "epoch": 2.6355425518381885, + "grad_norm": 3.0391414165496826, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9060611128807068, + "num_tokens": 804116723.0, + "step": 20718 + }, + { + "epoch": 2.635669762116779, + "grad_norm": 2.9056100845336914, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.8973026871681213, + "num_tokens": 804159353.0, + "step": 20719 + }, + { + "epoch": 2.6357969723953696, + "grad_norm": 2.886509895324707, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8968621492385864, + "num_tokens": 804199596.0, + "step": 20720 + }, + { + "epoch": 2.63592418267396, + "grad_norm": 2.8429994583129883, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9094390869140625, + "num_tokens": 804236223.0, + "step": 20721 + }, + { + "epoch": 2.6360513929525506, + "grad_norm": 2.751377582550049, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9058111906051636, + "num_tokens": 804276739.0, + "step": 20722 + }, + { + "epoch": 2.636178603231141, + "grad_norm": 3.196056842803955, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8963843584060669, + "num_tokens": 804311549.0, + "step": 20723 + }, + { + "epoch": 2.6363058135097317, + "grad_norm": 3.1956589221954346, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.8994920253753662, + "num_tokens": 804347336.0, + "step": 20724 + }, + { + "epoch": 2.6364330237883222, + "grad_norm": 2.859584331512451, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9078463315963745, + "num_tokens": 804385194.0, + "step": 20725 + }, + { + "epoch": 2.6365602340669128, + "grad_norm": 2.760122776031494, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9060221314430237, + "num_tokens": 804429032.0, + "step": 20726 + }, + { + "epoch": 2.636687444345503, + "grad_norm": 2.698336124420166, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9116779565811157, + "num_tokens": 804471435.0, + "step": 20727 + }, + { + "epoch": 2.636814654624094, + "grad_norm": 3.0691428184509277, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9043920636177063, + "num_tokens": 804507844.0, + "step": 20728 + }, + { + "epoch": 2.636941864902684, + "grad_norm": 3.0880520343780518, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9054257273674011, + "num_tokens": 804542084.0, + "step": 20729 + }, + { + "epoch": 2.637069075181275, + "grad_norm": 2.8365471363067627, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8967971205711365, + "num_tokens": 804587557.0, + "step": 20730 + }, + { + "epoch": 2.637196285459865, + "grad_norm": 2.8963687419891357, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9106058478355408, + "num_tokens": 804625197.0, + "step": 20731 + }, + { + "epoch": 2.637323495738456, + "grad_norm": 3.1456758975982666, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9024355411529541, + "num_tokens": 804663381.0, + "step": 20732 + }, + { + "epoch": 2.637450706017046, + "grad_norm": 2.866790771484375, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9090331196784973, + "num_tokens": 804701468.0, + "step": 20733 + }, + { + "epoch": 2.6375779162956365, + "grad_norm": 2.731670618057251, + "learning_rate": 1e-06, + "loss": 0.233, + "mean_token_accuracy": 0.9127296805381775, + "num_tokens": 804742757.0, + "step": 20734 + }, + { + "epoch": 2.637705126574227, + "grad_norm": 3.273674249649048, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.9129437804222107, + "num_tokens": 804773115.0, + "step": 20735 + }, + { + "epoch": 2.6378323368528176, + "grad_norm": 3.2075836658477783, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8916682600975037, + "num_tokens": 804810142.0, + "step": 20736 + }, + { + "epoch": 2.637959547131408, + "grad_norm": 2.8519082069396973, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.896406888961792, + "num_tokens": 804850408.0, + "step": 20737 + }, + { + "epoch": 2.6380867574099987, + "grad_norm": 2.8265247344970703, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9082759022712708, + "num_tokens": 804887224.0, + "step": 20738 + }, + { + "epoch": 2.638213967688589, + "grad_norm": 3.0664381980895996, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.898655891418457, + "num_tokens": 804924245.0, + "step": 20739 + }, + { + "epoch": 2.6383411779671797, + "grad_norm": 2.8161158561706543, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8977000713348389, + "num_tokens": 804964770.0, + "step": 20740 + }, + { + "epoch": 2.6384683882457702, + "grad_norm": 2.915501594543457, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9073597192764282, + "num_tokens": 804999110.0, + "step": 20741 + }, + { + "epoch": 2.6385955985243608, + "grad_norm": 2.8950071334838867, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9044670462608337, + "num_tokens": 805036034.0, + "step": 20742 + }, + { + "epoch": 2.6387228088029513, + "grad_norm": 2.697817087173462, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9068418145179749, + "num_tokens": 805075683.0, + "step": 20743 + }, + { + "epoch": 2.638850019081542, + "grad_norm": 2.7660319805145264, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.8989232778549194, + "num_tokens": 805114611.0, + "step": 20744 + }, + { + "epoch": 2.6389772293601323, + "grad_norm": 2.870018720626831, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9089754223823547, + "num_tokens": 805153001.0, + "step": 20745 + }, + { + "epoch": 2.639104439638723, + "grad_norm": 2.7492527961730957, + "learning_rate": 1e-06, + "loss": 0.2282, + "mean_token_accuracy": 0.9138533473014832, + "num_tokens": 805190577.0, + "step": 20746 + }, + { + "epoch": 2.6392316499173134, + "grad_norm": 3.00124192237854, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.911806583404541, + "num_tokens": 805223429.0, + "step": 20747 + }, + { + "epoch": 2.639358860195904, + "grad_norm": 2.792695999145508, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8962609171867371, + "num_tokens": 805267688.0, + "step": 20748 + }, + { + "epoch": 2.6394860704744945, + "grad_norm": 2.9412074089050293, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9080529808998108, + "num_tokens": 805301345.0, + "step": 20749 + }, + { + "epoch": 2.639613280753085, + "grad_norm": 2.976227283477783, + "learning_rate": 1e-06, + "loss": 0.2369, + "mean_token_accuracy": 0.9076839089393616, + "num_tokens": 805335248.0, + "step": 20750 + }, + { + "epoch": 2.6397404910316755, + "grad_norm": 3.157485008239746, + "learning_rate": 1e-06, + "loss": 0.2879, + "mean_token_accuracy": 0.8964885473251343, + "num_tokens": 805374196.0, + "step": 20751 + }, + { + "epoch": 2.6398677013102656, + "grad_norm": 2.7232205867767334, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9062236547470093, + "num_tokens": 805414958.0, + "step": 20752 + }, + { + "epoch": 2.6399949115888566, + "grad_norm": 3.0509796142578125, + "learning_rate": 1e-06, + "loss": 0.295, + "mean_token_accuracy": 0.8940682411193848, + "num_tokens": 805455085.0, + "step": 20753 + }, + { + "epoch": 2.6401221218674467, + "grad_norm": 2.7936112880706787, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9076787233352661, + "num_tokens": 805493957.0, + "step": 20754 + }, + { + "epoch": 2.6402493321460376, + "grad_norm": 2.5461854934692383, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9099897146224976, + "num_tokens": 805538902.0, + "step": 20755 + }, + { + "epoch": 2.6403765424246277, + "grad_norm": 2.8372411727905273, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9043485522270203, + "num_tokens": 805581424.0, + "step": 20756 + }, + { + "epoch": 2.6405037527032187, + "grad_norm": 2.938464641571045, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9049074649810791, + "num_tokens": 805618137.0, + "step": 20757 + }, + { + "epoch": 2.6406309629818088, + "grad_norm": 2.9310362339019775, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9050092697143555, + "num_tokens": 805658159.0, + "step": 20758 + }, + { + "epoch": 2.6407581732603993, + "grad_norm": 2.6782712936401367, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9064847826957703, + "num_tokens": 805697596.0, + "step": 20759 + }, + { + "epoch": 2.64088538353899, + "grad_norm": 2.771714210510254, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.903853178024292, + "num_tokens": 805740020.0, + "step": 20760 + }, + { + "epoch": 2.6410125938175804, + "grad_norm": 2.695617914199829, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9038938283920288, + "num_tokens": 805781928.0, + "step": 20761 + }, + { + "epoch": 2.641139804096171, + "grad_norm": 2.734915018081665, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9026113152503967, + "num_tokens": 805826486.0, + "step": 20762 + }, + { + "epoch": 2.6412670143747614, + "grad_norm": 2.711134910583496, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8936570286750793, + "num_tokens": 805873760.0, + "step": 20763 + }, + { + "epoch": 2.641394224653352, + "grad_norm": 2.795685052871704, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.9020864963531494, + "num_tokens": 805911517.0, + "step": 20764 + }, + { + "epoch": 2.6415214349319425, + "grad_norm": 2.8288607597351074, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8954629898071289, + "num_tokens": 805951306.0, + "step": 20765 + }, + { + "epoch": 2.641648645210533, + "grad_norm": 2.8404812812805176, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8960326910018921, + "num_tokens": 805993915.0, + "step": 20766 + }, + { + "epoch": 2.6417758554891235, + "grad_norm": 3.042895555496216, + "learning_rate": 1e-06, + "loss": 0.3066, + "mean_token_accuracy": 0.8876523971557617, + "num_tokens": 806037835.0, + "step": 20767 + }, + { + "epoch": 2.641903065767714, + "grad_norm": 2.86804461479187, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9035699367523193, + "num_tokens": 806077561.0, + "step": 20768 + }, + { + "epoch": 2.6420302760463046, + "grad_norm": 3.0356528759002686, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8987685441970825, + "num_tokens": 806113943.0, + "step": 20769 + }, + { + "epoch": 2.642157486324895, + "grad_norm": 2.8221118450164795, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9006869792938232, + "num_tokens": 806153179.0, + "step": 20770 + }, + { + "epoch": 2.6422846966034856, + "grad_norm": 3.000772714614868, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9008910655975342, + "num_tokens": 806189829.0, + "step": 20771 + }, + { + "epoch": 2.642411906882076, + "grad_norm": 3.1766674518585205, + "learning_rate": 1e-06, + "loss": 0.3017, + "mean_token_accuracy": 0.8907437324523926, + "num_tokens": 806228180.0, + "step": 20772 + }, + { + "epoch": 2.6425391171606667, + "grad_norm": 2.7322607040405273, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9064112305641174, + "num_tokens": 806268827.0, + "step": 20773 + }, + { + "epoch": 2.6426663274392572, + "grad_norm": 2.6629655361175537, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9060105681419373, + "num_tokens": 806315137.0, + "step": 20774 + }, + { + "epoch": 2.6427935377178478, + "grad_norm": 2.9247658252716064, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9058181643486023, + "num_tokens": 806352423.0, + "step": 20775 + }, + { + "epoch": 2.6429207479964383, + "grad_norm": 2.709246873855591, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.8995927572250366, + "num_tokens": 806394401.0, + "step": 20776 + }, + { + "epoch": 2.6430479582750284, + "grad_norm": 2.8730664253234863, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.905290961265564, + "num_tokens": 806435241.0, + "step": 20777 + }, + { + "epoch": 2.6431751685536193, + "grad_norm": 2.8013832569122314, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9095079302787781, + "num_tokens": 806475980.0, + "step": 20778 + }, + { + "epoch": 2.6433023788322094, + "grad_norm": 2.7638802528381348, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9025341868400574, + "num_tokens": 806520039.0, + "step": 20779 + }, + { + "epoch": 2.6434295891108004, + "grad_norm": 2.7900803089141846, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9039713144302368, + "num_tokens": 806561158.0, + "step": 20780 + }, + { + "epoch": 2.6435567993893905, + "grad_norm": 2.9106833934783936, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9019691944122314, + "num_tokens": 806598912.0, + "step": 20781 + }, + { + "epoch": 2.6436840096679814, + "grad_norm": 2.8870997428894043, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9032018184661865, + "num_tokens": 806636682.0, + "step": 20782 + }, + { + "epoch": 2.6438112199465715, + "grad_norm": 2.7941956520080566, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9125846028327942, + "num_tokens": 806676349.0, + "step": 20783 + }, + { + "epoch": 2.643938430225162, + "grad_norm": 2.9956719875335693, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9095286130905151, + "num_tokens": 806708963.0, + "step": 20784 + }, + { + "epoch": 2.6440656405037526, + "grad_norm": 2.9629621505737305, + "learning_rate": 1e-06, + "loss": 0.2621, + "mean_token_accuracy": 0.9018330574035645, + "num_tokens": 806748610.0, + "step": 20785 + }, + { + "epoch": 2.644192850782343, + "grad_norm": 3.002946138381958, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.901161253452301, + "num_tokens": 806785367.0, + "step": 20786 + }, + { + "epoch": 2.6443200610609336, + "grad_norm": 3.0080058574676514, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9008439779281616, + "num_tokens": 806822748.0, + "step": 20787 + }, + { + "epoch": 2.644447271339524, + "grad_norm": 2.8969779014587402, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9023022055625916, + "num_tokens": 806863445.0, + "step": 20788 + }, + { + "epoch": 2.6445744816181147, + "grad_norm": 2.587205171585083, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9121328592300415, + "num_tokens": 806908087.0, + "step": 20789 + }, + { + "epoch": 2.6447016918967052, + "grad_norm": 2.952238082885742, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9019157886505127, + "num_tokens": 806946717.0, + "step": 20790 + }, + { + "epoch": 2.6448289021752958, + "grad_norm": 2.963590145111084, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8935197591781616, + "num_tokens": 806990354.0, + "step": 20791 + }, + { + "epoch": 2.6449561124538863, + "grad_norm": 3.2902603149414062, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.9053707718849182, + "num_tokens": 807025123.0, + "step": 20792 + }, + { + "epoch": 2.645083322732477, + "grad_norm": 2.9984216690063477, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9071304202079773, + "num_tokens": 807062905.0, + "step": 20793 + }, + { + "epoch": 2.6452105330110673, + "grad_norm": 3.1857054233551025, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8979573845863342, + "num_tokens": 807101270.0, + "step": 20794 + }, + { + "epoch": 2.645337743289658, + "grad_norm": 2.664058208465576, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9116189479827881, + "num_tokens": 807142605.0, + "step": 20795 + }, + { + "epoch": 2.6454649535682484, + "grad_norm": 2.9364917278289795, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8997503519058228, + "num_tokens": 807180905.0, + "step": 20796 + }, + { + "epoch": 2.645592163846839, + "grad_norm": 2.7668659687042236, + "learning_rate": 1e-06, + "loss": 0.234, + "mean_token_accuracy": 0.9136288166046143, + "num_tokens": 807218398.0, + "step": 20797 + }, + { + "epoch": 2.6457193741254295, + "grad_norm": 2.7801764011383057, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9061681628227234, + "num_tokens": 807258160.0, + "step": 20798 + }, + { + "epoch": 2.64584658440402, + "grad_norm": 3.0860002040863037, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8921182155609131, + "num_tokens": 807301364.0, + "step": 20799 + }, + { + "epoch": 2.6459737946826105, + "grad_norm": 3.19950795173645, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9112772941589355, + "num_tokens": 807330525.0, + "step": 20800 + }, + { + "epoch": 2.646101004961201, + "grad_norm": 3.343575954437256, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8964760303497314, + "num_tokens": 807361693.0, + "step": 20801 + }, + { + "epoch": 2.646228215239791, + "grad_norm": 2.831902503967285, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.9085826873779297, + "num_tokens": 807397920.0, + "step": 20802 + }, + { + "epoch": 2.646355425518382, + "grad_norm": 2.963392734527588, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8969243168830872, + "num_tokens": 807438499.0, + "step": 20803 + }, + { + "epoch": 2.646482635796972, + "grad_norm": 2.805368185043335, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9064814448356628, + "num_tokens": 807480146.0, + "step": 20804 + }, + { + "epoch": 2.646609846075563, + "grad_norm": 2.8310296535491943, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9019099473953247, + "num_tokens": 807522682.0, + "step": 20805 + }, + { + "epoch": 2.6467370563541532, + "grad_norm": 2.7925422191619873, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9023436307907104, + "num_tokens": 807561992.0, + "step": 20806 + }, + { + "epoch": 2.6468642666327438, + "grad_norm": 3.154298782348633, + "learning_rate": 1e-06, + "loss": 0.2896, + "mean_token_accuracy": 0.8935597538948059, + "num_tokens": 807597960.0, + "step": 20807 + }, + { + "epoch": 2.6469914769113343, + "grad_norm": 2.697662591934204, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.901584267616272, + "num_tokens": 807642959.0, + "step": 20808 + }, + { + "epoch": 2.647118687189925, + "grad_norm": 2.9691519737243652, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8953240513801575, + "num_tokens": 807680702.0, + "step": 20809 + }, + { + "epoch": 2.6472458974685154, + "grad_norm": 2.6611340045928955, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.904937207698822, + "num_tokens": 807724497.0, + "step": 20810 + }, + { + "epoch": 2.647373107747106, + "grad_norm": 2.7114193439483643, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9046971201896667, + "num_tokens": 807764473.0, + "step": 20811 + }, + { + "epoch": 2.6475003180256964, + "grad_norm": 2.7422285079956055, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.9170035719871521, + "num_tokens": 807804583.0, + "step": 20812 + }, + { + "epoch": 2.647627528304287, + "grad_norm": 2.768040895462036, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8990963697433472, + "num_tokens": 807848697.0, + "step": 20813 + }, + { + "epoch": 2.6477547385828775, + "grad_norm": 2.9489662647247314, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9088627696037292, + "num_tokens": 807884169.0, + "step": 20814 + }, + { + "epoch": 2.647881948861468, + "grad_norm": 3.059192180633545, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8965263366699219, + "num_tokens": 807922264.0, + "step": 20815 + }, + { + "epoch": 2.6480091591400585, + "grad_norm": 2.884457588195801, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8934133052825928, + "num_tokens": 807967444.0, + "step": 20816 + }, + { + "epoch": 2.648136369418649, + "grad_norm": 2.7834856510162354, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8977727293968201, + "num_tokens": 808011843.0, + "step": 20817 + }, + { + "epoch": 2.6482635796972396, + "grad_norm": 2.7955210208892822, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9037241339683533, + "num_tokens": 808052684.0, + "step": 20818 + }, + { + "epoch": 2.64839078997583, + "grad_norm": 2.8116297721862793, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8952379822731018, + "num_tokens": 808095547.0, + "step": 20819 + }, + { + "epoch": 2.6485180002544206, + "grad_norm": 2.8424036502838135, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9084643125534058, + "num_tokens": 808136716.0, + "step": 20820 + }, + { + "epoch": 2.648645210533011, + "grad_norm": 3.2613937854766846, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9091585874557495, + "num_tokens": 808165790.0, + "step": 20821 + }, + { + "epoch": 2.6487724208116017, + "grad_norm": 2.939927816390991, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9100464582443237, + "num_tokens": 808198256.0, + "step": 20822 + }, + { + "epoch": 2.648899631090192, + "grad_norm": 2.9691460132598877, + "learning_rate": 1e-06, + "loss": 0.3062, + "mean_token_accuracy": 0.8872933387756348, + "num_tokens": 808240414.0, + "step": 20823 + }, + { + "epoch": 2.6490268413687827, + "grad_norm": 2.8709328174591064, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9038369655609131, + "num_tokens": 808281157.0, + "step": 20824 + }, + { + "epoch": 2.649154051647373, + "grad_norm": 2.968879222869873, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.89469975233078, + "num_tokens": 808322200.0, + "step": 20825 + }, + { + "epoch": 2.649281261925964, + "grad_norm": 2.8842833042144775, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8973524570465088, + "num_tokens": 808363147.0, + "step": 20826 + }, + { + "epoch": 2.649408472204554, + "grad_norm": 2.9699032306671143, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9074663519859314, + "num_tokens": 808398771.0, + "step": 20827 + }, + { + "epoch": 2.649535682483145, + "grad_norm": 2.7401621341705322, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9096649885177612, + "num_tokens": 808440989.0, + "step": 20828 + }, + { + "epoch": 2.649662892761735, + "grad_norm": 2.820432424545288, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9088839888572693, + "num_tokens": 808484348.0, + "step": 20829 + }, + { + "epoch": 2.649790103040326, + "grad_norm": 2.553922414779663, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9029634594917297, + "num_tokens": 808528419.0, + "step": 20830 + }, + { + "epoch": 2.649917313318916, + "grad_norm": 2.8979036808013916, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9036127924919128, + "num_tokens": 808572391.0, + "step": 20831 + }, + { + "epoch": 2.6500445235975065, + "grad_norm": 2.684624671936035, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9109784364700317, + "num_tokens": 808610709.0, + "step": 20832 + }, + { + "epoch": 2.650171733876097, + "grad_norm": 3.160313367843628, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9022015333175659, + "num_tokens": 808645171.0, + "step": 20833 + }, + { + "epoch": 2.6502989441546876, + "grad_norm": 2.7307024002075195, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9058695435523987, + "num_tokens": 808686574.0, + "step": 20834 + }, + { + "epoch": 2.650426154433278, + "grad_norm": 2.9066481590270996, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.897430419921875, + "num_tokens": 808727292.0, + "step": 20835 + }, + { + "epoch": 2.6505533647118686, + "grad_norm": 2.759993553161621, + "learning_rate": 1e-06, + "loss": 0.2271, + "mean_token_accuracy": 0.9168380498886108, + "num_tokens": 808762736.0, + "step": 20836 + }, + { + "epoch": 2.650680574990459, + "grad_norm": 2.9004502296447754, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9076310396194458, + "num_tokens": 808800727.0, + "step": 20837 + }, + { + "epoch": 2.6508077852690497, + "grad_norm": 2.864851713180542, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.8919067978858948, + "num_tokens": 808844086.0, + "step": 20838 + }, + { + "epoch": 2.6509349955476402, + "grad_norm": 3.120823383331299, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9135790467262268, + "num_tokens": 808878795.0, + "step": 20839 + }, + { + "epoch": 2.6510622058262308, + "grad_norm": 2.762199878692627, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9017056822776794, + "num_tokens": 808921023.0, + "step": 20840 + }, + { + "epoch": 2.6511894161048213, + "grad_norm": 2.9183318614959717, + "learning_rate": 1e-06, + "loss": 0.2185, + "mean_token_accuracy": 0.9197489619255066, + "num_tokens": 808954384.0, + "step": 20841 + }, + { + "epoch": 2.651316626383412, + "grad_norm": 3.1297664642333984, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.9017775058746338, + "num_tokens": 808989796.0, + "step": 20842 + }, + { + "epoch": 2.6514438366620023, + "grad_norm": 3.0482017993927, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9093238115310669, + "num_tokens": 809024268.0, + "step": 20843 + }, + { + "epoch": 2.651571046940593, + "grad_norm": 2.680403709411621, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9006912708282471, + "num_tokens": 809067825.0, + "step": 20844 + }, + { + "epoch": 2.6516982572191834, + "grad_norm": 2.947979688644409, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9102832078933716, + "num_tokens": 809105332.0, + "step": 20845 + }, + { + "epoch": 2.651825467497774, + "grad_norm": 3.289612293243408, + "learning_rate": 1e-06, + "loss": 0.3031, + "mean_token_accuracy": 0.8889041543006897, + "num_tokens": 809141560.0, + "step": 20846 + }, + { + "epoch": 2.6519526777763645, + "grad_norm": 3.04551100730896, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9076062440872192, + "num_tokens": 809174928.0, + "step": 20847 + }, + { + "epoch": 2.652079888054955, + "grad_norm": 2.861269950866699, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8939781785011292, + "num_tokens": 809214440.0, + "step": 20848 + }, + { + "epoch": 2.6522070983335455, + "grad_norm": 2.814340353012085, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9074828624725342, + "num_tokens": 809252849.0, + "step": 20849 + }, + { + "epoch": 2.6523343086121356, + "grad_norm": 2.6561734676361084, + "learning_rate": 1e-06, + "loss": 0.2334, + "mean_token_accuracy": 0.9136905074119568, + "num_tokens": 809296217.0, + "step": 20850 + }, + { + "epoch": 2.6524615188907266, + "grad_norm": 2.974600076675415, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9116739630699158, + "num_tokens": 809334633.0, + "step": 20851 + }, + { + "epoch": 2.6525887291693167, + "grad_norm": 2.814235210418701, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.891677975654602, + "num_tokens": 809375935.0, + "step": 20852 + }, + { + "epoch": 2.6527159394479076, + "grad_norm": 2.881218433380127, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8951576948165894, + "num_tokens": 809420241.0, + "step": 20853 + }, + { + "epoch": 2.6528431497264977, + "grad_norm": 2.7383627891540527, + "learning_rate": 1e-06, + "loss": 0.2314, + "mean_token_accuracy": 0.9133689403533936, + "num_tokens": 809457930.0, + "step": 20854 + }, + { + "epoch": 2.6529703600050887, + "grad_norm": 3.281524419784546, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.900692343711853, + "num_tokens": 809490764.0, + "step": 20855 + }, + { + "epoch": 2.6530975702836788, + "grad_norm": 2.9196252822875977, + "learning_rate": 1e-06, + "loss": 0.2394, + "mean_token_accuracy": 0.9105662107467651, + "num_tokens": 809525267.0, + "step": 20856 + }, + { + "epoch": 2.6532247805622693, + "grad_norm": 3.2861642837524414, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9040611982345581, + "num_tokens": 809556106.0, + "step": 20857 + }, + { + "epoch": 2.65335199084086, + "grad_norm": 2.7863619327545166, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9138666987419128, + "num_tokens": 809590721.0, + "step": 20858 + }, + { + "epoch": 2.6534792011194503, + "grad_norm": 2.9419894218444824, + "learning_rate": 1e-06, + "loss": 0.2313, + "mean_token_accuracy": 0.9093344211578369, + "num_tokens": 809626939.0, + "step": 20859 + }, + { + "epoch": 2.653606411398041, + "grad_norm": 2.7037086486816406, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9091523289680481, + "num_tokens": 809667102.0, + "step": 20860 + }, + { + "epoch": 2.6537336216766314, + "grad_norm": 2.815809726715088, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9032526016235352, + "num_tokens": 809709318.0, + "step": 20861 + }, + { + "epoch": 2.653860831955222, + "grad_norm": 2.7221133708953857, + "learning_rate": 1e-06, + "loss": 0.2323, + "mean_token_accuracy": 0.9121708273887634, + "num_tokens": 809746388.0, + "step": 20862 + }, + { + "epoch": 2.6539880422338125, + "grad_norm": 3.0169219970703125, + "learning_rate": 1e-06, + "loss": 0.2305, + "mean_token_accuracy": 0.9137176275253296, + "num_tokens": 809778391.0, + "step": 20863 + }, + { + "epoch": 2.654115252512403, + "grad_norm": 2.9343771934509277, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9054616093635559, + "num_tokens": 809813330.0, + "step": 20864 + }, + { + "epoch": 2.6542424627909935, + "grad_norm": 2.812142848968506, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9031165838241577, + "num_tokens": 809852178.0, + "step": 20865 + }, + { + "epoch": 2.654369673069584, + "grad_norm": 2.8413589000701904, + "learning_rate": 1e-06, + "loss": 0.2272, + "mean_token_accuracy": 0.9165845513343811, + "num_tokens": 809884637.0, + "step": 20866 + }, + { + "epoch": 2.6544968833481746, + "grad_norm": 2.7023377418518066, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9085463285446167, + "num_tokens": 809925786.0, + "step": 20867 + }, + { + "epoch": 2.654624093626765, + "grad_norm": 2.9723355770111084, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8983463644981384, + "num_tokens": 809963354.0, + "step": 20868 + }, + { + "epoch": 2.6547513039053556, + "grad_norm": 3.4327940940856934, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9044767022132874, + "num_tokens": 809996603.0, + "step": 20869 + }, + { + "epoch": 2.654878514183946, + "grad_norm": 3.1243302822113037, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9071147441864014, + "num_tokens": 810039595.0, + "step": 20870 + }, + { + "epoch": 2.6550057244625367, + "grad_norm": 2.9306068420410156, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9039329290390015, + "num_tokens": 810077928.0, + "step": 20871 + }, + { + "epoch": 2.655132934741127, + "grad_norm": 2.9308414459228516, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9044605493545532, + "num_tokens": 810115033.0, + "step": 20872 + }, + { + "epoch": 2.6552601450197177, + "grad_norm": 2.913213014602661, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9118880033493042, + "num_tokens": 810150684.0, + "step": 20873 + }, + { + "epoch": 2.6553873552983083, + "grad_norm": 2.692406177520752, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9022567272186279, + "num_tokens": 810197856.0, + "step": 20874 + }, + { + "epoch": 2.6555145655768984, + "grad_norm": 2.8569276332855225, + "learning_rate": 1e-06, + "loss": 0.2302, + "mean_token_accuracy": 0.9153625965118408, + "num_tokens": 810233533.0, + "step": 20875 + }, + { + "epoch": 2.6556417758554893, + "grad_norm": 2.957763671875, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.8995318412780762, + "num_tokens": 810269221.0, + "step": 20876 + }, + { + "epoch": 2.6557689861340794, + "grad_norm": 2.8032703399658203, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9043899774551392, + "num_tokens": 810309513.0, + "step": 20877 + }, + { + "epoch": 2.6558961964126704, + "grad_norm": 3.178968906402588, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.9034278392791748, + "num_tokens": 810341450.0, + "step": 20878 + }, + { + "epoch": 2.6560234066912605, + "grad_norm": 3.105799436569214, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9063021540641785, + "num_tokens": 810375622.0, + "step": 20879 + }, + { + "epoch": 2.6561506169698514, + "grad_norm": 2.8916773796081543, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8883890509605408, + "num_tokens": 810413620.0, + "step": 20880 + }, + { + "epoch": 2.6562778272484415, + "grad_norm": 2.602240800857544, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9063702821731567, + "num_tokens": 810458680.0, + "step": 20881 + }, + { + "epoch": 2.656405037527032, + "grad_norm": 3.1917812824249268, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9022151231765747, + "num_tokens": 810490923.0, + "step": 20882 + }, + { + "epoch": 2.6565322478056226, + "grad_norm": 2.934349775314331, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9091703295707703, + "num_tokens": 810527908.0, + "step": 20883 + }, + { + "epoch": 2.656659458084213, + "grad_norm": 3.0145440101623535, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8982006311416626, + "num_tokens": 810564664.0, + "step": 20884 + }, + { + "epoch": 2.6567866683628036, + "grad_norm": 2.7505428791046143, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9115290641784668, + "num_tokens": 810604148.0, + "step": 20885 + }, + { + "epoch": 2.656913878641394, + "grad_norm": 2.986053705215454, + "learning_rate": 1e-06, + "loss": 0.2416, + "mean_token_accuracy": 0.9108227491378784, + "num_tokens": 810639874.0, + "step": 20886 + }, + { + "epoch": 2.6570410889199847, + "grad_norm": 3.149090528488159, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.893904447555542, + "num_tokens": 810674218.0, + "step": 20887 + }, + { + "epoch": 2.6571682991985752, + "grad_norm": 2.9004416465759277, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9024684429168701, + "num_tokens": 810714269.0, + "step": 20888 + }, + { + "epoch": 2.6572955094771658, + "grad_norm": 2.8231027126312256, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9082728624343872, + "num_tokens": 810755637.0, + "step": 20889 + }, + { + "epoch": 2.6574227197557563, + "grad_norm": 2.961203098297119, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8982526063919067, + "num_tokens": 810794167.0, + "step": 20890 + }, + { + "epoch": 2.657549930034347, + "grad_norm": 3.0934293270111084, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8925442695617676, + "num_tokens": 810836299.0, + "step": 20891 + }, + { + "epoch": 2.6576771403129373, + "grad_norm": 3.082688808441162, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.8990203142166138, + "num_tokens": 810872484.0, + "step": 20892 + }, + { + "epoch": 2.657804350591528, + "grad_norm": 2.596423387527466, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9095706343650818, + "num_tokens": 810920477.0, + "step": 20893 + }, + { + "epoch": 2.6579315608701184, + "grad_norm": 2.7473390102386475, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8975787162780762, + "num_tokens": 810962657.0, + "step": 20894 + }, + { + "epoch": 2.658058771148709, + "grad_norm": 2.9111251831054688, + "learning_rate": 1e-06, + "loss": 0.2858, + "mean_token_accuracy": 0.8991299867630005, + "num_tokens": 811001669.0, + "step": 20895 + }, + { + "epoch": 2.6581859814272994, + "grad_norm": 2.9718801975250244, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9044292569160461, + "num_tokens": 811042696.0, + "step": 20896 + }, + { + "epoch": 2.65831319170589, + "grad_norm": 2.6518537998199463, + "learning_rate": 1e-06, + "loss": 0.2394, + "mean_token_accuracy": 0.9101018309593201, + "num_tokens": 811084505.0, + "step": 20897 + }, + { + "epoch": 2.6584404019844805, + "grad_norm": 2.8227267265319824, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9072155952453613, + "num_tokens": 811122217.0, + "step": 20898 + }, + { + "epoch": 2.658567612263071, + "grad_norm": 2.8005082607269287, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9120813608169556, + "num_tokens": 811162507.0, + "step": 20899 + }, + { + "epoch": 2.658694822541661, + "grad_norm": 2.6738040447235107, + "learning_rate": 1e-06, + "loss": 0.2342, + "mean_token_accuracy": 0.9153273105621338, + "num_tokens": 811204145.0, + "step": 20900 + }, + { + "epoch": 2.658822032820252, + "grad_norm": 2.6682496070861816, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9079729914665222, + "num_tokens": 811246289.0, + "step": 20901 + }, + { + "epoch": 2.658949243098842, + "grad_norm": 2.8860652446746826, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9095553755760193, + "num_tokens": 811284076.0, + "step": 20902 + }, + { + "epoch": 2.659076453377433, + "grad_norm": 2.7602202892303467, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9003947973251343, + "num_tokens": 811327497.0, + "step": 20903 + }, + { + "epoch": 2.6592036636560232, + "grad_norm": 2.876661777496338, + "learning_rate": 1e-06, + "loss": 0.237, + "mean_token_accuracy": 0.9112887978553772, + "num_tokens": 811365702.0, + "step": 20904 + }, + { + "epoch": 2.6593308739346138, + "grad_norm": 2.7929975986480713, + "learning_rate": 1e-06, + "loss": 0.2809, + "mean_token_accuracy": 0.8959708213806152, + "num_tokens": 811407291.0, + "step": 20905 + }, + { + "epoch": 2.6594580842132043, + "grad_norm": 2.977786064147949, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9006372094154358, + "num_tokens": 811447008.0, + "step": 20906 + }, + { + "epoch": 2.659585294491795, + "grad_norm": 3.0503344535827637, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9031786322593689, + "num_tokens": 811483315.0, + "step": 20907 + }, + { + "epoch": 2.6597125047703853, + "grad_norm": 2.9469268321990967, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8949340581893921, + "num_tokens": 811524657.0, + "step": 20908 + }, + { + "epoch": 2.659839715048976, + "grad_norm": 2.976844310760498, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9054447412490845, + "num_tokens": 811562218.0, + "step": 20909 + }, + { + "epoch": 2.6599669253275664, + "grad_norm": 3.0767314434051514, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8999886512756348, + "num_tokens": 811598313.0, + "step": 20910 + }, + { + "epoch": 2.660094135606157, + "grad_norm": 2.8260083198547363, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9048641920089722, + "num_tokens": 811639993.0, + "step": 20911 + }, + { + "epoch": 2.6602213458847475, + "grad_norm": 2.9885082244873047, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8986705541610718, + "num_tokens": 811677945.0, + "step": 20912 + }, + { + "epoch": 2.660348556163338, + "grad_norm": 2.6899378299713135, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.9126559495925903, + "num_tokens": 811718712.0, + "step": 20913 + }, + { + "epoch": 2.6604757664419285, + "grad_norm": 3.214754819869995, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9046964645385742, + "num_tokens": 811752359.0, + "step": 20914 + }, + { + "epoch": 2.660602976720519, + "grad_norm": 3.061180830001831, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9054743051528931, + "num_tokens": 811787554.0, + "step": 20915 + }, + { + "epoch": 2.6607301869991096, + "grad_norm": 3.0367767810821533, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9031999111175537, + "num_tokens": 811824889.0, + "step": 20916 + }, + { + "epoch": 2.6608573972777, + "grad_norm": 2.6705479621887207, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9117741584777832, + "num_tokens": 811865156.0, + "step": 20917 + }, + { + "epoch": 2.6609846075562906, + "grad_norm": 2.715287208557129, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9062546491622925, + "num_tokens": 811905374.0, + "step": 20918 + }, + { + "epoch": 2.661111817834881, + "grad_norm": 2.707049608230591, + "learning_rate": 1e-06, + "loss": 0.2346, + "mean_token_accuracy": 0.9117240309715271, + "num_tokens": 811943567.0, + "step": 20919 + }, + { + "epoch": 2.6612390281134717, + "grad_norm": 2.788816213607788, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9081076979637146, + "num_tokens": 811983040.0, + "step": 20920 + }, + { + "epoch": 2.661366238392062, + "grad_norm": 2.7845876216888428, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9047422409057617, + "num_tokens": 812021927.0, + "step": 20921 + }, + { + "epoch": 2.6614934486706527, + "grad_norm": 2.493743658065796, + "learning_rate": 1e-06, + "loss": 0.2133, + "mean_token_accuracy": 0.9203940629959106, + "num_tokens": 812059702.0, + "step": 20922 + }, + { + "epoch": 2.661620658949243, + "grad_norm": 2.8206236362457275, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9140286445617676, + "num_tokens": 812098362.0, + "step": 20923 + }, + { + "epoch": 2.661747869227834, + "grad_norm": 2.647362470626831, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9073847532272339, + "num_tokens": 812142231.0, + "step": 20924 + }, + { + "epoch": 2.661875079506424, + "grad_norm": 2.9287147521972656, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9045074582099915, + "num_tokens": 812175646.0, + "step": 20925 + }, + { + "epoch": 2.662002289785015, + "grad_norm": 2.655306816101074, + "learning_rate": 1e-06, + "loss": 0.2252, + "mean_token_accuracy": 0.9156228303909302, + "num_tokens": 812216159.0, + "step": 20926 + }, + { + "epoch": 2.662129500063605, + "grad_norm": 2.6534621715545654, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.9082143306732178, + "num_tokens": 812260423.0, + "step": 20927 + }, + { + "epoch": 2.662256710342196, + "grad_norm": 2.9731264114379883, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8962634801864624, + "num_tokens": 812299990.0, + "step": 20928 + }, + { + "epoch": 2.662383920620786, + "grad_norm": 2.842041492462158, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.902442455291748, + "num_tokens": 812343431.0, + "step": 20929 + }, + { + "epoch": 2.6625111308993765, + "grad_norm": 2.9670424461364746, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.8930917978286743, + "num_tokens": 812386286.0, + "step": 20930 + }, + { + "epoch": 2.662638341177967, + "grad_norm": 3.048748731613159, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8904114961624146, + "num_tokens": 812425210.0, + "step": 20931 + }, + { + "epoch": 2.6627655514565576, + "grad_norm": 2.8579671382904053, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9114495515823364, + "num_tokens": 812463572.0, + "step": 20932 + }, + { + "epoch": 2.662892761735148, + "grad_norm": 2.7565159797668457, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9053161144256592, + "num_tokens": 812503038.0, + "step": 20933 + }, + { + "epoch": 2.6630199720137386, + "grad_norm": 2.83408522605896, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9023404121398926, + "num_tokens": 812541717.0, + "step": 20934 + }, + { + "epoch": 2.663147182292329, + "grad_norm": 2.8550384044647217, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.908602237701416, + "num_tokens": 812576500.0, + "step": 20935 + }, + { + "epoch": 2.6632743925709197, + "grad_norm": 2.7320168018341064, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9080368280410767, + "num_tokens": 812616723.0, + "step": 20936 + }, + { + "epoch": 2.66340160284951, + "grad_norm": 2.8285107612609863, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9130977392196655, + "num_tokens": 812653382.0, + "step": 20937 + }, + { + "epoch": 2.6635288131281007, + "grad_norm": 2.8559792041778564, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.9132189750671387, + "num_tokens": 812689673.0, + "step": 20938 + }, + { + "epoch": 2.6636560234066913, + "grad_norm": 3.0579910278320312, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9062944650650024, + "num_tokens": 812726144.0, + "step": 20939 + }, + { + "epoch": 2.663783233685282, + "grad_norm": 2.525050640106201, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9087730646133423, + "num_tokens": 812773211.0, + "step": 20940 + }, + { + "epoch": 2.6639104439638723, + "grad_norm": 2.95096755027771, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.911217451095581, + "num_tokens": 812809236.0, + "step": 20941 + }, + { + "epoch": 2.664037654242463, + "grad_norm": 3.0375938415527344, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9089367985725403, + "num_tokens": 812840418.0, + "step": 20942 + }, + { + "epoch": 2.6641648645210534, + "grad_norm": 2.836120367050171, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9054912328720093, + "num_tokens": 812879015.0, + "step": 20943 + }, + { + "epoch": 2.664292074799644, + "grad_norm": 2.8792035579681396, + "learning_rate": 1e-06, + "loss": 0.2979, + "mean_token_accuracy": 0.8911868929862976, + "num_tokens": 812921231.0, + "step": 20944 + }, + { + "epoch": 2.6644192850782344, + "grad_norm": 2.779895782470703, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9075940251350403, + "num_tokens": 812965394.0, + "step": 20945 + }, + { + "epoch": 2.664546495356825, + "grad_norm": 2.9587459564208984, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9087547659873962, + "num_tokens": 813001545.0, + "step": 20946 + }, + { + "epoch": 2.6646737056354155, + "grad_norm": 3.028773546218872, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9052087068557739, + "num_tokens": 813039264.0, + "step": 20947 + }, + { + "epoch": 2.6648009159140056, + "grad_norm": 2.916339159011841, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.8955392837524414, + "num_tokens": 813077575.0, + "step": 20948 + }, + { + "epoch": 2.6649281261925966, + "grad_norm": 3.0204005241394043, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9075056910514832, + "num_tokens": 813112000.0, + "step": 20949 + }, + { + "epoch": 2.6650553364711866, + "grad_norm": 3.0698578357696533, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8983436226844788, + "num_tokens": 813145463.0, + "step": 20950 + }, + { + "epoch": 2.6651825467497776, + "grad_norm": 2.8368489742279053, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9065345525741577, + "num_tokens": 813185237.0, + "step": 20951 + }, + { + "epoch": 2.6653097570283677, + "grad_norm": 2.813701629638672, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.9007834196090698, + "num_tokens": 813225143.0, + "step": 20952 + }, + { + "epoch": 2.6654369673069587, + "grad_norm": 2.6106507778167725, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.906741201877594, + "num_tokens": 813269752.0, + "step": 20953 + }, + { + "epoch": 2.6655641775855488, + "grad_norm": 2.7436094284057617, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9088188409805298, + "num_tokens": 813310759.0, + "step": 20954 + }, + { + "epoch": 2.6656913878641393, + "grad_norm": 3.026449203491211, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9023250341415405, + "num_tokens": 813347689.0, + "step": 20955 + }, + { + "epoch": 2.66581859814273, + "grad_norm": 2.973720073699951, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8895894289016724, + "num_tokens": 813388241.0, + "step": 20956 + }, + { + "epoch": 2.6659458084213203, + "grad_norm": 2.9751994609832764, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9089755415916443, + "num_tokens": 813426982.0, + "step": 20957 + }, + { + "epoch": 2.666073018699911, + "grad_norm": 2.997804880142212, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9076549410820007, + "num_tokens": 813463168.0, + "step": 20958 + }, + { + "epoch": 2.6662002289785014, + "grad_norm": 2.6998696327209473, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.903317928314209, + "num_tokens": 813508869.0, + "step": 20959 + }, + { + "epoch": 2.666327439257092, + "grad_norm": 2.8170166015625, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9099088907241821, + "num_tokens": 813546126.0, + "step": 20960 + }, + { + "epoch": 2.6664546495356825, + "grad_norm": 2.6542611122131348, + "learning_rate": 1e-06, + "loss": 0.2376, + "mean_token_accuracy": 0.9114305973052979, + "num_tokens": 813589292.0, + "step": 20961 + }, + { + "epoch": 2.666581859814273, + "grad_norm": 2.7682273387908936, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.90968918800354, + "num_tokens": 813630887.0, + "step": 20962 + }, + { + "epoch": 2.6667090700928635, + "grad_norm": 2.8351776599884033, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9075852632522583, + "num_tokens": 813668587.0, + "step": 20963 + }, + { + "epoch": 2.666836280371454, + "grad_norm": 3.011497974395752, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9049840569496155, + "num_tokens": 813704576.0, + "step": 20964 + }, + { + "epoch": 2.6669634906500446, + "grad_norm": 2.588871717453003, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9096695780754089, + "num_tokens": 813749376.0, + "step": 20965 + }, + { + "epoch": 2.667090700928635, + "grad_norm": 2.951871633529663, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9067769050598145, + "num_tokens": 813784695.0, + "step": 20966 + }, + { + "epoch": 2.6672179112072256, + "grad_norm": 2.765620470046997, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9088202118873596, + "num_tokens": 813822125.0, + "step": 20967 + }, + { + "epoch": 2.667345121485816, + "grad_norm": 2.8732147216796875, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9058797359466553, + "num_tokens": 813858191.0, + "step": 20968 + }, + { + "epoch": 2.6674723317644067, + "grad_norm": 2.7089498043060303, + "learning_rate": 1e-06, + "loss": 0.2248, + "mean_token_accuracy": 0.9147650599479675, + "num_tokens": 813898869.0, + "step": 20969 + }, + { + "epoch": 2.667599542042997, + "grad_norm": 2.5996079444885254, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9103280305862427, + "num_tokens": 813940264.0, + "step": 20970 + }, + { + "epoch": 2.6677267523215877, + "grad_norm": 2.849273443222046, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9058581590652466, + "num_tokens": 813975389.0, + "step": 20971 + }, + { + "epoch": 2.6678539626001783, + "grad_norm": 2.9170405864715576, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.900762140750885, + "num_tokens": 814013512.0, + "step": 20972 + }, + { + "epoch": 2.6679811728787683, + "grad_norm": 2.90702486038208, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9093804359436035, + "num_tokens": 814051593.0, + "step": 20973 + }, + { + "epoch": 2.6681083831573593, + "grad_norm": 2.9166836738586426, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9110419750213623, + "num_tokens": 814089143.0, + "step": 20974 + }, + { + "epoch": 2.6682355934359494, + "grad_norm": 2.978436231613159, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.8998667597770691, + "num_tokens": 814129303.0, + "step": 20975 + }, + { + "epoch": 2.6683628037145404, + "grad_norm": 2.7274136543273926, + "learning_rate": 1e-06, + "loss": 0.2322, + "mean_token_accuracy": 0.9139211177825928, + "num_tokens": 814166858.0, + "step": 20976 + }, + { + "epoch": 2.6684900139931305, + "grad_norm": 2.900235652923584, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.90079665184021, + "num_tokens": 814207555.0, + "step": 20977 + }, + { + "epoch": 2.668617224271721, + "grad_norm": 2.995361566543579, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9012004137039185, + "num_tokens": 814245384.0, + "step": 20978 + }, + { + "epoch": 2.6687444345503115, + "grad_norm": 2.850947856903076, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9066058993339539, + "num_tokens": 814279981.0, + "step": 20979 + }, + { + "epoch": 2.668871644828902, + "grad_norm": 2.673415422439575, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9043166637420654, + "num_tokens": 814323007.0, + "step": 20980 + }, + { + "epoch": 2.6689988551074926, + "grad_norm": 2.9451863765716553, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8981781601905823, + "num_tokens": 814362829.0, + "step": 20981 + }, + { + "epoch": 2.669126065386083, + "grad_norm": 2.6991477012634277, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.903313159942627, + "num_tokens": 814404198.0, + "step": 20982 + }, + { + "epoch": 2.6692532756646736, + "grad_norm": 2.8111836910247803, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9043874740600586, + "num_tokens": 814443294.0, + "step": 20983 + }, + { + "epoch": 2.669380485943264, + "grad_norm": 2.86745285987854, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9060103893280029, + "num_tokens": 814480711.0, + "step": 20984 + }, + { + "epoch": 2.6695076962218547, + "grad_norm": 2.9265339374542236, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9009988307952881, + "num_tokens": 814519697.0, + "step": 20985 + }, + { + "epoch": 2.669634906500445, + "grad_norm": 2.5006260871887207, + "learning_rate": 1e-06, + "loss": 0.2193, + "mean_token_accuracy": 0.9180053472518921, + "num_tokens": 814560584.0, + "step": 20986 + }, + { + "epoch": 2.6697621167790357, + "grad_norm": 2.9342987537384033, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.8990957736968994, + "num_tokens": 814596616.0, + "step": 20987 + }, + { + "epoch": 2.6698893270576263, + "grad_norm": 2.9152300357818604, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8836581707000732, + "num_tokens": 814640947.0, + "step": 20988 + }, + { + "epoch": 2.670016537336217, + "grad_norm": 3.0082671642303467, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9120833277702332, + "num_tokens": 814675207.0, + "step": 20989 + }, + { + "epoch": 2.6701437476148073, + "grad_norm": 3.1652426719665527, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.902512788772583, + "num_tokens": 814706435.0, + "step": 20990 + }, + { + "epoch": 2.670270957893398, + "grad_norm": 2.7410309314727783, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9112650156021118, + "num_tokens": 814749384.0, + "step": 20991 + }, + { + "epoch": 2.6703981681719884, + "grad_norm": 3.26416015625, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8958253860473633, + "num_tokens": 814785014.0, + "step": 20992 + }, + { + "epoch": 2.670525378450579, + "grad_norm": 3.1750569343566895, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9029018878936768, + "num_tokens": 814816855.0, + "step": 20993 + }, + { + "epoch": 2.6706525887291694, + "grad_norm": 2.945826530456543, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9033585786819458, + "num_tokens": 814852163.0, + "step": 20994 + }, + { + "epoch": 2.67077979900776, + "grad_norm": 2.9529857635498047, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.8994183540344238, + "num_tokens": 814889738.0, + "step": 20995 + }, + { + "epoch": 2.6709070092863505, + "grad_norm": 3.057871103286743, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9148558378219604, + "num_tokens": 814924096.0, + "step": 20996 + }, + { + "epoch": 2.671034219564941, + "grad_norm": 3.0261456966400146, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9032006859779358, + "num_tokens": 814960689.0, + "step": 20997 + }, + { + "epoch": 2.671161429843531, + "grad_norm": 2.698930501937866, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9087159633636475, + "num_tokens": 815002788.0, + "step": 20998 + }, + { + "epoch": 2.671288640122122, + "grad_norm": 2.6660051345825195, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9062731862068176, + "num_tokens": 815046781.0, + "step": 20999 + }, + { + "epoch": 2.671415850400712, + "grad_norm": 2.753095865249634, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9059193134307861, + "num_tokens": 815085659.0, + "step": 21000 + }, + { + "epoch": 2.671543060679303, + "grad_norm": 3.1163434982299805, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9029444456100464, + "num_tokens": 815120495.0, + "step": 21001 + }, + { + "epoch": 2.6716702709578932, + "grad_norm": 2.985323429107666, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.899566113948822, + "num_tokens": 815156115.0, + "step": 21002 + }, + { + "epoch": 2.6717974812364838, + "grad_norm": 2.7592968940734863, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9099620580673218, + "num_tokens": 815195021.0, + "step": 21003 + }, + { + "epoch": 2.6719246915150743, + "grad_norm": 3.0671749114990234, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9058934450149536, + "num_tokens": 815231461.0, + "step": 21004 + }, + { + "epoch": 2.672051901793665, + "grad_norm": 2.807910442352295, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9080687761306763, + "num_tokens": 815270872.0, + "step": 21005 + }, + { + "epoch": 2.6721791120722553, + "grad_norm": 3.050490617752075, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8860714435577393, + "num_tokens": 815314774.0, + "step": 21006 + }, + { + "epoch": 2.672306322350846, + "grad_norm": 3.023353338241577, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9050598740577698, + "num_tokens": 815350836.0, + "step": 21007 + }, + { + "epoch": 2.6724335326294364, + "grad_norm": 2.89101243019104, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9069677591323853, + "num_tokens": 815390562.0, + "step": 21008 + }, + { + "epoch": 2.672560742908027, + "grad_norm": 2.9016001224517822, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9118425846099854, + "num_tokens": 815426540.0, + "step": 21009 + }, + { + "epoch": 2.6726879531866174, + "grad_norm": 2.7011260986328125, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9051413536071777, + "num_tokens": 815469366.0, + "step": 21010 + }, + { + "epoch": 2.672815163465208, + "grad_norm": 2.7968289852142334, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9049196243286133, + "num_tokens": 815508379.0, + "step": 21011 + }, + { + "epoch": 2.6729423737437985, + "grad_norm": 2.6803817749023438, + "learning_rate": 1e-06, + "loss": 0.2411, + "mean_token_accuracy": 0.9101956486701965, + "num_tokens": 815545735.0, + "step": 21012 + }, + { + "epoch": 2.673069584022389, + "grad_norm": 3.027630090713501, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9075658321380615, + "num_tokens": 815581613.0, + "step": 21013 + }, + { + "epoch": 2.6731967943009796, + "grad_norm": 3.005760431289673, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9034211039543152, + "num_tokens": 815617301.0, + "step": 21014 + }, + { + "epoch": 2.67332400457957, + "grad_norm": 2.888425350189209, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9048033952713013, + "num_tokens": 815652817.0, + "step": 21015 + }, + { + "epoch": 2.6734512148581606, + "grad_norm": 2.9019265174865723, + "learning_rate": 1e-06, + "loss": 0.2794, + "mean_token_accuracy": 0.8995048999786377, + "num_tokens": 815693101.0, + "step": 21016 + }, + { + "epoch": 2.673578425136751, + "grad_norm": 3.1056745052337646, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8904379606246948, + "num_tokens": 815732251.0, + "step": 21017 + }, + { + "epoch": 2.6737056354153417, + "grad_norm": 2.7340211868286133, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9111778736114502, + "num_tokens": 815775384.0, + "step": 21018 + }, + { + "epoch": 2.673832845693932, + "grad_norm": 2.910311460494995, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.8996458649635315, + "num_tokens": 815818357.0, + "step": 21019 + }, + { + "epoch": 2.6739600559725227, + "grad_norm": 2.7822744846343994, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9136489629745483, + "num_tokens": 815855944.0, + "step": 21020 + }, + { + "epoch": 2.674087266251113, + "grad_norm": 2.6730196475982666, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9037882089614868, + "num_tokens": 815900345.0, + "step": 21021 + }, + { + "epoch": 2.674214476529704, + "grad_norm": 2.5720624923706055, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9126278162002563, + "num_tokens": 815942309.0, + "step": 21022 + }, + { + "epoch": 2.674341686808294, + "grad_norm": 3.0701053142547607, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9052493572235107, + "num_tokens": 815976739.0, + "step": 21023 + }, + { + "epoch": 2.674468897086885, + "grad_norm": 2.807598829269409, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9071935415267944, + "num_tokens": 816013609.0, + "step": 21024 + }, + { + "epoch": 2.674596107365475, + "grad_norm": 2.765540361404419, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9080582857131958, + "num_tokens": 816052709.0, + "step": 21025 + }, + { + "epoch": 2.674723317644066, + "grad_norm": 2.971581220626831, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9018888473510742, + "num_tokens": 816092580.0, + "step": 21026 + }, + { + "epoch": 2.674850527922656, + "grad_norm": 3.081759452819824, + "learning_rate": 1e-06, + "loss": 0.2339, + "mean_token_accuracy": 0.9118121862411499, + "num_tokens": 816124654.0, + "step": 21027 + }, + { + "epoch": 2.6749777382012465, + "grad_norm": 2.893371820449829, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.902323842048645, + "num_tokens": 816161320.0, + "step": 21028 + }, + { + "epoch": 2.675104948479837, + "grad_norm": 2.839453935623169, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9019461274147034, + "num_tokens": 816202133.0, + "step": 21029 + }, + { + "epoch": 2.6752321587584276, + "grad_norm": 2.8323299884796143, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9069798588752747, + "num_tokens": 816241611.0, + "step": 21030 + }, + { + "epoch": 2.675359369037018, + "grad_norm": 2.872373342514038, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.891846239566803, + "num_tokens": 816284251.0, + "step": 21031 + }, + { + "epoch": 2.6754865793156086, + "grad_norm": 3.1311168670654297, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9065176844596863, + "num_tokens": 816319182.0, + "step": 21032 + }, + { + "epoch": 2.675613789594199, + "grad_norm": 3.0022568702697754, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9090102314949036, + "num_tokens": 816354890.0, + "step": 21033 + }, + { + "epoch": 2.6757409998727897, + "grad_norm": 3.055248498916626, + "learning_rate": 1e-06, + "loss": 0.2487, + "mean_token_accuracy": 0.9080275297164917, + "num_tokens": 816390851.0, + "step": 21034 + }, + { + "epoch": 2.67586821015138, + "grad_norm": 2.650420665740967, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9095431566238403, + "num_tokens": 816432766.0, + "step": 21035 + }, + { + "epoch": 2.6759954204299707, + "grad_norm": 2.811519145965576, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8976292610168457, + "num_tokens": 816472046.0, + "step": 21036 + }, + { + "epoch": 2.6761226307085613, + "grad_norm": 2.6009604930877686, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9119143486022949, + "num_tokens": 816512325.0, + "step": 21037 + }, + { + "epoch": 2.676249840987152, + "grad_norm": 2.7865755558013916, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8968144059181213, + "num_tokens": 816557198.0, + "step": 21038 + }, + { + "epoch": 2.6763770512657423, + "grad_norm": 2.722184419631958, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9042747020721436, + "num_tokens": 816599143.0, + "step": 21039 + }, + { + "epoch": 2.676504261544333, + "grad_norm": 2.721132755279541, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9094013571739197, + "num_tokens": 816640585.0, + "step": 21040 + }, + { + "epoch": 2.6766314718229234, + "grad_norm": 2.730030059814453, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9019622206687927, + "num_tokens": 816685023.0, + "step": 21041 + }, + { + "epoch": 2.676758682101514, + "grad_norm": 2.9388628005981445, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.8997734785079956, + "num_tokens": 816721553.0, + "step": 21042 + }, + { + "epoch": 2.6768858923801044, + "grad_norm": 2.8247175216674805, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9129478931427002, + "num_tokens": 816756784.0, + "step": 21043 + }, + { + "epoch": 2.677013102658695, + "grad_norm": 3.1703381538391113, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.9026086330413818, + "num_tokens": 816789169.0, + "step": 21044 + }, + { + "epoch": 2.6771403129372855, + "grad_norm": 3.0300652980804443, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8934487104415894, + "num_tokens": 816829196.0, + "step": 21045 + }, + { + "epoch": 2.6772675232158756, + "grad_norm": 3.220332145690918, + "learning_rate": 1e-06, + "loss": 0.3097, + "mean_token_accuracy": 0.8912383317947388, + "num_tokens": 816867657.0, + "step": 21046 + }, + { + "epoch": 2.6773947334944665, + "grad_norm": 2.762321710586548, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9097402095794678, + "num_tokens": 816907518.0, + "step": 21047 + }, + { + "epoch": 2.6775219437730566, + "grad_norm": 3.2810096740722656, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9010252952575684, + "num_tokens": 816939543.0, + "step": 21048 + }, + { + "epoch": 2.6776491540516476, + "grad_norm": 2.787637948989868, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9107294082641602, + "num_tokens": 816982780.0, + "step": 21049 + }, + { + "epoch": 2.6777763643302377, + "grad_norm": 3.0482234954833984, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8931236863136292, + "num_tokens": 817021341.0, + "step": 21050 + }, + { + "epoch": 2.6779035746088287, + "grad_norm": 2.9412574768066406, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9027566313743591, + "num_tokens": 817059507.0, + "step": 21051 + }, + { + "epoch": 2.6780307848874187, + "grad_norm": 3.177865743637085, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8953779339790344, + "num_tokens": 817096514.0, + "step": 21052 + }, + { + "epoch": 2.6781579951660093, + "grad_norm": 3.041534185409546, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.906427264213562, + "num_tokens": 817135772.0, + "step": 21053 + }, + { + "epoch": 2.6782852054446, + "grad_norm": 3.246094226837158, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8979138135910034, + "num_tokens": 817172044.0, + "step": 21054 + }, + { + "epoch": 2.6784124157231903, + "grad_norm": 2.924079656600952, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9004648327827454, + "num_tokens": 817213768.0, + "step": 21055 + }, + { + "epoch": 2.678539626001781, + "grad_norm": 2.8319928646087646, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9084845781326294, + "num_tokens": 817253051.0, + "step": 21056 + }, + { + "epoch": 2.6786668362803714, + "grad_norm": 2.6283841133117676, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9052311182022095, + "num_tokens": 817295407.0, + "step": 21057 + }, + { + "epoch": 2.678794046558962, + "grad_norm": 3.0037198066711426, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9077761769294739, + "num_tokens": 817331642.0, + "step": 21058 + }, + { + "epoch": 2.6789212568375524, + "grad_norm": 2.8445935249328613, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9099329710006714, + "num_tokens": 817368770.0, + "step": 21059 + }, + { + "epoch": 2.679048467116143, + "grad_norm": 3.0241189002990723, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9016762375831604, + "num_tokens": 817410399.0, + "step": 21060 + }, + { + "epoch": 2.6791756773947335, + "grad_norm": 2.898620843887329, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9021403789520264, + "num_tokens": 817449897.0, + "step": 21061 + }, + { + "epoch": 2.679302887673324, + "grad_norm": 2.4334378242492676, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9119497537612915, + "num_tokens": 817497572.0, + "step": 21062 + }, + { + "epoch": 2.6794300979519146, + "grad_norm": 3.0032732486724854, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8972715735435486, + "num_tokens": 817535761.0, + "step": 21063 + }, + { + "epoch": 2.679557308230505, + "grad_norm": 2.623853921890259, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9088024497032166, + "num_tokens": 817577836.0, + "step": 21064 + }, + { + "epoch": 2.6796845185090956, + "grad_norm": 2.76633882522583, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9061428308486938, + "num_tokens": 817620524.0, + "step": 21065 + }, + { + "epoch": 2.679811728787686, + "grad_norm": 2.973857879638672, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.895919919013977, + "num_tokens": 817661822.0, + "step": 21066 + }, + { + "epoch": 2.6799389390662767, + "grad_norm": 2.761507034301758, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9044755697250366, + "num_tokens": 817703528.0, + "step": 21067 + }, + { + "epoch": 2.680066149344867, + "grad_norm": 3.154663324356079, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8990459442138672, + "num_tokens": 817738685.0, + "step": 21068 + }, + { + "epoch": 2.6801933596234577, + "grad_norm": 2.830223321914673, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9010732173919678, + "num_tokens": 817778804.0, + "step": 21069 + }, + { + "epoch": 2.6803205699020483, + "grad_norm": 2.910238265991211, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8928883671760559, + "num_tokens": 817820801.0, + "step": 21070 + }, + { + "epoch": 2.6804477801806383, + "grad_norm": 2.9596378803253174, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.900272786617279, + "num_tokens": 817859108.0, + "step": 21071 + }, + { + "epoch": 2.6805749904592293, + "grad_norm": 2.769414186477661, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8979653120040894, + "num_tokens": 817902895.0, + "step": 21072 + }, + { + "epoch": 2.6807022007378194, + "grad_norm": 2.594851016998291, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.8987312316894531, + "num_tokens": 817948256.0, + "step": 21073 + }, + { + "epoch": 2.6808294110164104, + "grad_norm": 2.6713433265686035, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9137375950813293, + "num_tokens": 817990578.0, + "step": 21074 + }, + { + "epoch": 2.6809566212950005, + "grad_norm": 2.7948288917541504, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.9166398644447327, + "num_tokens": 818023471.0, + "step": 21075 + }, + { + "epoch": 2.681083831573591, + "grad_norm": 3.1501147747039795, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8994455337524414, + "num_tokens": 818057993.0, + "step": 21076 + }, + { + "epoch": 2.6812110418521815, + "grad_norm": 2.7443196773529053, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9097527861595154, + "num_tokens": 818093386.0, + "step": 21077 + }, + { + "epoch": 2.681338252130772, + "grad_norm": 2.6871988773345947, + "learning_rate": 1e-06, + "loss": 0.2317, + "mean_token_accuracy": 0.9114602208137512, + "num_tokens": 818134118.0, + "step": 21078 + }, + { + "epoch": 2.6814654624093626, + "grad_norm": 3.163557767868042, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8959488868713379, + "num_tokens": 818173665.0, + "step": 21079 + }, + { + "epoch": 2.681592672687953, + "grad_norm": 2.9132542610168457, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9034238457679749, + "num_tokens": 818211280.0, + "step": 21080 + }, + { + "epoch": 2.6817198829665436, + "grad_norm": 2.8606455326080322, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9032702445983887, + "num_tokens": 818248587.0, + "step": 21081 + }, + { + "epoch": 2.681847093245134, + "grad_norm": 3.0832266807556152, + "learning_rate": 1e-06, + "loss": 0.2734, + "mean_token_accuracy": 0.8994250297546387, + "num_tokens": 818281699.0, + "step": 21082 + }, + { + "epoch": 2.6819743035237247, + "grad_norm": 2.769331932067871, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9056591391563416, + "num_tokens": 818319382.0, + "step": 21083 + }, + { + "epoch": 2.682101513802315, + "grad_norm": 2.9096546173095703, + "learning_rate": 1e-06, + "loss": 0.226, + "mean_token_accuracy": 0.9155555367469788, + "num_tokens": 818356741.0, + "step": 21084 + }, + { + "epoch": 2.6822287240809057, + "grad_norm": 2.8568475246429443, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9021595120429993, + "num_tokens": 818397708.0, + "step": 21085 + }, + { + "epoch": 2.6823559343594963, + "grad_norm": 3.108555555343628, + "learning_rate": 1e-06, + "loss": 0.2932, + "mean_token_accuracy": 0.8943558931350708, + "num_tokens": 818439333.0, + "step": 21086 + }, + { + "epoch": 2.682483144638087, + "grad_norm": 3.128521203994751, + "learning_rate": 1e-06, + "loss": 0.2364, + "mean_token_accuracy": 0.9128220081329346, + "num_tokens": 818471744.0, + "step": 21087 + }, + { + "epoch": 2.6826103549166773, + "grad_norm": 3.012205123901367, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.8982604146003723, + "num_tokens": 818511797.0, + "step": 21088 + }, + { + "epoch": 2.682737565195268, + "grad_norm": 3.1429293155670166, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9003506302833557, + "num_tokens": 818546610.0, + "step": 21089 + }, + { + "epoch": 2.6828647754738584, + "grad_norm": 2.921808958053589, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.893854022026062, + "num_tokens": 818590637.0, + "step": 21090 + }, + { + "epoch": 2.682991985752449, + "grad_norm": 3.102185010910034, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8975962996482849, + "num_tokens": 818627012.0, + "step": 21091 + }, + { + "epoch": 2.6831191960310394, + "grad_norm": 3.0578033924102783, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9019317030906677, + "num_tokens": 818661976.0, + "step": 21092 + }, + { + "epoch": 2.68324640630963, + "grad_norm": 3.1336326599121094, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.8998267650604248, + "num_tokens": 818698290.0, + "step": 21093 + }, + { + "epoch": 2.6833736165882205, + "grad_norm": 2.5414633750915527, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9126478433609009, + "num_tokens": 818741633.0, + "step": 21094 + }, + { + "epoch": 2.683500826866811, + "grad_norm": 3.2701120376586914, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9059838056564331, + "num_tokens": 818770233.0, + "step": 21095 + }, + { + "epoch": 2.683628037145401, + "grad_norm": 3.061112642288208, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8984192609786987, + "num_tokens": 818804555.0, + "step": 21096 + }, + { + "epoch": 2.683755247423992, + "grad_norm": 2.538423538208008, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9061130881309509, + "num_tokens": 818850896.0, + "step": 21097 + }, + { + "epoch": 2.683882457702582, + "grad_norm": 2.554412364959717, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9048271179199219, + "num_tokens": 818895915.0, + "step": 21098 + }, + { + "epoch": 2.684009667981173, + "grad_norm": 3.251120090484619, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8951904773712158, + "num_tokens": 818928789.0, + "step": 21099 + }, + { + "epoch": 2.684136878259763, + "grad_norm": 2.8569047451019287, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9117082357406616, + "num_tokens": 818966550.0, + "step": 21100 + }, + { + "epoch": 2.6842640885383537, + "grad_norm": 2.620682716369629, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.904003381729126, + "num_tokens": 819010681.0, + "step": 21101 + }, + { + "epoch": 2.6843912988169443, + "grad_norm": 2.7373650074005127, + "learning_rate": 1e-06, + "loss": 0.2376, + "mean_token_accuracy": 0.9141737222671509, + "num_tokens": 819049715.0, + "step": 21102 + }, + { + "epoch": 2.684518509095535, + "grad_norm": 2.7639405727386475, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9042170643806458, + "num_tokens": 819094714.0, + "step": 21103 + }, + { + "epoch": 2.6846457193741253, + "grad_norm": 2.690051555633545, + "learning_rate": 1e-06, + "loss": 0.2217, + "mean_token_accuracy": 0.9155076742172241, + "num_tokens": 819132646.0, + "step": 21104 + }, + { + "epoch": 2.684772929652716, + "grad_norm": 2.8787057399749756, + "learning_rate": 1e-06, + "loss": 0.229, + "mean_token_accuracy": 0.9136887788772583, + "num_tokens": 819169517.0, + "step": 21105 + }, + { + "epoch": 2.6849001399313064, + "grad_norm": 3.023049831390381, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9012589454650879, + "num_tokens": 819206878.0, + "step": 21106 + }, + { + "epoch": 2.685027350209897, + "grad_norm": 2.834277391433716, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9124651551246643, + "num_tokens": 819245800.0, + "step": 21107 + }, + { + "epoch": 2.6851545604884874, + "grad_norm": 2.8111886978149414, + "learning_rate": 1e-06, + "loss": 0.2487, + "mean_token_accuracy": 0.9090744256973267, + "num_tokens": 819282701.0, + "step": 21108 + }, + { + "epoch": 2.685281770767078, + "grad_norm": 2.6542320251464844, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.905713677406311, + "num_tokens": 819323050.0, + "step": 21109 + }, + { + "epoch": 2.6854089810456685, + "grad_norm": 3.0372474193573, + "learning_rate": 1e-06, + "loss": 0.292, + "mean_token_accuracy": 0.8947535753250122, + "num_tokens": 819361922.0, + "step": 21110 + }, + { + "epoch": 2.685536191324259, + "grad_norm": 2.9242892265319824, + "learning_rate": 1e-06, + "loss": 0.3041, + "mean_token_accuracy": 0.8879965543746948, + "num_tokens": 819404434.0, + "step": 21111 + }, + { + "epoch": 2.6856634016028496, + "grad_norm": 3.1150126457214355, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.898335337638855, + "num_tokens": 819439739.0, + "step": 21112 + }, + { + "epoch": 2.68579061188144, + "grad_norm": 2.8127317428588867, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9048827886581421, + "num_tokens": 819479814.0, + "step": 21113 + }, + { + "epoch": 2.6859178221600306, + "grad_norm": 2.8021938800811768, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9071388840675354, + "num_tokens": 819520947.0, + "step": 21114 + }, + { + "epoch": 2.686045032438621, + "grad_norm": 2.9042232036590576, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9025700092315674, + "num_tokens": 819558826.0, + "step": 21115 + }, + { + "epoch": 2.6861722427172117, + "grad_norm": 3.1897873878479004, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.89830082654953, + "num_tokens": 819591921.0, + "step": 21116 + }, + { + "epoch": 2.686299452995802, + "grad_norm": 2.979766845703125, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9024624228477478, + "num_tokens": 819628336.0, + "step": 21117 + }, + { + "epoch": 2.6864266632743927, + "grad_norm": 2.8674957752227783, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9092212319374084, + "num_tokens": 819666431.0, + "step": 21118 + }, + { + "epoch": 2.686553873552983, + "grad_norm": 3.1295418739318848, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.905958354473114, + "num_tokens": 819698598.0, + "step": 21119 + }, + { + "epoch": 2.686681083831574, + "grad_norm": 3.0614988803863525, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8973511457443237, + "num_tokens": 819738685.0, + "step": 21120 + }, + { + "epoch": 2.686808294110164, + "grad_norm": 2.8347630500793457, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8978030681610107, + "num_tokens": 819781298.0, + "step": 21121 + }, + { + "epoch": 2.686935504388755, + "grad_norm": 2.9036097526550293, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9036623239517212, + "num_tokens": 819819803.0, + "step": 21122 + }, + { + "epoch": 2.687062714667345, + "grad_norm": 2.9207510948181152, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8945443630218506, + "num_tokens": 819859794.0, + "step": 21123 + }, + { + "epoch": 2.687189924945936, + "grad_norm": 3.0041420459747314, + "learning_rate": 1e-06, + "loss": 0.2746, + "mean_token_accuracy": 0.8998030424118042, + "num_tokens": 819899601.0, + "step": 21124 + }, + { + "epoch": 2.687317135224526, + "grad_norm": 2.947296142578125, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.9133201241493225, + "num_tokens": 819932229.0, + "step": 21125 + }, + { + "epoch": 2.6874443455031165, + "grad_norm": 2.9686994552612305, + "learning_rate": 1e-06, + "loss": 0.2857, + "mean_token_accuracy": 0.8974279761314392, + "num_tokens": 819970150.0, + "step": 21126 + }, + { + "epoch": 2.687571555781707, + "grad_norm": 2.7676265239715576, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9007992744445801, + "num_tokens": 820012193.0, + "step": 21127 + }, + { + "epoch": 2.6876987660602976, + "grad_norm": 2.9193124771118164, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9031029939651489, + "num_tokens": 820050403.0, + "step": 21128 + }, + { + "epoch": 2.687825976338888, + "grad_norm": 3.002368688583374, + "learning_rate": 1e-06, + "loss": 0.279, + "mean_token_accuracy": 0.8994912505149841, + "num_tokens": 820089930.0, + "step": 21129 + }, + { + "epoch": 2.6879531866174786, + "grad_norm": 2.984369993209839, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.903745174407959, + "num_tokens": 820129758.0, + "step": 21130 + }, + { + "epoch": 2.688080396896069, + "grad_norm": 2.831840753555298, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9100708365440369, + "num_tokens": 820168983.0, + "step": 21131 + }, + { + "epoch": 2.6882076071746597, + "grad_norm": 2.746089220046997, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.906778872013092, + "num_tokens": 820211916.0, + "step": 21132 + }, + { + "epoch": 2.68833481745325, + "grad_norm": 2.9641661643981934, + "learning_rate": 1e-06, + "loss": 0.3087, + "mean_token_accuracy": 0.8882477283477783, + "num_tokens": 820254978.0, + "step": 21133 + }, + { + "epoch": 2.6884620277318407, + "grad_norm": 2.9954922199249268, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.8986769914627075, + "num_tokens": 820290021.0, + "step": 21134 + }, + { + "epoch": 2.6885892380104313, + "grad_norm": 2.989168167114258, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8932288885116577, + "num_tokens": 820328177.0, + "step": 21135 + }, + { + "epoch": 2.688716448289022, + "grad_norm": 2.800657033920288, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9076640009880066, + "num_tokens": 820364628.0, + "step": 21136 + }, + { + "epoch": 2.6888436585676123, + "grad_norm": 3.0735270977020264, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9027135372161865, + "num_tokens": 820401541.0, + "step": 21137 + }, + { + "epoch": 2.688970868846203, + "grad_norm": 2.8703291416168213, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8951111435890198, + "num_tokens": 820442413.0, + "step": 21138 + }, + { + "epoch": 2.6890980791247934, + "grad_norm": 3.0568792819976807, + "learning_rate": 1e-06, + "loss": 0.2188, + "mean_token_accuracy": 0.9167729616165161, + "num_tokens": 820473501.0, + "step": 21139 + }, + { + "epoch": 2.689225289403384, + "grad_norm": 3.043017864227295, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9010195732116699, + "num_tokens": 820508519.0, + "step": 21140 + }, + { + "epoch": 2.6893524996819744, + "grad_norm": 2.9642908573150635, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8949418663978577, + "num_tokens": 820549135.0, + "step": 21141 + }, + { + "epoch": 2.689479709960565, + "grad_norm": 2.9810705184936523, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9026674032211304, + "num_tokens": 820587087.0, + "step": 21142 + }, + { + "epoch": 2.6896069202391555, + "grad_norm": 2.8752994537353516, + "learning_rate": 1e-06, + "loss": 0.2163, + "mean_token_accuracy": 0.9154761433601379, + "num_tokens": 820620335.0, + "step": 21143 + }, + { + "epoch": 2.6897341305177456, + "grad_norm": 3.0068445205688477, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8948831558227539, + "num_tokens": 820659824.0, + "step": 21144 + }, + { + "epoch": 2.6898613407963365, + "grad_norm": 2.8842318058013916, + "learning_rate": 1e-06, + "loss": 0.2487, + "mean_token_accuracy": 0.9102185368537903, + "num_tokens": 820696250.0, + "step": 21145 + }, + { + "epoch": 2.6899885510749266, + "grad_norm": 3.4322855472564697, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.8962583541870117, + "num_tokens": 820727582.0, + "step": 21146 + }, + { + "epoch": 2.6901157613535176, + "grad_norm": 2.766613245010376, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9006176590919495, + "num_tokens": 820772156.0, + "step": 21147 + }, + { + "epoch": 2.6902429716321077, + "grad_norm": 2.6741297245025635, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9059098958969116, + "num_tokens": 820812075.0, + "step": 21148 + }, + { + "epoch": 2.6903701819106987, + "grad_norm": 3.046104669570923, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8979531526565552, + "num_tokens": 820851057.0, + "step": 21149 + }, + { + "epoch": 2.6904973921892887, + "grad_norm": 2.8351123332977295, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9010366201400757, + "num_tokens": 820893107.0, + "step": 21150 + }, + { + "epoch": 2.6906246024678793, + "grad_norm": 3.038642644882202, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9063529968261719, + "num_tokens": 820923546.0, + "step": 21151 + }, + { + "epoch": 2.69075181274647, + "grad_norm": 2.97688889503479, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8977301120758057, + "num_tokens": 820960160.0, + "step": 21152 + }, + { + "epoch": 2.6908790230250603, + "grad_norm": 2.793292284011841, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9063937067985535, + "num_tokens": 821001416.0, + "step": 21153 + }, + { + "epoch": 2.691006233303651, + "grad_norm": 3.035466432571411, + "learning_rate": 1e-06, + "loss": 0.3111, + "mean_token_accuracy": 0.8853704929351807, + "num_tokens": 821041253.0, + "step": 21154 + }, + { + "epoch": 2.6911334435822414, + "grad_norm": 2.8087382316589355, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.9000515341758728, + "num_tokens": 821082676.0, + "step": 21155 + }, + { + "epoch": 2.691260653860832, + "grad_norm": 2.951789379119873, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9090808629989624, + "num_tokens": 821117719.0, + "step": 21156 + }, + { + "epoch": 2.6913878641394224, + "grad_norm": 2.870288372039795, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9025878310203552, + "num_tokens": 821157290.0, + "step": 21157 + }, + { + "epoch": 2.691515074418013, + "grad_norm": 2.7392003536224365, + "learning_rate": 1e-06, + "loss": 0.242, + "mean_token_accuracy": 0.9111276268959045, + "num_tokens": 821197047.0, + "step": 21158 + }, + { + "epoch": 2.6916422846966035, + "grad_norm": 3.0842645168304443, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8969991207122803, + "num_tokens": 821234969.0, + "step": 21159 + }, + { + "epoch": 2.691769494975194, + "grad_norm": 2.9800117015838623, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.8938883543014526, + "num_tokens": 821275495.0, + "step": 21160 + }, + { + "epoch": 2.6918967052537845, + "grad_norm": 2.971654176712036, + "learning_rate": 1e-06, + "loss": 0.2915, + "mean_token_accuracy": 0.8939310312271118, + "num_tokens": 821317793.0, + "step": 21161 + }, + { + "epoch": 2.692023915532375, + "grad_norm": 2.695286512374878, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9081153869628906, + "num_tokens": 821360546.0, + "step": 21162 + }, + { + "epoch": 2.6921511258109656, + "grad_norm": 2.890333414077759, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.906535804271698, + "num_tokens": 821397773.0, + "step": 21163 + }, + { + "epoch": 2.692278336089556, + "grad_norm": 2.817607879638672, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8992666602134705, + "num_tokens": 821439746.0, + "step": 21164 + }, + { + "epoch": 2.6924055463681467, + "grad_norm": 2.8422257900238037, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9118843078613281, + "num_tokens": 821478507.0, + "step": 21165 + }, + { + "epoch": 2.692532756646737, + "grad_norm": 2.781132221221924, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9004055261611938, + "num_tokens": 821520818.0, + "step": 21166 + }, + { + "epoch": 2.6926599669253277, + "grad_norm": 2.9447457790374756, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9016457200050354, + "num_tokens": 821560576.0, + "step": 21167 + }, + { + "epoch": 2.6927871772039182, + "grad_norm": 3.0204949378967285, + "learning_rate": 1e-06, + "loss": 0.3081, + "mean_token_accuracy": 0.8907384872436523, + "num_tokens": 821598839.0, + "step": 21168 + }, + { + "epoch": 2.6929143874825083, + "grad_norm": 2.6139752864837646, + "learning_rate": 1e-06, + "loss": 0.2059, + "mean_token_accuracy": 0.9238186478614807, + "num_tokens": 821640285.0, + "step": 21169 + }, + { + "epoch": 2.6930415977610993, + "grad_norm": 2.9348549842834473, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9094212055206299, + "num_tokens": 821674147.0, + "step": 21170 + }, + { + "epoch": 2.6931688080396894, + "grad_norm": 2.931661605834961, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9088881015777588, + "num_tokens": 821711576.0, + "step": 21171 + }, + { + "epoch": 2.6932960183182804, + "grad_norm": 2.66892671585083, + "learning_rate": 1e-06, + "loss": 0.2265, + "mean_token_accuracy": 0.9160078763961792, + "num_tokens": 821752928.0, + "step": 21172 + }, + { + "epoch": 2.6934232285968704, + "grad_norm": 2.747490644454956, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9082419276237488, + "num_tokens": 821792726.0, + "step": 21173 + }, + { + "epoch": 2.693550438875461, + "grad_norm": 2.9469242095947266, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.908859372138977, + "num_tokens": 821830580.0, + "step": 21174 + }, + { + "epoch": 2.6936776491540515, + "grad_norm": 3.0494964122772217, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9027242660522461, + "num_tokens": 821864742.0, + "step": 21175 + }, + { + "epoch": 2.693804859432642, + "grad_norm": 3.4291882514953613, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.895557165145874, + "num_tokens": 821898748.0, + "step": 21176 + }, + { + "epoch": 2.6939320697112326, + "grad_norm": 2.5843119621276855, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9028686881065369, + "num_tokens": 821945553.0, + "step": 21177 + }, + { + "epoch": 2.694059279989823, + "grad_norm": 3.128772258758545, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.9128673076629639, + "num_tokens": 821983518.0, + "step": 21178 + }, + { + "epoch": 2.6941864902684136, + "grad_norm": 3.1133837699890137, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9017588496208191, + "num_tokens": 822018804.0, + "step": 21179 + }, + { + "epoch": 2.694313700547004, + "grad_norm": 2.7653446197509766, + "learning_rate": 1e-06, + "loss": 0.2349, + "mean_token_accuracy": 0.9122480154037476, + "num_tokens": 822058953.0, + "step": 21180 + }, + { + "epoch": 2.6944409108255947, + "grad_norm": 2.7902278900146484, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9075756072998047, + "num_tokens": 822095093.0, + "step": 21181 + }, + { + "epoch": 2.694568121104185, + "grad_norm": 2.8647263050079346, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9082856178283691, + "num_tokens": 822133517.0, + "step": 21182 + }, + { + "epoch": 2.6946953313827757, + "grad_norm": 2.814826250076294, + "learning_rate": 1e-06, + "loss": 0.2337, + "mean_token_accuracy": 0.9119318723678589, + "num_tokens": 822168851.0, + "step": 21183 + }, + { + "epoch": 2.6948225416613663, + "grad_norm": 2.6469526290893555, + "learning_rate": 1e-06, + "loss": 0.2283, + "mean_token_accuracy": 0.9153494238853455, + "num_tokens": 822208354.0, + "step": 21184 + }, + { + "epoch": 2.694949751939957, + "grad_norm": 2.8449454307556152, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9074167013168335, + "num_tokens": 822246134.0, + "step": 21185 + }, + { + "epoch": 2.6950769622185473, + "grad_norm": 2.7021515369415283, + "learning_rate": 1e-06, + "loss": 0.2363, + "mean_token_accuracy": 0.9131724238395691, + "num_tokens": 822284810.0, + "step": 21186 + }, + { + "epoch": 2.695204172497138, + "grad_norm": 2.8087892532348633, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9082680940628052, + "num_tokens": 822326575.0, + "step": 21187 + }, + { + "epoch": 2.6953313827757284, + "grad_norm": 3.028331756591797, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.9021131992340088, + "num_tokens": 822367644.0, + "step": 21188 + }, + { + "epoch": 2.695458593054319, + "grad_norm": 2.9553091526031494, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9056042432785034, + "num_tokens": 822407259.0, + "step": 21189 + }, + { + "epoch": 2.6955858033329094, + "grad_norm": 3.011749505996704, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.8985087275505066, + "num_tokens": 822444065.0, + "step": 21190 + }, + { + "epoch": 2.6957130136115, + "grad_norm": 2.843724250793457, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9131132960319519, + "num_tokens": 822479452.0, + "step": 21191 + }, + { + "epoch": 2.6958402238900905, + "grad_norm": 2.9785327911376953, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.8996100425720215, + "num_tokens": 822518150.0, + "step": 21192 + }, + { + "epoch": 2.695967434168681, + "grad_norm": 3.320606231689453, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.8978395462036133, + "num_tokens": 822553397.0, + "step": 21193 + }, + { + "epoch": 2.696094644447271, + "grad_norm": 3.0638980865478516, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8934599757194519, + "num_tokens": 822592852.0, + "step": 21194 + }, + { + "epoch": 2.696221854725862, + "grad_norm": 2.892397165298462, + "learning_rate": 1e-06, + "loss": 0.2827, + "mean_token_accuracy": 0.8976792693138123, + "num_tokens": 822637998.0, + "step": 21195 + }, + { + "epoch": 2.696349065004452, + "grad_norm": 2.984980344772339, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.8992139101028442, + "num_tokens": 822677008.0, + "step": 21196 + }, + { + "epoch": 2.696476275283043, + "grad_norm": 2.7788851261138916, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.900250256061554, + "num_tokens": 822718790.0, + "step": 21197 + }, + { + "epoch": 2.696603485561633, + "grad_norm": 3.0138132572174072, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9034737944602966, + "num_tokens": 822752984.0, + "step": 21198 + }, + { + "epoch": 2.6967306958402237, + "grad_norm": 2.8620216846466064, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.90907883644104, + "num_tokens": 822788194.0, + "step": 21199 + }, + { + "epoch": 2.6968579061188143, + "grad_norm": 3.1396896839141846, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9060924053192139, + "num_tokens": 822820732.0, + "step": 21200 + }, + { + "epoch": 2.696985116397405, + "grad_norm": 2.7199723720550537, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9132471084594727, + "num_tokens": 822860138.0, + "step": 21201 + }, + { + "epoch": 2.6971123266759953, + "grad_norm": 3.18997859954834, + "learning_rate": 1e-06, + "loss": 0.2715, + "mean_token_accuracy": 0.8952201008796692, + "num_tokens": 822895411.0, + "step": 21202 + }, + { + "epoch": 2.697239536954586, + "grad_norm": 2.8873610496520996, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.8968669176101685, + "num_tokens": 822935961.0, + "step": 21203 + }, + { + "epoch": 2.6973667472331764, + "grad_norm": 3.0979201793670654, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9024779796600342, + "num_tokens": 822969435.0, + "step": 21204 + }, + { + "epoch": 2.697493957511767, + "grad_norm": 3.0491321086883545, + "learning_rate": 1e-06, + "loss": 0.2443, + "mean_token_accuracy": 0.9068807363510132, + "num_tokens": 823001637.0, + "step": 21205 + }, + { + "epoch": 2.6976211677903574, + "grad_norm": 3.093400716781616, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9041617512702942, + "num_tokens": 823036848.0, + "step": 21206 + }, + { + "epoch": 2.697748378068948, + "grad_norm": 3.2671024799346924, + "learning_rate": 1e-06, + "loss": 0.306, + "mean_token_accuracy": 0.8894916772842407, + "num_tokens": 823073037.0, + "step": 21207 + }, + { + "epoch": 2.6978755883475385, + "grad_norm": 2.8094215393066406, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9159430265426636, + "num_tokens": 823113469.0, + "step": 21208 + }, + { + "epoch": 2.698002798626129, + "grad_norm": 3.1453168392181396, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9022274017333984, + "num_tokens": 823147861.0, + "step": 21209 + }, + { + "epoch": 2.6981300089047195, + "grad_norm": 2.8538150787353516, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9044016003608704, + "num_tokens": 823187168.0, + "step": 21210 + }, + { + "epoch": 2.69825721918331, + "grad_norm": 2.553590774536133, + "learning_rate": 1e-06, + "loss": 0.2244, + "mean_token_accuracy": 0.9142962694168091, + "num_tokens": 823228242.0, + "step": 21211 + }, + { + "epoch": 2.6983844294619006, + "grad_norm": 2.835965871810913, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.911677360534668, + "num_tokens": 823264507.0, + "step": 21212 + }, + { + "epoch": 2.698511639740491, + "grad_norm": 2.8034329414367676, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9070730805397034, + "num_tokens": 823301274.0, + "step": 21213 + }, + { + "epoch": 2.6986388500190817, + "grad_norm": 2.8148233890533447, + "learning_rate": 1e-06, + "loss": 0.2445, + "mean_token_accuracy": 0.9065209627151489, + "num_tokens": 823339831.0, + "step": 21214 + }, + { + "epoch": 2.698766060297672, + "grad_norm": 2.7480366230010986, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9069299697875977, + "num_tokens": 823379905.0, + "step": 21215 + }, + { + "epoch": 2.6988932705762627, + "grad_norm": 3.438403844833374, + "learning_rate": 1e-06, + "loss": 0.2959, + "mean_token_accuracy": 0.8937400579452515, + "num_tokens": 823412466.0, + "step": 21216 + }, + { + "epoch": 2.699020480854853, + "grad_norm": 3.147780179977417, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8974061012268066, + "num_tokens": 823449209.0, + "step": 21217 + }, + { + "epoch": 2.6991476911334438, + "grad_norm": 2.9111979007720947, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9007123112678528, + "num_tokens": 823487242.0, + "step": 21218 + }, + { + "epoch": 2.699274901412034, + "grad_norm": 3.1109068393707275, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8958336114883423, + "num_tokens": 823523109.0, + "step": 21219 + }, + { + "epoch": 2.699402111690625, + "grad_norm": 2.808739423751831, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.8997821807861328, + "num_tokens": 823562967.0, + "step": 21220 + }, + { + "epoch": 2.699529321969215, + "grad_norm": 2.570110559463501, + "learning_rate": 1e-06, + "loss": 0.2304, + "mean_token_accuracy": 0.9145537614822388, + "num_tokens": 823607335.0, + "step": 21221 + }, + { + "epoch": 2.699656532247806, + "grad_norm": 2.673466444015503, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9023904800415039, + "num_tokens": 823652733.0, + "step": 21222 + }, + { + "epoch": 2.699783742526396, + "grad_norm": 3.068655490875244, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9091781377792358, + "num_tokens": 823685203.0, + "step": 21223 + }, + { + "epoch": 2.6999109528049865, + "grad_norm": 2.590664863586426, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9078899621963501, + "num_tokens": 823730720.0, + "step": 21224 + }, + { + "epoch": 2.700038163083577, + "grad_norm": 2.861546039581299, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9028563499450684, + "num_tokens": 823768392.0, + "step": 21225 + }, + { + "epoch": 2.7001653733621676, + "grad_norm": 2.5552010536193848, + "learning_rate": 1e-06, + "loss": 0.1954, + "mean_token_accuracy": 0.9263756275177002, + "num_tokens": 823810337.0, + "step": 21226 + }, + { + "epoch": 2.700292583640758, + "grad_norm": 2.7581512928009033, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9069411754608154, + "num_tokens": 823849502.0, + "step": 21227 + }, + { + "epoch": 2.7004197939193486, + "grad_norm": 2.77925705909729, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9052786827087402, + "num_tokens": 823891731.0, + "step": 21228 + }, + { + "epoch": 2.700547004197939, + "grad_norm": 2.8913803100585938, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9058685302734375, + "num_tokens": 823930057.0, + "step": 21229 + }, + { + "epoch": 2.7006742144765297, + "grad_norm": 2.6002840995788574, + "learning_rate": 1e-06, + "loss": 0.2234, + "mean_token_accuracy": 0.916869044303894, + "num_tokens": 823968435.0, + "step": 21230 + }, + { + "epoch": 2.70080142475512, + "grad_norm": 3.0726218223571777, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8947272896766663, + "num_tokens": 824005817.0, + "step": 21231 + }, + { + "epoch": 2.7009286350337107, + "grad_norm": 2.537550687789917, + "learning_rate": 1e-06, + "loss": 0.2311, + "mean_token_accuracy": 0.9138290882110596, + "num_tokens": 824051403.0, + "step": 21232 + }, + { + "epoch": 2.7010558453123013, + "grad_norm": 3.1595969200134277, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8935743570327759, + "num_tokens": 824094886.0, + "step": 21233 + }, + { + "epoch": 2.701183055590892, + "grad_norm": 2.7768144607543945, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.9028054475784302, + "num_tokens": 824138164.0, + "step": 21234 + }, + { + "epoch": 2.7013102658694823, + "grad_norm": 2.9903252124786377, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.907451331615448, + "num_tokens": 824171997.0, + "step": 21235 + }, + { + "epoch": 2.701437476148073, + "grad_norm": 3.2022504806518555, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9043428897857666, + "num_tokens": 824203264.0, + "step": 21236 + }, + { + "epoch": 2.7015646864266634, + "grad_norm": 2.5960817337036133, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9113585948944092, + "num_tokens": 824245552.0, + "step": 21237 + }, + { + "epoch": 2.701691896705254, + "grad_norm": 2.7027173042297363, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.9007520079612732, + "num_tokens": 824291479.0, + "step": 21238 + }, + { + "epoch": 2.7018191069838444, + "grad_norm": 2.5373663902282715, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9123144149780273, + "num_tokens": 824334271.0, + "step": 21239 + }, + { + "epoch": 2.701946317262435, + "grad_norm": 2.667832374572754, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9097067713737488, + "num_tokens": 824378006.0, + "step": 21240 + }, + { + "epoch": 2.7020735275410255, + "grad_norm": 2.8847897052764893, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9070237874984741, + "num_tokens": 824415136.0, + "step": 21241 + }, + { + "epoch": 2.7022007378196156, + "grad_norm": 3.132352113723755, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9019576907157898, + "num_tokens": 824449053.0, + "step": 21242 + }, + { + "epoch": 2.7023279480982065, + "grad_norm": 2.895108461380005, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8986799716949463, + "num_tokens": 824490168.0, + "step": 21243 + }, + { + "epoch": 2.7024551583767966, + "grad_norm": 2.8713326454162598, + "learning_rate": 1e-06, + "loss": 0.2272, + "mean_token_accuracy": 0.9139107465744019, + "num_tokens": 824526590.0, + "step": 21244 + }, + { + "epoch": 2.7025823686553876, + "grad_norm": 2.9180266857147217, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9015181064605713, + "num_tokens": 824567826.0, + "step": 21245 + }, + { + "epoch": 2.7027095789339777, + "grad_norm": 2.5756890773773193, + "learning_rate": 1e-06, + "loss": 0.229, + "mean_token_accuracy": 0.9142019152641296, + "num_tokens": 824610592.0, + "step": 21246 + }, + { + "epoch": 2.7028367892125686, + "grad_norm": 3.076488494873047, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.891201913356781, + "num_tokens": 824652799.0, + "step": 21247 + }, + { + "epoch": 2.7029639994911587, + "grad_norm": 3.1175317764282227, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9067026972770691, + "num_tokens": 824690667.0, + "step": 21248 + }, + { + "epoch": 2.7030912097697493, + "grad_norm": 2.791991949081421, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9034090042114258, + "num_tokens": 824729390.0, + "step": 21249 + }, + { + "epoch": 2.70321842004834, + "grad_norm": 2.9530084133148193, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9067920446395874, + "num_tokens": 824768540.0, + "step": 21250 + }, + { + "epoch": 2.7033456303269303, + "grad_norm": 2.544408082962036, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9104241132736206, + "num_tokens": 824812763.0, + "step": 21251 + }, + { + "epoch": 2.703472840605521, + "grad_norm": 2.81044864654541, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8890863060951233, + "num_tokens": 824858533.0, + "step": 21252 + }, + { + "epoch": 2.7036000508841114, + "grad_norm": 2.8612492084503174, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8991336822509766, + "num_tokens": 824900987.0, + "step": 21253 + }, + { + "epoch": 2.703727261162702, + "grad_norm": 2.9928064346313477, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9089082479476929, + "num_tokens": 824938097.0, + "step": 21254 + }, + { + "epoch": 2.7038544714412924, + "grad_norm": 2.865351915359497, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8992233872413635, + "num_tokens": 824974537.0, + "step": 21255 + }, + { + "epoch": 2.703981681719883, + "grad_norm": 2.7299294471740723, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9033820033073425, + "num_tokens": 825014237.0, + "step": 21256 + }, + { + "epoch": 2.7041088919984735, + "grad_norm": 3.1037232875823975, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.903742790222168, + "num_tokens": 825047700.0, + "step": 21257 + }, + { + "epoch": 2.704236102277064, + "grad_norm": 2.5153064727783203, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9119425415992737, + "num_tokens": 825089055.0, + "step": 21258 + }, + { + "epoch": 2.7043633125556545, + "grad_norm": 2.843435287475586, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8930415511131287, + "num_tokens": 825131714.0, + "step": 21259 + }, + { + "epoch": 2.704490522834245, + "grad_norm": 2.81462025642395, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9049727320671082, + "num_tokens": 825172814.0, + "step": 21260 + }, + { + "epoch": 2.7046177331128356, + "grad_norm": 2.9570584297180176, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9054179787635803, + "num_tokens": 825205692.0, + "step": 21261 + }, + { + "epoch": 2.704744943391426, + "grad_norm": 2.6439602375030518, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.913388729095459, + "num_tokens": 825242082.0, + "step": 21262 + }, + { + "epoch": 2.7048721536700167, + "grad_norm": 2.9804418087005615, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9090148210525513, + "num_tokens": 825275731.0, + "step": 21263 + }, + { + "epoch": 2.704999363948607, + "grad_norm": 2.8718528747558594, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9017612338066101, + "num_tokens": 825316739.0, + "step": 21264 + }, + { + "epoch": 2.7051265742271977, + "grad_norm": 2.8902547359466553, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9071471691131592, + "num_tokens": 825354190.0, + "step": 21265 + }, + { + "epoch": 2.7052537845057882, + "grad_norm": 2.553959369659424, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9089429378509521, + "num_tokens": 825397935.0, + "step": 21266 + }, + { + "epoch": 2.7053809947843783, + "grad_norm": 3.0489425659179688, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9015036821365356, + "num_tokens": 825433601.0, + "step": 21267 + }, + { + "epoch": 2.7055082050629693, + "grad_norm": 2.8994691371917725, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9063321948051453, + "num_tokens": 825470626.0, + "step": 21268 + }, + { + "epoch": 2.7056354153415594, + "grad_norm": 2.8277366161346436, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.9081593751907349, + "num_tokens": 825508585.0, + "step": 21269 + }, + { + "epoch": 2.7057626256201504, + "grad_norm": 3.004183530807495, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9129432439804077, + "num_tokens": 825549511.0, + "step": 21270 + }, + { + "epoch": 2.7058898358987404, + "grad_norm": 3.119152069091797, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9009004235267639, + "num_tokens": 825584896.0, + "step": 21271 + }, + { + "epoch": 2.706017046177331, + "grad_norm": 3.2235748767852783, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9103763103485107, + "num_tokens": 825616440.0, + "step": 21272 + }, + { + "epoch": 2.7061442564559215, + "grad_norm": 2.8293392658233643, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9073381423950195, + "num_tokens": 825652987.0, + "step": 21273 + }, + { + "epoch": 2.706271466734512, + "grad_norm": 3.102534532546997, + "learning_rate": 1e-06, + "loss": 0.2911, + "mean_token_accuracy": 0.8948715925216675, + "num_tokens": 825692895.0, + "step": 21274 + }, + { + "epoch": 2.7063986770131025, + "grad_norm": 2.901218891143799, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9096953272819519, + "num_tokens": 825726403.0, + "step": 21275 + }, + { + "epoch": 2.706525887291693, + "grad_norm": 2.9018757343292236, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9027146100997925, + "num_tokens": 825765419.0, + "step": 21276 + }, + { + "epoch": 2.7066530975702836, + "grad_norm": 2.8919320106506348, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.8951430320739746, + "num_tokens": 825806626.0, + "step": 21277 + }, + { + "epoch": 2.706780307848874, + "grad_norm": 2.780758857727051, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8924171328544617, + "num_tokens": 825852851.0, + "step": 21278 + }, + { + "epoch": 2.7069075181274647, + "grad_norm": 2.999675989151001, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9051088094711304, + "num_tokens": 825890235.0, + "step": 21279 + }, + { + "epoch": 2.707034728406055, + "grad_norm": 2.7394838333129883, + "learning_rate": 1e-06, + "loss": 0.2487, + "mean_token_accuracy": 0.9084634184837341, + "num_tokens": 825932713.0, + "step": 21280 + }, + { + "epoch": 2.7071619386846457, + "grad_norm": 2.639540910720825, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9043689966201782, + "num_tokens": 825976658.0, + "step": 21281 + }, + { + "epoch": 2.7072891489632362, + "grad_norm": 2.705329656600952, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9004610776901245, + "num_tokens": 826024361.0, + "step": 21282 + }, + { + "epoch": 2.7074163592418268, + "grad_norm": 2.845867395401001, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.9004601240158081, + "num_tokens": 826063947.0, + "step": 21283 + }, + { + "epoch": 2.7075435695204173, + "grad_norm": 3.0273396968841553, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8937243819236755, + "num_tokens": 826103875.0, + "step": 21284 + }, + { + "epoch": 2.707670779799008, + "grad_norm": 2.8966052532196045, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9130299687385559, + "num_tokens": 826138421.0, + "step": 21285 + }, + { + "epoch": 2.7077979900775984, + "grad_norm": 2.8576183319091797, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9099943041801453, + "num_tokens": 826177990.0, + "step": 21286 + }, + { + "epoch": 2.707925200356189, + "grad_norm": 2.717613697052002, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9080625772476196, + "num_tokens": 826218608.0, + "step": 21287 + }, + { + "epoch": 2.7080524106347794, + "grad_norm": 2.8043386936187744, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8946837186813354, + "num_tokens": 826263107.0, + "step": 21288 + }, + { + "epoch": 2.70817962091337, + "grad_norm": 3.445798635482788, + "learning_rate": 1e-06, + "loss": 0.3107, + "mean_token_accuracy": 0.8888357877731323, + "num_tokens": 826298782.0, + "step": 21289 + }, + { + "epoch": 2.7083068311919605, + "grad_norm": 2.570845603942871, + "learning_rate": 1e-06, + "loss": 0.2161, + "mean_token_accuracy": 0.9180444478988647, + "num_tokens": 826340855.0, + "step": 21290 + }, + { + "epoch": 2.708434041470551, + "grad_norm": 2.88531494140625, + "learning_rate": 1e-06, + "loss": 0.3035, + "mean_token_accuracy": 0.8917932510375977, + "num_tokens": 826381868.0, + "step": 21291 + }, + { + "epoch": 2.708561251749141, + "grad_norm": 2.564831018447876, + "learning_rate": 1e-06, + "loss": 0.2325, + "mean_token_accuracy": 0.9144554734230042, + "num_tokens": 826424753.0, + "step": 21292 + }, + { + "epoch": 2.708688462027732, + "grad_norm": 2.9597792625427246, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9090050458908081, + "num_tokens": 826457089.0, + "step": 21293 + }, + { + "epoch": 2.708815672306322, + "grad_norm": 2.764660596847534, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9002530574798584, + "num_tokens": 826497264.0, + "step": 21294 + }, + { + "epoch": 2.708942882584913, + "grad_norm": 3.0100278854370117, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9049980640411377, + "num_tokens": 826534887.0, + "step": 21295 + }, + { + "epoch": 2.709070092863503, + "grad_norm": 2.823859930038452, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9003718495368958, + "num_tokens": 826573430.0, + "step": 21296 + }, + { + "epoch": 2.7091973031420937, + "grad_norm": 2.7847557067871094, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9132851958274841, + "num_tokens": 826614240.0, + "step": 21297 + }, + { + "epoch": 2.7093245134206843, + "grad_norm": 2.8069305419921875, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9073987007141113, + "num_tokens": 826652531.0, + "step": 21298 + }, + { + "epoch": 2.709451723699275, + "grad_norm": 2.980350971221924, + "learning_rate": 1e-06, + "loss": 0.2244, + "mean_token_accuracy": 0.9158024787902832, + "num_tokens": 826685114.0, + "step": 21299 + }, + { + "epoch": 2.7095789339778653, + "grad_norm": 2.982421636581421, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9112086296081543, + "num_tokens": 826720705.0, + "step": 21300 + }, + { + "epoch": 2.709706144256456, + "grad_norm": 3.0603535175323486, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8937557935714722, + "num_tokens": 826759880.0, + "step": 21301 + }, + { + "epoch": 2.7098333545350464, + "grad_norm": 2.5407652854919434, + "learning_rate": 1e-06, + "loss": 0.2148, + "mean_token_accuracy": 0.9193471670150757, + "num_tokens": 826801625.0, + "step": 21302 + }, + { + "epoch": 2.709960564813637, + "grad_norm": 3.203366994857788, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8911144137382507, + "num_tokens": 826841302.0, + "step": 21303 + }, + { + "epoch": 2.7100877750922274, + "grad_norm": 2.6731903553009033, + "learning_rate": 1e-06, + "loss": 0.2295, + "mean_token_accuracy": 0.9159966707229614, + "num_tokens": 826883068.0, + "step": 21304 + }, + { + "epoch": 2.710214985370818, + "grad_norm": 3.0879786014556885, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9049704074859619, + "num_tokens": 826919843.0, + "step": 21305 + }, + { + "epoch": 2.7103421956494085, + "grad_norm": 3.1566317081451416, + "learning_rate": 1e-06, + "loss": 0.2411, + "mean_token_accuracy": 0.909395694732666, + "num_tokens": 826953025.0, + "step": 21306 + }, + { + "epoch": 2.710469405927999, + "grad_norm": 2.8771209716796875, + "learning_rate": 1e-06, + "loss": 0.305, + "mean_token_accuracy": 0.8901217579841614, + "num_tokens": 826997861.0, + "step": 21307 + }, + { + "epoch": 2.7105966162065895, + "grad_norm": 3.2457613945007324, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9002200961112976, + "num_tokens": 827028995.0, + "step": 21308 + }, + { + "epoch": 2.71072382648518, + "grad_norm": 2.866936206817627, + "learning_rate": 1e-06, + "loss": 0.3019, + "mean_token_accuracy": 0.8882012963294983, + "num_tokens": 827075145.0, + "step": 21309 + }, + { + "epoch": 2.7108510367637706, + "grad_norm": 2.9225192070007324, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9147899150848389, + "num_tokens": 827107973.0, + "step": 21310 + }, + { + "epoch": 2.710978247042361, + "grad_norm": 2.75929856300354, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9002523422241211, + "num_tokens": 827152212.0, + "step": 21311 + }, + { + "epoch": 2.7111054573209517, + "grad_norm": 3.0065219402313232, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9092898368835449, + "num_tokens": 827189816.0, + "step": 21312 + }, + { + "epoch": 2.711232667599542, + "grad_norm": 2.7536494731903076, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9114639759063721, + "num_tokens": 827230881.0, + "step": 21313 + }, + { + "epoch": 2.7113598778781327, + "grad_norm": 2.8018689155578613, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8971654176712036, + "num_tokens": 827273654.0, + "step": 21314 + }, + { + "epoch": 2.711487088156723, + "grad_norm": 2.9984891414642334, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9069013595581055, + "num_tokens": 827308923.0, + "step": 21315 + }, + { + "epoch": 2.7116142984353138, + "grad_norm": 2.961005687713623, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9031543731689453, + "num_tokens": 827347063.0, + "step": 21316 + }, + { + "epoch": 2.711741508713904, + "grad_norm": 2.7808313369750977, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9095546007156372, + "num_tokens": 827384973.0, + "step": 21317 + }, + { + "epoch": 2.711868718992495, + "grad_norm": 2.726517915725708, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9008597135543823, + "num_tokens": 827428355.0, + "step": 21318 + }, + { + "epoch": 2.711995929271085, + "grad_norm": 3.160043954849243, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9012656807899475, + "num_tokens": 827460538.0, + "step": 21319 + }, + { + "epoch": 2.712123139549676, + "grad_norm": 2.800377368927002, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.8981986045837402, + "num_tokens": 827499258.0, + "step": 21320 + }, + { + "epoch": 2.712250349828266, + "grad_norm": 2.7428553104400635, + "learning_rate": 1e-06, + "loss": 0.2229, + "mean_token_accuracy": 0.9162976741790771, + "num_tokens": 827535016.0, + "step": 21321 + }, + { + "epoch": 2.7123775601068565, + "grad_norm": 3.056910276412964, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8945815563201904, + "num_tokens": 827571783.0, + "step": 21322 + }, + { + "epoch": 2.712504770385447, + "grad_norm": 2.5947365760803223, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9160248637199402, + "num_tokens": 827617117.0, + "step": 21323 + }, + { + "epoch": 2.7126319806640375, + "grad_norm": 2.8167881965637207, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9014236927032471, + "num_tokens": 827659817.0, + "step": 21324 + }, + { + "epoch": 2.712759190942628, + "grad_norm": 2.7278940677642822, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9012983441352844, + "num_tokens": 827707283.0, + "step": 21325 + }, + { + "epoch": 2.7128864012212186, + "grad_norm": 2.977538824081421, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9039521813392639, + "num_tokens": 827738976.0, + "step": 21326 + }, + { + "epoch": 2.713013611499809, + "grad_norm": 2.7717103958129883, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9085482358932495, + "num_tokens": 827782386.0, + "step": 21327 + }, + { + "epoch": 2.7131408217783997, + "grad_norm": 2.825331211090088, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.8954195976257324, + "num_tokens": 827826568.0, + "step": 21328 + }, + { + "epoch": 2.71326803205699, + "grad_norm": 2.8826303482055664, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.9135411977767944, + "num_tokens": 827862236.0, + "step": 21329 + }, + { + "epoch": 2.7133952423355807, + "grad_norm": 2.7045347690582275, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8971445560455322, + "num_tokens": 827908086.0, + "step": 21330 + }, + { + "epoch": 2.7135224526141712, + "grad_norm": 2.7211174964904785, + "learning_rate": 1e-06, + "loss": 0.2165, + "mean_token_accuracy": 0.9212796092033386, + "num_tokens": 827945245.0, + "step": 21331 + }, + { + "epoch": 2.7136496628927618, + "grad_norm": 3.1693220138549805, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8999919891357422, + "num_tokens": 827980989.0, + "step": 21332 + }, + { + "epoch": 2.7137768731713523, + "grad_norm": 3.0019564628601074, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9073632955551147, + "num_tokens": 828015510.0, + "step": 21333 + }, + { + "epoch": 2.713904083449943, + "grad_norm": 3.0098137855529785, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9026254415512085, + "num_tokens": 828053153.0, + "step": 21334 + }, + { + "epoch": 2.7140312937285334, + "grad_norm": 2.80625057220459, + "learning_rate": 1e-06, + "loss": 0.2339, + "mean_token_accuracy": 0.9144452810287476, + "num_tokens": 828090184.0, + "step": 21335 + }, + { + "epoch": 2.714158504007124, + "grad_norm": 3.023608446121216, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8964762687683105, + "num_tokens": 828131157.0, + "step": 21336 + }, + { + "epoch": 2.7142857142857144, + "grad_norm": 2.7584376335144043, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.9136370420455933, + "num_tokens": 828171254.0, + "step": 21337 + }, + { + "epoch": 2.714412924564305, + "grad_norm": 2.99082350730896, + "learning_rate": 1e-06, + "loss": 0.2828, + "mean_token_accuracy": 0.8955473899841309, + "num_tokens": 828210508.0, + "step": 21338 + }, + { + "epoch": 2.7145401348428955, + "grad_norm": 2.9702165126800537, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9045220017433167, + "num_tokens": 828249956.0, + "step": 21339 + }, + { + "epoch": 2.7146673451214856, + "grad_norm": 2.794802665710449, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9085643291473389, + "num_tokens": 828289115.0, + "step": 21340 + }, + { + "epoch": 2.7147945554000765, + "grad_norm": 2.863661050796509, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.902657151222229, + "num_tokens": 828328143.0, + "step": 21341 + }, + { + "epoch": 2.7149217656786666, + "grad_norm": 2.7845451831817627, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9034358859062195, + "num_tokens": 828368410.0, + "step": 21342 + }, + { + "epoch": 2.7150489759572576, + "grad_norm": 2.7717642784118652, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9065530300140381, + "num_tokens": 828407291.0, + "step": 21343 + }, + { + "epoch": 2.7151761862358477, + "grad_norm": 2.907325029373169, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8959962129592896, + "num_tokens": 828443734.0, + "step": 21344 + }, + { + "epoch": 2.7153033965144386, + "grad_norm": 2.8167834281921387, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.911145806312561, + "num_tokens": 828481432.0, + "step": 21345 + }, + { + "epoch": 2.7154306067930287, + "grad_norm": 2.8836793899536133, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9015249013900757, + "num_tokens": 828519961.0, + "step": 21346 + }, + { + "epoch": 2.7155578170716193, + "grad_norm": 3.1294407844543457, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9079691171646118, + "num_tokens": 828557283.0, + "step": 21347 + }, + { + "epoch": 2.71568502735021, + "grad_norm": 3.228825330734253, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8919893503189087, + "num_tokens": 828596884.0, + "step": 21348 + }, + { + "epoch": 2.7158122376288003, + "grad_norm": 2.8841781616210938, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9048357605934143, + "num_tokens": 828638842.0, + "step": 21349 + }, + { + "epoch": 2.715939447907391, + "grad_norm": 2.795468807220459, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9093756079673767, + "num_tokens": 828683378.0, + "step": 21350 + }, + { + "epoch": 2.7160666581859814, + "grad_norm": 2.6540892124176025, + "learning_rate": 1e-06, + "loss": 0.2215, + "mean_token_accuracy": 0.9165236949920654, + "num_tokens": 828723963.0, + "step": 21351 + }, + { + "epoch": 2.716193868464572, + "grad_norm": 2.880342721939087, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.907856285572052, + "num_tokens": 828759104.0, + "step": 21352 + }, + { + "epoch": 2.7163210787431624, + "grad_norm": 2.865217685699463, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9019923210144043, + "num_tokens": 828799986.0, + "step": 21353 + }, + { + "epoch": 2.716448289021753, + "grad_norm": 3.185328722000122, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8967435956001282, + "num_tokens": 828839199.0, + "step": 21354 + }, + { + "epoch": 2.7165754993003435, + "grad_norm": 3.0953173637390137, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.903599739074707, + "num_tokens": 828875266.0, + "step": 21355 + }, + { + "epoch": 2.716702709578934, + "grad_norm": 3.216794729232788, + "learning_rate": 1e-06, + "loss": 0.313, + "mean_token_accuracy": 0.8865156769752502, + "num_tokens": 828915098.0, + "step": 21356 + }, + { + "epoch": 2.7168299198575245, + "grad_norm": 3.200390100479126, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8959076404571533, + "num_tokens": 828950344.0, + "step": 21357 + }, + { + "epoch": 2.716957130136115, + "grad_norm": 2.7491016387939453, + "learning_rate": 1e-06, + "loss": 0.221, + "mean_token_accuracy": 0.9153738021850586, + "num_tokens": 828985650.0, + "step": 21358 + }, + { + "epoch": 2.7170843404147056, + "grad_norm": 3.1115517616271973, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.914246141910553, + "num_tokens": 829023932.0, + "step": 21359 + }, + { + "epoch": 2.717211550693296, + "grad_norm": 2.701425552368164, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9075841903686523, + "num_tokens": 829066200.0, + "step": 21360 + }, + { + "epoch": 2.7173387609718866, + "grad_norm": 2.6723108291625977, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.901110053062439, + "num_tokens": 829109604.0, + "step": 21361 + }, + { + "epoch": 2.717465971250477, + "grad_norm": 2.8823533058166504, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9114721417427063, + "num_tokens": 829147312.0, + "step": 21362 + }, + { + "epoch": 2.7175931815290677, + "grad_norm": 2.7800333499908447, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.894756555557251, + "num_tokens": 829192044.0, + "step": 21363 + }, + { + "epoch": 2.7177203918076582, + "grad_norm": 2.7834417819976807, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.9030052423477173, + "num_tokens": 829234927.0, + "step": 21364 + }, + { + "epoch": 2.7178476020862483, + "grad_norm": 2.9223718643188477, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.8947976231575012, + "num_tokens": 829276686.0, + "step": 21365 + }, + { + "epoch": 2.7179748123648393, + "grad_norm": 3.032447099685669, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.9072767496109009, + "num_tokens": 829312196.0, + "step": 21366 + }, + { + "epoch": 2.7181020226434294, + "grad_norm": 3.0509233474731445, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9043468832969666, + "num_tokens": 829345317.0, + "step": 21367 + }, + { + "epoch": 2.7182292329220203, + "grad_norm": 3.1397273540496826, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9056762456893921, + "num_tokens": 829381885.0, + "step": 21368 + }, + { + "epoch": 2.7183564432006104, + "grad_norm": 2.9686026573181152, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9112741947174072, + "num_tokens": 829416101.0, + "step": 21369 + }, + { + "epoch": 2.718483653479201, + "grad_norm": 2.7939727306365967, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9035476446151733, + "num_tokens": 829458976.0, + "step": 21370 + }, + { + "epoch": 2.7186108637577915, + "grad_norm": 2.913628339767456, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8965308666229248, + "num_tokens": 829498157.0, + "step": 21371 + }, + { + "epoch": 2.718738074036382, + "grad_norm": 2.794788360595703, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9091308116912842, + "num_tokens": 829537321.0, + "step": 21372 + }, + { + "epoch": 2.7188652843149725, + "grad_norm": 2.8765709400177, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9106862545013428, + "num_tokens": 829571998.0, + "step": 21373 + }, + { + "epoch": 2.718992494593563, + "grad_norm": 3.145453929901123, + "learning_rate": 1e-06, + "loss": 0.2914, + "mean_token_accuracy": 0.8952440619468689, + "num_tokens": 829609100.0, + "step": 21374 + }, + { + "epoch": 2.7191197048721536, + "grad_norm": 3.075359582901001, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9052685499191284, + "num_tokens": 829645218.0, + "step": 21375 + }, + { + "epoch": 2.719246915150744, + "grad_norm": 3.0842394828796387, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9108408689498901, + "num_tokens": 829682441.0, + "step": 21376 + }, + { + "epoch": 2.7193741254293347, + "grad_norm": 2.898953914642334, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9056726694107056, + "num_tokens": 829718775.0, + "step": 21377 + }, + { + "epoch": 2.719501335707925, + "grad_norm": 2.9252307415008545, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9048646688461304, + "num_tokens": 829759676.0, + "step": 21378 + }, + { + "epoch": 2.7196285459865157, + "grad_norm": 2.608290672302246, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9047120213508606, + "num_tokens": 829803311.0, + "step": 21379 + }, + { + "epoch": 2.7197557562651062, + "grad_norm": 3.162017583847046, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8942914009094238, + "num_tokens": 829840457.0, + "step": 21380 + }, + { + "epoch": 2.7198829665436968, + "grad_norm": 2.896688222885132, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.908412754535675, + "num_tokens": 829877268.0, + "step": 21381 + }, + { + "epoch": 2.7200101768222873, + "grad_norm": 2.815906524658203, + "learning_rate": 1e-06, + "loss": 0.2905, + "mean_token_accuracy": 0.8947151899337769, + "num_tokens": 829921619.0, + "step": 21382 + }, + { + "epoch": 2.720137387100878, + "grad_norm": 3.234487295150757, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8915866613388062, + "num_tokens": 829954595.0, + "step": 21383 + }, + { + "epoch": 2.7202645973794684, + "grad_norm": 2.821326971054077, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9120602011680603, + "num_tokens": 829989826.0, + "step": 21384 + }, + { + "epoch": 2.720391807658059, + "grad_norm": 2.791656970977783, + "learning_rate": 1e-06, + "loss": 0.2253, + "mean_token_accuracy": 0.9164907336235046, + "num_tokens": 830027490.0, + "step": 21385 + }, + { + "epoch": 2.7205190179366494, + "grad_norm": 3.020317316055298, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8962392807006836, + "num_tokens": 830066008.0, + "step": 21386 + }, + { + "epoch": 2.72064622821524, + "grad_norm": 3.01806640625, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9077856540679932, + "num_tokens": 830101365.0, + "step": 21387 + }, + { + "epoch": 2.7207734384938305, + "grad_norm": 2.5135865211486816, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9120351672172546, + "num_tokens": 830146095.0, + "step": 21388 + }, + { + "epoch": 2.720900648772421, + "grad_norm": 2.5798568725585938, + "learning_rate": 1e-06, + "loss": 0.2169, + "mean_token_accuracy": 0.918060302734375, + "num_tokens": 830188048.0, + "step": 21389 + }, + { + "epoch": 2.721027859051011, + "grad_norm": 2.962620973587036, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9065321087837219, + "num_tokens": 830226235.0, + "step": 21390 + }, + { + "epoch": 2.721155069329602, + "grad_norm": 2.958413600921631, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9055367112159729, + "num_tokens": 830263699.0, + "step": 21391 + }, + { + "epoch": 2.721282279608192, + "grad_norm": 2.9879283905029297, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8917380571365356, + "num_tokens": 830301943.0, + "step": 21392 + }, + { + "epoch": 2.721409489886783, + "grad_norm": 2.8430371284484863, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9038732051849365, + "num_tokens": 830343619.0, + "step": 21393 + }, + { + "epoch": 2.721536700165373, + "grad_norm": 3.0902199745178223, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8976484537124634, + "num_tokens": 830381910.0, + "step": 21394 + }, + { + "epoch": 2.7216639104439637, + "grad_norm": 3.039247989654541, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9040272235870361, + "num_tokens": 830415470.0, + "step": 21395 + }, + { + "epoch": 2.7217911207225542, + "grad_norm": 3.0463297367095947, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9012045860290527, + "num_tokens": 830453377.0, + "step": 21396 + }, + { + "epoch": 2.7219183310011448, + "grad_norm": 2.795611619949341, + "learning_rate": 1e-06, + "loss": 0.2405, + "mean_token_accuracy": 0.9116135239601135, + "num_tokens": 830493615.0, + "step": 21397 + }, + { + "epoch": 2.7220455412797353, + "grad_norm": 2.6997742652893066, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9029823541641235, + "num_tokens": 830539217.0, + "step": 21398 + }, + { + "epoch": 2.722172751558326, + "grad_norm": 2.8322694301605225, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9012668132781982, + "num_tokens": 830579267.0, + "step": 21399 + }, + { + "epoch": 2.7222999618369164, + "grad_norm": 2.9469406604766846, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8987705707550049, + "num_tokens": 830618892.0, + "step": 21400 + }, + { + "epoch": 2.722427172115507, + "grad_norm": 2.8028650283813477, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9066348671913147, + "num_tokens": 830659393.0, + "step": 21401 + }, + { + "epoch": 2.7225543823940974, + "grad_norm": 2.698446273803711, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9062873125076294, + "num_tokens": 830700830.0, + "step": 21402 + }, + { + "epoch": 2.722681592672688, + "grad_norm": 2.708174705505371, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9080402851104736, + "num_tokens": 830742098.0, + "step": 21403 + }, + { + "epoch": 2.7228088029512785, + "grad_norm": 2.895624876022339, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.8968441486358643, + "num_tokens": 830779623.0, + "step": 21404 + }, + { + "epoch": 2.722936013229869, + "grad_norm": 2.9008445739746094, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9041268825531006, + "num_tokens": 830820955.0, + "step": 21405 + }, + { + "epoch": 2.7230632235084595, + "grad_norm": 2.7453410625457764, + "learning_rate": 1e-06, + "loss": 0.2851, + "mean_token_accuracy": 0.8963056206703186, + "num_tokens": 830869350.0, + "step": 21406 + }, + { + "epoch": 2.72319043378705, + "grad_norm": 2.979433298110962, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.895142674446106, + "num_tokens": 830910198.0, + "step": 21407 + }, + { + "epoch": 2.7233176440656406, + "grad_norm": 3.1358675956726074, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9017183780670166, + "num_tokens": 830943575.0, + "step": 21408 + }, + { + "epoch": 2.723444854344231, + "grad_norm": 3.035475492477417, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.9000023603439331, + "num_tokens": 830981056.0, + "step": 21409 + }, + { + "epoch": 2.7235720646228216, + "grad_norm": 2.889124870300293, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9071484804153442, + "num_tokens": 831015261.0, + "step": 21410 + }, + { + "epoch": 2.723699274901412, + "grad_norm": 2.915332555770874, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9079694151878357, + "num_tokens": 831055427.0, + "step": 21411 + }, + { + "epoch": 2.7238264851800027, + "grad_norm": 3.1095190048217773, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8979952335357666, + "num_tokens": 831091172.0, + "step": 21412 + }, + { + "epoch": 2.723953695458593, + "grad_norm": 2.760951042175293, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9013967514038086, + "num_tokens": 831133776.0, + "step": 21413 + }, + { + "epoch": 2.7240809057371838, + "grad_norm": 2.9794044494628906, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9038147926330566, + "num_tokens": 831168744.0, + "step": 21414 + }, + { + "epoch": 2.724208116015774, + "grad_norm": 2.7143146991729736, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9044135808944702, + "num_tokens": 831212920.0, + "step": 21415 + }, + { + "epoch": 2.724335326294365, + "grad_norm": 2.614297389984131, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9075497388839722, + "num_tokens": 831254940.0, + "step": 21416 + }, + { + "epoch": 2.724462536572955, + "grad_norm": 2.960691213607788, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.9045538902282715, + "num_tokens": 831290066.0, + "step": 21417 + }, + { + "epoch": 2.724589746851546, + "grad_norm": 2.996854305267334, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9000344276428223, + "num_tokens": 831326151.0, + "step": 21418 + }, + { + "epoch": 2.724716957130136, + "grad_norm": 2.7696235179901123, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9048478603363037, + "num_tokens": 831367316.0, + "step": 21419 + }, + { + "epoch": 2.7248441674087265, + "grad_norm": 3.1309447288513184, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.898421049118042, + "num_tokens": 831404036.0, + "step": 21420 + }, + { + "epoch": 2.724971377687317, + "grad_norm": 2.895951271057129, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8980222940444946, + "num_tokens": 831444759.0, + "step": 21421 + }, + { + "epoch": 2.7250985879659075, + "grad_norm": 2.6438212394714355, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9016962647438049, + "num_tokens": 831488513.0, + "step": 21422 + }, + { + "epoch": 2.725225798244498, + "grad_norm": 2.9556519985198975, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8982176780700684, + "num_tokens": 831528720.0, + "step": 21423 + }, + { + "epoch": 2.7253530085230886, + "grad_norm": 2.827448844909668, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9095561504364014, + "num_tokens": 831565989.0, + "step": 21424 + }, + { + "epoch": 2.725480218801679, + "grad_norm": 2.9599804878234863, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9051497578620911, + "num_tokens": 831603102.0, + "step": 21425 + }, + { + "epoch": 2.7256074290802697, + "grad_norm": 2.767949104309082, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.9126850366592407, + "num_tokens": 831640910.0, + "step": 21426 + }, + { + "epoch": 2.72573463935886, + "grad_norm": 3.0247244834899902, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8955241441726685, + "num_tokens": 831678107.0, + "step": 21427 + }, + { + "epoch": 2.7258618496374507, + "grad_norm": 2.9244418144226074, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.8955869078636169, + "num_tokens": 831718480.0, + "step": 21428 + }, + { + "epoch": 2.7259890599160412, + "grad_norm": 2.96873140335083, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8995014429092407, + "num_tokens": 831760294.0, + "step": 21429 + }, + { + "epoch": 2.7261162701946318, + "grad_norm": 2.929680585861206, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8950988054275513, + "num_tokens": 831801444.0, + "step": 21430 + }, + { + "epoch": 2.7262434804732223, + "grad_norm": 2.7959699630737305, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.9125254154205322, + "num_tokens": 831839579.0, + "step": 21431 + }, + { + "epoch": 2.726370690751813, + "grad_norm": 2.6401307582855225, + "learning_rate": 1e-06, + "loss": 0.2443, + "mean_token_accuracy": 0.9071625471115112, + "num_tokens": 831882851.0, + "step": 21432 + }, + { + "epoch": 2.7264979010304033, + "grad_norm": 2.80773663520813, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9021648168563843, + "num_tokens": 831922006.0, + "step": 21433 + }, + { + "epoch": 2.726625111308994, + "grad_norm": 3.0905308723449707, + "learning_rate": 1e-06, + "loss": 0.2897, + "mean_token_accuracy": 0.8935893177986145, + "num_tokens": 831961081.0, + "step": 21434 + }, + { + "epoch": 2.7267523215875844, + "grad_norm": 2.9097540378570557, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.904921293258667, + "num_tokens": 831995975.0, + "step": 21435 + }, + { + "epoch": 2.726879531866175, + "grad_norm": 2.7896482944488525, + "learning_rate": 1e-06, + "loss": 0.2339, + "mean_token_accuracy": 0.9128856658935547, + "num_tokens": 832033282.0, + "step": 21436 + }, + { + "epoch": 2.7270067421447655, + "grad_norm": 2.8417110443115234, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9009451866149902, + "num_tokens": 832074244.0, + "step": 21437 + }, + { + "epoch": 2.7271339524233555, + "grad_norm": 3.005708932876587, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9057732820510864, + "num_tokens": 832112636.0, + "step": 21438 + }, + { + "epoch": 2.7272611627019465, + "grad_norm": 2.604971170425415, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9123749732971191, + "num_tokens": 832155818.0, + "step": 21439 + }, + { + "epoch": 2.7273883729805366, + "grad_norm": 2.786625862121582, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9101890325546265, + "num_tokens": 832196502.0, + "step": 21440 + }, + { + "epoch": 2.7275155832591276, + "grad_norm": 3.0041539669036865, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9013689756393433, + "num_tokens": 832232761.0, + "step": 21441 + }, + { + "epoch": 2.7276427935377177, + "grad_norm": 2.969062328338623, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9047806262969971, + "num_tokens": 832272867.0, + "step": 21442 + }, + { + "epoch": 2.7277700038163086, + "grad_norm": 2.990924835205078, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9041934013366699, + "num_tokens": 832308271.0, + "step": 21443 + }, + { + "epoch": 2.7278972140948987, + "grad_norm": 3.0822036266326904, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8991093635559082, + "num_tokens": 832341701.0, + "step": 21444 + }, + { + "epoch": 2.7280244243734892, + "grad_norm": 2.86217999458313, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9070773124694824, + "num_tokens": 832380187.0, + "step": 21445 + }, + { + "epoch": 2.7281516346520798, + "grad_norm": 2.8365330696105957, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9020843505859375, + "num_tokens": 832418644.0, + "step": 21446 + }, + { + "epoch": 2.7282788449306703, + "grad_norm": 2.8787949085235596, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8993143439292908, + "num_tokens": 832459272.0, + "step": 21447 + }, + { + "epoch": 2.728406055209261, + "grad_norm": 3.1410226821899414, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.895311176776886, + "num_tokens": 832493667.0, + "step": 21448 + }, + { + "epoch": 2.7285332654878514, + "grad_norm": 3.1822643280029297, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8893675804138184, + "num_tokens": 832529909.0, + "step": 21449 + }, + { + "epoch": 2.728660475766442, + "grad_norm": 3.049351453781128, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9056156277656555, + "num_tokens": 832566225.0, + "step": 21450 + }, + { + "epoch": 2.7287876860450324, + "grad_norm": 2.8056256771087646, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9025915861129761, + "num_tokens": 832608032.0, + "step": 21451 + }, + { + "epoch": 2.728914896323623, + "grad_norm": 3.0698161125183105, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8939599990844727, + "num_tokens": 832645264.0, + "step": 21452 + }, + { + "epoch": 2.7290421066022135, + "grad_norm": 2.8377532958984375, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9076685309410095, + "num_tokens": 832683488.0, + "step": 21453 + }, + { + "epoch": 2.729169316880804, + "grad_norm": 3.134938955307007, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8976482152938843, + "num_tokens": 832718146.0, + "step": 21454 + }, + { + "epoch": 2.7292965271593945, + "grad_norm": 3.0177948474884033, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9088354706764221, + "num_tokens": 832756485.0, + "step": 21455 + }, + { + "epoch": 2.729423737437985, + "grad_norm": 3.260932207107544, + "learning_rate": 1e-06, + "loss": 0.2929, + "mean_token_accuracy": 0.8936859369277954, + "num_tokens": 832793850.0, + "step": 21456 + }, + { + "epoch": 2.7295509477165756, + "grad_norm": 2.8747763633728027, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9017455577850342, + "num_tokens": 832832185.0, + "step": 21457 + }, + { + "epoch": 2.729678157995166, + "grad_norm": 2.8854548931121826, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9051420092582703, + "num_tokens": 832872582.0, + "step": 21458 + }, + { + "epoch": 2.7298053682737566, + "grad_norm": 2.962045669555664, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9024940729141235, + "num_tokens": 832910936.0, + "step": 21459 + }, + { + "epoch": 2.729932578552347, + "grad_norm": 2.8769679069519043, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9074395895004272, + "num_tokens": 832950502.0, + "step": 21460 + }, + { + "epoch": 2.7300597888309377, + "grad_norm": 2.9389045238494873, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8970426321029663, + "num_tokens": 832993047.0, + "step": 21461 + }, + { + "epoch": 2.7301869991095282, + "grad_norm": 2.667264223098755, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9119933843612671, + "num_tokens": 833032562.0, + "step": 21462 + }, + { + "epoch": 2.7303142093881183, + "grad_norm": 3.056892156600952, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9093321561813354, + "num_tokens": 833067343.0, + "step": 21463 + }, + { + "epoch": 2.7304414196667093, + "grad_norm": 2.5542399883270264, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9045912027359009, + "num_tokens": 833111164.0, + "step": 21464 + }, + { + "epoch": 2.7305686299452994, + "grad_norm": 3.162287712097168, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9017361998558044, + "num_tokens": 833145307.0, + "step": 21465 + }, + { + "epoch": 2.7306958402238903, + "grad_norm": 3.099621295928955, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9032722115516663, + "num_tokens": 833180780.0, + "step": 21466 + }, + { + "epoch": 2.7308230505024804, + "grad_norm": 3.019645929336548, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9069281816482544, + "num_tokens": 833216129.0, + "step": 21467 + }, + { + "epoch": 2.730950260781071, + "grad_norm": 3.1483640670776367, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.8978748321533203, + "num_tokens": 833252813.0, + "step": 21468 + }, + { + "epoch": 2.7310774710596615, + "grad_norm": 3.0843875408172607, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8932949304580688, + "num_tokens": 833294142.0, + "step": 21469 + }, + { + "epoch": 2.731204681338252, + "grad_norm": 3.19867205619812, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9032521843910217, + "num_tokens": 833326611.0, + "step": 21470 + }, + { + "epoch": 2.7313318916168425, + "grad_norm": 3.352574348449707, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8973082900047302, + "num_tokens": 833366168.0, + "step": 21471 + }, + { + "epoch": 2.731459101895433, + "grad_norm": 3.050204277038574, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9054923057556152, + "num_tokens": 833402111.0, + "step": 21472 + }, + { + "epoch": 2.7315863121740236, + "grad_norm": 2.8667845726013184, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9054180383682251, + "num_tokens": 833439175.0, + "step": 21473 + }, + { + "epoch": 2.731713522452614, + "grad_norm": 2.9565601348876953, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9089900255203247, + "num_tokens": 833473007.0, + "step": 21474 + }, + { + "epoch": 2.7318407327312046, + "grad_norm": 2.638122081756592, + "learning_rate": 1e-06, + "loss": 0.2362, + "mean_token_accuracy": 0.9135018587112427, + "num_tokens": 833512427.0, + "step": 21475 + }, + { + "epoch": 2.731967943009795, + "grad_norm": 2.8565547466278076, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.903672993183136, + "num_tokens": 833552530.0, + "step": 21476 + }, + { + "epoch": 2.7320951532883857, + "grad_norm": 2.8030242919921875, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9086658954620361, + "num_tokens": 833590883.0, + "step": 21477 + }, + { + "epoch": 2.7322223635669762, + "grad_norm": 2.6437292098999023, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9005078077316284, + "num_tokens": 833635229.0, + "step": 21478 + }, + { + "epoch": 2.7323495738455668, + "grad_norm": 2.9457404613494873, + "learning_rate": 1e-06, + "loss": 0.2941, + "mean_token_accuracy": 0.8931899070739746, + "num_tokens": 833676448.0, + "step": 21479 + }, + { + "epoch": 2.7324767841241573, + "grad_norm": 2.7639546394348145, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9138168692588806, + "num_tokens": 833713010.0, + "step": 21480 + }, + { + "epoch": 2.732603994402748, + "grad_norm": 2.982527256011963, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9029221534729004, + "num_tokens": 833747159.0, + "step": 21481 + }, + { + "epoch": 2.7327312046813383, + "grad_norm": 2.5507266521453857, + "learning_rate": 1e-06, + "loss": 0.2185, + "mean_token_accuracy": 0.9184449911117554, + "num_tokens": 833787723.0, + "step": 21482 + }, + { + "epoch": 2.732858414959929, + "grad_norm": 2.6586856842041016, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.907692015171051, + "num_tokens": 833828722.0, + "step": 21483 + }, + { + "epoch": 2.7329856252385194, + "grad_norm": 2.6746115684509277, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9002459645271301, + "num_tokens": 833872687.0, + "step": 21484 + }, + { + "epoch": 2.73311283551711, + "grad_norm": 2.878875970840454, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.8996655344963074, + "num_tokens": 833911831.0, + "step": 21485 + }, + { + "epoch": 2.7332400457957005, + "grad_norm": 2.712805986404419, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.901518702507019, + "num_tokens": 833958291.0, + "step": 21486 + }, + { + "epoch": 2.733367256074291, + "grad_norm": 2.7357053756713867, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9144091606140137, + "num_tokens": 833994999.0, + "step": 21487 + }, + { + "epoch": 2.733494466352881, + "grad_norm": 2.719085454940796, + "learning_rate": 1e-06, + "loss": 0.226, + "mean_token_accuracy": 0.9160552024841309, + "num_tokens": 834033894.0, + "step": 21488 + }, + { + "epoch": 2.733621676631472, + "grad_norm": 2.8705830574035645, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9062975645065308, + "num_tokens": 834070344.0, + "step": 21489 + }, + { + "epoch": 2.733748886910062, + "grad_norm": 2.701571226119995, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9081049561500549, + "num_tokens": 834109050.0, + "step": 21490 + }, + { + "epoch": 2.733876097188653, + "grad_norm": 2.8552541732788086, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9067699909210205, + "num_tokens": 834148553.0, + "step": 21491 + }, + { + "epoch": 2.734003307467243, + "grad_norm": 2.6675407886505127, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9077455997467041, + "num_tokens": 834191387.0, + "step": 21492 + }, + { + "epoch": 2.7341305177458337, + "grad_norm": 2.929687261581421, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9068911075592041, + "num_tokens": 834228405.0, + "step": 21493 + }, + { + "epoch": 2.7342577280244242, + "grad_norm": 2.858421564102173, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.9100503921508789, + "num_tokens": 834264591.0, + "step": 21494 + }, + { + "epoch": 2.7343849383030148, + "grad_norm": 2.5840163230895996, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9089966416358948, + "num_tokens": 834310139.0, + "step": 21495 + }, + { + "epoch": 2.7345121485816053, + "grad_norm": 3.0657904148101807, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9090293645858765, + "num_tokens": 834344211.0, + "step": 21496 + }, + { + "epoch": 2.734639358860196, + "grad_norm": 3.028164863586426, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.8972496390342712, + "num_tokens": 834382732.0, + "step": 21497 + }, + { + "epoch": 2.7347665691387864, + "grad_norm": 2.881422281265259, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9046754837036133, + "num_tokens": 834421111.0, + "step": 21498 + }, + { + "epoch": 2.734893779417377, + "grad_norm": 2.9253029823303223, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9002101421356201, + "num_tokens": 834461307.0, + "step": 21499 + }, + { + "epoch": 2.7350209896959674, + "grad_norm": 2.6322200298309326, + "learning_rate": 1e-06, + "loss": 0.245, + "mean_token_accuracy": 0.9090070724487305, + "num_tokens": 834508605.0, + "step": 21500 + }, + { + "epoch": 2.735148199974558, + "grad_norm": 2.828172206878662, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.9096105098724365, + "num_tokens": 834549188.0, + "step": 21501 + }, + { + "epoch": 2.7352754102531485, + "grad_norm": 2.779447078704834, + "learning_rate": 1e-06, + "loss": 0.2306, + "mean_token_accuracy": 0.9148483872413635, + "num_tokens": 834586916.0, + "step": 21502 + }, + { + "epoch": 2.735402620531739, + "grad_norm": 3.0038955211639404, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.912955641746521, + "num_tokens": 834623198.0, + "step": 21503 + }, + { + "epoch": 2.7355298308103295, + "grad_norm": 2.89909029006958, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8952969908714294, + "num_tokens": 834665817.0, + "step": 21504 + }, + { + "epoch": 2.73565704108892, + "grad_norm": 2.8191986083984375, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9035784006118774, + "num_tokens": 834704032.0, + "step": 21505 + }, + { + "epoch": 2.7357842513675106, + "grad_norm": 3.0851871967315674, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9062017798423767, + "num_tokens": 834738451.0, + "step": 21506 + }, + { + "epoch": 2.735911461646101, + "grad_norm": 2.8507938385009766, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.90462726354599, + "num_tokens": 834780744.0, + "step": 21507 + }, + { + "epoch": 2.7360386719246916, + "grad_norm": 2.90409779548645, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.902317464351654, + "num_tokens": 834819791.0, + "step": 21508 + }, + { + "epoch": 2.736165882203282, + "grad_norm": 2.743380546569824, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.9140843749046326, + "num_tokens": 834860504.0, + "step": 21509 + }, + { + "epoch": 2.7362930924818727, + "grad_norm": 2.8268942832946777, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.9113283753395081, + "num_tokens": 834897522.0, + "step": 21510 + }, + { + "epoch": 2.7364203027604628, + "grad_norm": 2.6053831577301025, + "learning_rate": 1e-06, + "loss": 0.2318, + "mean_token_accuracy": 0.9119448661804199, + "num_tokens": 834941877.0, + "step": 21511 + }, + { + "epoch": 2.7365475130390537, + "grad_norm": 3.135298252105713, + "learning_rate": 1e-06, + "loss": 0.32, + "mean_token_accuracy": 0.8868470788002014, + "num_tokens": 834984995.0, + "step": 21512 + }, + { + "epoch": 2.736674723317644, + "grad_norm": 3.143906354904175, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9045243263244629, + "num_tokens": 835016721.0, + "step": 21513 + }, + { + "epoch": 2.736801933596235, + "grad_norm": 3.1076176166534424, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8962692022323608, + "num_tokens": 835052548.0, + "step": 21514 + }, + { + "epoch": 2.736929143874825, + "grad_norm": 2.8329250812530518, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9101153016090393, + "num_tokens": 835092832.0, + "step": 21515 + }, + { + "epoch": 2.737056354153416, + "grad_norm": 2.988860607147217, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.907903790473938, + "num_tokens": 835128086.0, + "step": 21516 + }, + { + "epoch": 2.737183564432006, + "grad_norm": 2.82814359664917, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9028132557868958, + "num_tokens": 835169109.0, + "step": 21517 + }, + { + "epoch": 2.7373107747105965, + "grad_norm": 2.90264892578125, + "learning_rate": 1e-06, + "loss": 0.27, + "mean_token_accuracy": 0.898777961730957, + "num_tokens": 835211040.0, + "step": 21518 + }, + { + "epoch": 2.737437984989187, + "grad_norm": 3.2656421661376953, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8998943567276001, + "num_tokens": 835245169.0, + "step": 21519 + }, + { + "epoch": 2.7375651952677775, + "grad_norm": 2.9636478424072266, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9094889163970947, + "num_tokens": 835280535.0, + "step": 21520 + }, + { + "epoch": 2.737692405546368, + "grad_norm": 3.2095272541046143, + "learning_rate": 1e-06, + "loss": 0.2989, + "mean_token_accuracy": 0.8903375267982483, + "num_tokens": 835315954.0, + "step": 21521 + }, + { + "epoch": 2.7378196158249586, + "grad_norm": 2.8061792850494385, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.8999344110488892, + "num_tokens": 835355568.0, + "step": 21522 + }, + { + "epoch": 2.737946826103549, + "grad_norm": 3.2590177059173584, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8932545185089111, + "num_tokens": 835389255.0, + "step": 21523 + }, + { + "epoch": 2.7380740363821396, + "grad_norm": 3.0217928886413574, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8998509049415588, + "num_tokens": 835423206.0, + "step": 21524 + }, + { + "epoch": 2.73820124666073, + "grad_norm": 2.856410264968872, + "learning_rate": 1e-06, + "loss": 0.2252, + "mean_token_accuracy": 0.914747953414917, + "num_tokens": 835457008.0, + "step": 21525 + }, + { + "epoch": 2.7383284569393207, + "grad_norm": 2.8307323455810547, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.899031937122345, + "num_tokens": 835499989.0, + "step": 21526 + }, + { + "epoch": 2.7384556672179112, + "grad_norm": 2.845067262649536, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9110461473464966, + "num_tokens": 835538007.0, + "step": 21527 + }, + { + "epoch": 2.7385828774965018, + "grad_norm": 2.6489787101745605, + "learning_rate": 1e-06, + "loss": 0.29, + "mean_token_accuracy": 0.8937217593193054, + "num_tokens": 835585834.0, + "step": 21528 + }, + { + "epoch": 2.7387100877750923, + "grad_norm": 2.7629761695861816, + "learning_rate": 1e-06, + "loss": 0.2718, + "mean_token_accuracy": 0.9013662934303284, + "num_tokens": 835630451.0, + "step": 21529 + }, + { + "epoch": 2.738837298053683, + "grad_norm": 2.8457229137420654, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9060748815536499, + "num_tokens": 835667556.0, + "step": 21530 + }, + { + "epoch": 2.7389645083322733, + "grad_norm": 2.9645795822143555, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9010840058326721, + "num_tokens": 835702459.0, + "step": 21531 + }, + { + "epoch": 2.739091718610864, + "grad_norm": 2.871387004852295, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.8994903564453125, + "num_tokens": 835744231.0, + "step": 21532 + }, + { + "epoch": 2.7392189288894544, + "grad_norm": 2.7503409385681152, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.901382565498352, + "num_tokens": 835789006.0, + "step": 21533 + }, + { + "epoch": 2.739346139168045, + "grad_norm": 3.0120203495025635, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.8995038866996765, + "num_tokens": 835825775.0, + "step": 21534 + }, + { + "epoch": 2.7394733494466355, + "grad_norm": 2.686821222305298, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.8986388444900513, + "num_tokens": 835870525.0, + "step": 21535 + }, + { + "epoch": 2.7396005597252255, + "grad_norm": 3.1404478549957275, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.8979724049568176, + "num_tokens": 835902923.0, + "step": 21536 + }, + { + "epoch": 2.7397277700038165, + "grad_norm": 2.6862950325012207, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9022570848464966, + "num_tokens": 835949559.0, + "step": 21537 + }, + { + "epoch": 2.7398549802824066, + "grad_norm": 2.942746162414551, + "learning_rate": 1e-06, + "loss": 0.3016, + "mean_token_accuracy": 0.8878699541091919, + "num_tokens": 835990179.0, + "step": 21538 + }, + { + "epoch": 2.7399821905609976, + "grad_norm": 3.0390353202819824, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.9003292322158813, + "num_tokens": 836029280.0, + "step": 21539 + }, + { + "epoch": 2.7401094008395877, + "grad_norm": 2.9768431186676025, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9053325653076172, + "num_tokens": 836063050.0, + "step": 21540 + }, + { + "epoch": 2.7402366111181786, + "grad_norm": 2.5733001232147217, + "learning_rate": 1e-06, + "loss": 0.2202, + "mean_token_accuracy": 0.9156595468521118, + "num_tokens": 836104416.0, + "step": 21541 + }, + { + "epoch": 2.7403638213967687, + "grad_norm": 2.9926178455352783, + "learning_rate": 1e-06, + "loss": 0.264, + "mean_token_accuracy": 0.9040166735649109, + "num_tokens": 836143351.0, + "step": 21542 + }, + { + "epoch": 2.7404910316753592, + "grad_norm": 2.8989009857177734, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9043583869934082, + "num_tokens": 836182683.0, + "step": 21543 + }, + { + "epoch": 2.7406182419539498, + "grad_norm": 3.1195223331451416, + "learning_rate": 1e-06, + "loss": 0.2988, + "mean_token_accuracy": 0.8919392824172974, + "num_tokens": 836223672.0, + "step": 21544 + }, + { + "epoch": 2.7407454522325403, + "grad_norm": 2.9244070053100586, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9008791446685791, + "num_tokens": 836262639.0, + "step": 21545 + }, + { + "epoch": 2.740872662511131, + "grad_norm": 2.8819236755371094, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9011747241020203, + "num_tokens": 836301126.0, + "step": 21546 + }, + { + "epoch": 2.7409998727897213, + "grad_norm": 2.5901124477386475, + "learning_rate": 1e-06, + "loss": 0.2318, + "mean_token_accuracy": 0.9136319756507874, + "num_tokens": 836344069.0, + "step": 21547 + }, + { + "epoch": 2.741127083068312, + "grad_norm": 2.8169641494750977, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9027183055877686, + "num_tokens": 836386222.0, + "step": 21548 + }, + { + "epoch": 2.7412542933469024, + "grad_norm": 2.9455370903015137, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8979817628860474, + "num_tokens": 836428343.0, + "step": 21549 + }, + { + "epoch": 2.741381503625493, + "grad_norm": 2.7428505420684814, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9073978662490845, + "num_tokens": 836465819.0, + "step": 21550 + }, + { + "epoch": 2.7415087139040835, + "grad_norm": 2.831235408782959, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.907799482345581, + "num_tokens": 836506507.0, + "step": 21551 + }, + { + "epoch": 2.741635924182674, + "grad_norm": 3.0705606937408447, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9024534225463867, + "num_tokens": 836542252.0, + "step": 21552 + }, + { + "epoch": 2.7417631344612645, + "grad_norm": 2.7718005180358887, + "learning_rate": 1e-06, + "loss": 0.2264, + "mean_token_accuracy": 0.9159010648727417, + "num_tokens": 836580595.0, + "step": 21553 + }, + { + "epoch": 2.741890344739855, + "grad_norm": 2.80017352104187, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9087274074554443, + "num_tokens": 836622917.0, + "step": 21554 + }, + { + "epoch": 2.7420175550184456, + "grad_norm": 2.8634088039398193, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.903508186340332, + "num_tokens": 836663267.0, + "step": 21555 + }, + { + "epoch": 2.742144765297036, + "grad_norm": 2.8567113876342773, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9045710563659668, + "num_tokens": 836704103.0, + "step": 21556 + }, + { + "epoch": 2.7422719755756266, + "grad_norm": 3.071486473083496, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9007415771484375, + "num_tokens": 836738234.0, + "step": 21557 + }, + { + "epoch": 2.742399185854217, + "grad_norm": 2.6305899620056152, + "learning_rate": 1e-06, + "loss": 0.2177, + "mean_token_accuracy": 0.917776882648468, + "num_tokens": 836775572.0, + "step": 21558 + }, + { + "epoch": 2.7425263961328077, + "grad_norm": 2.6088480949401855, + "learning_rate": 1e-06, + "loss": 0.2229, + "mean_token_accuracy": 0.9173084497451782, + "num_tokens": 836815957.0, + "step": 21559 + }, + { + "epoch": 2.742653606411398, + "grad_norm": 2.9263391494750977, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9014580845832825, + "num_tokens": 836855235.0, + "step": 21560 + }, + { + "epoch": 2.7427808166899883, + "grad_norm": 2.600667715072632, + "learning_rate": 1e-06, + "loss": 0.2172, + "mean_token_accuracy": 0.9180443286895752, + "num_tokens": 836895085.0, + "step": 21561 + }, + { + "epoch": 2.7429080269685793, + "grad_norm": 2.9909446239471436, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8999606966972351, + "num_tokens": 836934005.0, + "step": 21562 + }, + { + "epoch": 2.7430352372471694, + "grad_norm": 2.970285177230835, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9079307317733765, + "num_tokens": 836970685.0, + "step": 21563 + }, + { + "epoch": 2.7431624475257603, + "grad_norm": 2.8976054191589355, + "learning_rate": 1e-06, + "loss": 0.2432, + "mean_token_accuracy": 0.9105894565582275, + "num_tokens": 837008203.0, + "step": 21564 + }, + { + "epoch": 2.7432896578043504, + "grad_norm": 3.313045024871826, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9050741195678711, + "num_tokens": 837037231.0, + "step": 21565 + }, + { + "epoch": 2.743416868082941, + "grad_norm": 2.717825412750244, + "learning_rate": 1e-06, + "loss": 0.2244, + "mean_token_accuracy": 0.9126505851745605, + "num_tokens": 837075695.0, + "step": 21566 + }, + { + "epoch": 2.7435440783615315, + "grad_norm": 2.7256174087524414, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9065855741500854, + "num_tokens": 837119138.0, + "step": 21567 + }, + { + "epoch": 2.743671288640122, + "grad_norm": 2.7534732818603516, + "learning_rate": 1e-06, + "loss": 0.2595, + "mean_token_accuracy": 0.903907060623169, + "num_tokens": 837164730.0, + "step": 21568 + }, + { + "epoch": 2.7437984989187125, + "grad_norm": 2.9059829711914062, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9050963521003723, + "num_tokens": 837202004.0, + "step": 21569 + }, + { + "epoch": 2.743925709197303, + "grad_norm": 2.735323905944824, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9017173051834106, + "num_tokens": 837245346.0, + "step": 21570 + }, + { + "epoch": 2.7440529194758936, + "grad_norm": 2.692004919052124, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9117957949638367, + "num_tokens": 837284988.0, + "step": 21571 + }, + { + "epoch": 2.744180129754484, + "grad_norm": 3.0234949588775635, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8958746194839478, + "num_tokens": 837323048.0, + "step": 21572 + }, + { + "epoch": 2.7443073400330746, + "grad_norm": 3.0016958713531494, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8979340195655823, + "num_tokens": 837361990.0, + "step": 21573 + }, + { + "epoch": 2.744434550311665, + "grad_norm": 3.235419750213623, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8985501527786255, + "num_tokens": 837394427.0, + "step": 21574 + }, + { + "epoch": 2.7445617605902557, + "grad_norm": 2.8633410930633545, + "learning_rate": 1e-06, + "loss": 0.242, + "mean_token_accuracy": 0.9087709188461304, + "num_tokens": 837430185.0, + "step": 21575 + }, + { + "epoch": 2.7446889708688462, + "grad_norm": 3.1223392486572266, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8958696126937866, + "num_tokens": 837467996.0, + "step": 21576 + }, + { + "epoch": 2.7448161811474368, + "grad_norm": 2.8465816974639893, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9051337242126465, + "num_tokens": 837505608.0, + "step": 21577 + }, + { + "epoch": 2.7449433914260273, + "grad_norm": 2.989689826965332, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9050147533416748, + "num_tokens": 837540873.0, + "step": 21578 + }, + { + "epoch": 2.745070601704618, + "grad_norm": 3.161184310913086, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9136878252029419, + "num_tokens": 837571063.0, + "step": 21579 + }, + { + "epoch": 2.7451978119832083, + "grad_norm": 3.282033681869507, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8967037796974182, + "num_tokens": 837605539.0, + "step": 21580 + }, + { + "epoch": 2.745325022261799, + "grad_norm": 2.850191831588745, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8951708078384399, + "num_tokens": 837650340.0, + "step": 21581 + }, + { + "epoch": 2.7454522325403894, + "grad_norm": 2.7514572143554688, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9036851525306702, + "num_tokens": 837692479.0, + "step": 21582 + }, + { + "epoch": 2.74557944281898, + "grad_norm": 2.9626965522766113, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9003578424453735, + "num_tokens": 837731526.0, + "step": 21583 + }, + { + "epoch": 2.7457066530975704, + "grad_norm": 3.0046286582946777, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9051001071929932, + "num_tokens": 837765412.0, + "step": 21584 + }, + { + "epoch": 2.745833863376161, + "grad_norm": 2.9122462272644043, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9013580083847046, + "num_tokens": 837807659.0, + "step": 21585 + }, + { + "epoch": 2.745961073654751, + "grad_norm": 2.816333770751953, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9056621789932251, + "num_tokens": 837850714.0, + "step": 21586 + }, + { + "epoch": 2.746088283933342, + "grad_norm": 2.98260235786438, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9122586846351624, + "num_tokens": 837887088.0, + "step": 21587 + }, + { + "epoch": 2.746215494211932, + "grad_norm": 2.9489693641662598, + "learning_rate": 1e-06, + "loss": 0.2226, + "mean_token_accuracy": 0.9162136316299438, + "num_tokens": 837918903.0, + "step": 21588 + }, + { + "epoch": 2.746342704490523, + "grad_norm": 3.1585745811462402, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.9038386344909668, + "num_tokens": 837955147.0, + "step": 21589 + }, + { + "epoch": 2.746469914769113, + "grad_norm": 2.855491876602173, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9076507091522217, + "num_tokens": 837990491.0, + "step": 21590 + }, + { + "epoch": 2.7465971250477037, + "grad_norm": 3.2057242393493652, + "learning_rate": 1e-06, + "loss": 0.3174, + "mean_token_accuracy": 0.883267343044281, + "num_tokens": 838027781.0, + "step": 21591 + }, + { + "epoch": 2.7467243353262942, + "grad_norm": 3.1304357051849365, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9030255079269409, + "num_tokens": 838061348.0, + "step": 21592 + }, + { + "epoch": 2.7468515456048848, + "grad_norm": 2.8195605278015137, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9011658430099487, + "num_tokens": 838102742.0, + "step": 21593 + }, + { + "epoch": 2.7469787558834753, + "grad_norm": 2.7266433238983154, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9061135053634644, + "num_tokens": 838144809.0, + "step": 21594 + }, + { + "epoch": 2.747105966162066, + "grad_norm": 2.8833634853363037, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9060400128364563, + "num_tokens": 838179986.0, + "step": 21595 + }, + { + "epoch": 2.7472331764406563, + "grad_norm": 2.7485268115997314, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.904195249080658, + "num_tokens": 838221661.0, + "step": 21596 + }, + { + "epoch": 2.747360386719247, + "grad_norm": 2.589914083480835, + "learning_rate": 1e-06, + "loss": 0.2258, + "mean_token_accuracy": 0.9158428907394409, + "num_tokens": 838261418.0, + "step": 21597 + }, + { + "epoch": 2.7474875969978374, + "grad_norm": 2.9796621799468994, + "learning_rate": 1e-06, + "loss": 0.3086, + "mean_token_accuracy": 0.8848152160644531, + "num_tokens": 838303094.0, + "step": 21598 + }, + { + "epoch": 2.747614807276428, + "grad_norm": 2.551595449447632, + "learning_rate": 1e-06, + "loss": 0.2224, + "mean_token_accuracy": 0.9172720909118652, + "num_tokens": 838347219.0, + "step": 21599 + }, + { + "epoch": 2.7477420175550185, + "grad_norm": 2.895545721054077, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9048357605934143, + "num_tokens": 838385583.0, + "step": 21600 + }, + { + "epoch": 2.747869227833609, + "grad_norm": 2.5413589477539062, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9153878688812256, + "num_tokens": 838431216.0, + "step": 21601 + }, + { + "epoch": 2.7479964381121995, + "grad_norm": 3.1617579460144043, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9026397466659546, + "num_tokens": 838461042.0, + "step": 21602 + }, + { + "epoch": 2.74812364839079, + "grad_norm": 2.9113337993621826, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.8986256122589111, + "num_tokens": 838498834.0, + "step": 21603 + }, + { + "epoch": 2.7482508586693806, + "grad_norm": 2.6936328411102295, + "learning_rate": 1e-06, + "loss": 0.228, + "mean_token_accuracy": 0.915667712688446, + "num_tokens": 838533391.0, + "step": 21604 + }, + { + "epoch": 2.748378068947971, + "grad_norm": 2.7312793731689453, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9042670130729675, + "num_tokens": 838575989.0, + "step": 21605 + }, + { + "epoch": 2.7485052792265616, + "grad_norm": 2.7733893394470215, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9085727334022522, + "num_tokens": 838616891.0, + "step": 21606 + }, + { + "epoch": 2.748632489505152, + "grad_norm": 3.004660129547119, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9115563035011292, + "num_tokens": 838648623.0, + "step": 21607 + }, + { + "epoch": 2.7487596997837427, + "grad_norm": 2.7482948303222656, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.908776581287384, + "num_tokens": 838688429.0, + "step": 21608 + }, + { + "epoch": 2.7488869100623328, + "grad_norm": 2.940749406814575, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9080482125282288, + "num_tokens": 838722170.0, + "step": 21609 + }, + { + "epoch": 2.7490141203409237, + "grad_norm": 3.009995222091675, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.8954670429229736, + "num_tokens": 838759190.0, + "step": 21610 + }, + { + "epoch": 2.749141330619514, + "grad_norm": 2.606477975845337, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8998367786407471, + "num_tokens": 838802891.0, + "step": 21611 + }, + { + "epoch": 2.749268540898105, + "grad_norm": 2.7798476219177246, + "learning_rate": 1e-06, + "loss": 0.3094, + "mean_token_accuracy": 0.8896257281303406, + "num_tokens": 838851082.0, + "step": 21612 + }, + { + "epoch": 2.749395751176695, + "grad_norm": 2.7875282764434814, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.898705244064331, + "num_tokens": 838894328.0, + "step": 21613 + }, + { + "epoch": 2.749522961455286, + "grad_norm": 2.738452196121216, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9079070687294006, + "num_tokens": 838933071.0, + "step": 21614 + }, + { + "epoch": 2.749650171733876, + "grad_norm": 2.7518861293792725, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9128785133361816, + "num_tokens": 838974656.0, + "step": 21615 + }, + { + "epoch": 2.7497773820124665, + "grad_norm": 2.736988067626953, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9066956043243408, + "num_tokens": 839017584.0, + "step": 21616 + }, + { + "epoch": 2.749904592291057, + "grad_norm": 2.7906572818756104, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.9147962927818298, + "num_tokens": 839055721.0, + "step": 21617 + }, + { + "epoch": 2.7500318025696475, + "grad_norm": 3.4149587154388428, + "learning_rate": 1e-06, + "loss": 0.3231, + "mean_token_accuracy": 0.8860743641853333, + "num_tokens": 839089441.0, + "step": 21618 + }, + { + "epoch": 2.750159012848238, + "grad_norm": 2.925511121749878, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.905468225479126, + "num_tokens": 839125500.0, + "step": 21619 + }, + { + "epoch": 2.7502862231268286, + "grad_norm": 2.9182934761047363, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.900046169757843, + "num_tokens": 839164638.0, + "step": 21620 + }, + { + "epoch": 2.750413433405419, + "grad_norm": 2.7529284954071045, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9070218801498413, + "num_tokens": 839203772.0, + "step": 21621 + }, + { + "epoch": 2.7505406436840096, + "grad_norm": 2.7603399753570557, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9106993079185486, + "num_tokens": 839241896.0, + "step": 21622 + }, + { + "epoch": 2.7506678539626, + "grad_norm": 2.8623716831207275, + "learning_rate": 1e-06, + "loss": 0.2218, + "mean_token_accuracy": 0.914960503578186, + "num_tokens": 839277349.0, + "step": 21623 + }, + { + "epoch": 2.7507950642411907, + "grad_norm": 2.7617299556732178, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.9106839895248413, + "num_tokens": 839313474.0, + "step": 21624 + }, + { + "epoch": 2.750922274519781, + "grad_norm": 3.2403368949890137, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8959541320800781, + "num_tokens": 839348220.0, + "step": 21625 + }, + { + "epoch": 2.7510494847983717, + "grad_norm": 2.7450942993164062, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9075257778167725, + "num_tokens": 839390053.0, + "step": 21626 + }, + { + "epoch": 2.7511766950769623, + "grad_norm": 3.4570465087890625, + "learning_rate": 1e-06, + "loss": 0.2855, + "mean_token_accuracy": 0.8983174562454224, + "num_tokens": 839422658.0, + "step": 21627 + }, + { + "epoch": 2.751303905355553, + "grad_norm": 3.4609875679016113, + "learning_rate": 1e-06, + "loss": 0.3129, + "mean_token_accuracy": 0.8898950219154358, + "num_tokens": 839455811.0, + "step": 21628 + }, + { + "epoch": 2.7514311156341433, + "grad_norm": 2.836956024169922, + "learning_rate": 1e-06, + "loss": 0.2329, + "mean_token_accuracy": 0.9154015183448792, + "num_tokens": 839492460.0, + "step": 21629 + }, + { + "epoch": 2.751558325912734, + "grad_norm": 2.9148945808410645, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9035360813140869, + "num_tokens": 839530346.0, + "step": 21630 + }, + { + "epoch": 2.7516855361913244, + "grad_norm": 2.552262306213379, + "learning_rate": 1e-06, + "loss": 0.2203, + "mean_token_accuracy": 0.9171109199523926, + "num_tokens": 839573168.0, + "step": 21631 + }, + { + "epoch": 2.751812746469915, + "grad_norm": 3.136981725692749, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.9008751511573792, + "num_tokens": 839605219.0, + "step": 21632 + }, + { + "epoch": 2.7519399567485054, + "grad_norm": 2.8274028301239014, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9045734405517578, + "num_tokens": 839645212.0, + "step": 21633 + }, + { + "epoch": 2.7520671670270955, + "grad_norm": 2.8886585235595703, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9077130556106567, + "num_tokens": 839680693.0, + "step": 21634 + }, + { + "epoch": 2.7521943773056865, + "grad_norm": 2.9483349323272705, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.904613733291626, + "num_tokens": 839717636.0, + "step": 21635 + }, + { + "epoch": 2.7523215875842766, + "grad_norm": 3.5532679557800293, + "learning_rate": 1e-06, + "loss": 0.334, + "mean_token_accuracy": 0.8793991804122925, + "num_tokens": 839750946.0, + "step": 21636 + }, + { + "epoch": 2.7524487978628676, + "grad_norm": 3.0291543006896973, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8906997442245483, + "num_tokens": 839789465.0, + "step": 21637 + }, + { + "epoch": 2.7525760081414576, + "grad_norm": 3.1951379776000977, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9041312336921692, + "num_tokens": 839821739.0, + "step": 21638 + }, + { + "epoch": 2.7527032184200486, + "grad_norm": 3.1256656646728516, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.895554780960083, + "num_tokens": 839857799.0, + "step": 21639 + }, + { + "epoch": 2.7528304286986387, + "grad_norm": 2.5868372917175293, + "learning_rate": 1e-06, + "loss": 0.2422, + "mean_token_accuracy": 0.9090510010719299, + "num_tokens": 839901945.0, + "step": 21640 + }, + { + "epoch": 2.7529576389772292, + "grad_norm": 3.06037974357605, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9023044109344482, + "num_tokens": 839939207.0, + "step": 21641 + }, + { + "epoch": 2.7530848492558198, + "grad_norm": 3.113107204437256, + "learning_rate": 1e-06, + "loss": 0.2709, + "mean_token_accuracy": 0.899202823638916, + "num_tokens": 839972912.0, + "step": 21642 + }, + { + "epoch": 2.7532120595344103, + "grad_norm": 2.990046739578247, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9015641808509827, + "num_tokens": 840010162.0, + "step": 21643 + }, + { + "epoch": 2.753339269813001, + "grad_norm": 2.9352617263793945, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9044944047927856, + "num_tokens": 840046877.0, + "step": 21644 + }, + { + "epoch": 2.7534664800915913, + "grad_norm": 3.0362634658813477, + "learning_rate": 1e-06, + "loss": 0.2426, + "mean_token_accuracy": 0.9101718664169312, + "num_tokens": 840079633.0, + "step": 21645 + }, + { + "epoch": 2.753593690370182, + "grad_norm": 2.6492843627929688, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9066439867019653, + "num_tokens": 840123231.0, + "step": 21646 + }, + { + "epoch": 2.7537209006487724, + "grad_norm": 2.6872031688690186, + "learning_rate": 1e-06, + "loss": 0.2192, + "mean_token_accuracy": 0.9186330437660217, + "num_tokens": 840159634.0, + "step": 21647 + }, + { + "epoch": 2.753848110927363, + "grad_norm": 3.0179078578948975, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9016227126121521, + "num_tokens": 840197256.0, + "step": 21648 + }, + { + "epoch": 2.7539753212059535, + "grad_norm": 2.749480962753296, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.9077954292297363, + "num_tokens": 840239370.0, + "step": 21649 + }, + { + "epoch": 2.754102531484544, + "grad_norm": 2.972628593444824, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.8992127776145935, + "num_tokens": 840279724.0, + "step": 21650 + }, + { + "epoch": 2.7542297417631345, + "grad_norm": 2.8713631629943848, + "learning_rate": 1e-06, + "loss": 0.2562, + "mean_token_accuracy": 0.9056189656257629, + "num_tokens": 840318505.0, + "step": 21651 + }, + { + "epoch": 2.754356952041725, + "grad_norm": 3.2062573432922363, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.889244019985199, + "num_tokens": 840358973.0, + "step": 21652 + }, + { + "epoch": 2.7544841623203156, + "grad_norm": 3.008024215698242, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8963468074798584, + "num_tokens": 840399310.0, + "step": 21653 + }, + { + "epoch": 2.754611372598906, + "grad_norm": 2.849311113357544, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9109072089195251, + "num_tokens": 840440252.0, + "step": 21654 + }, + { + "epoch": 2.7547385828774966, + "grad_norm": 2.7849695682525635, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9042055606842041, + "num_tokens": 840478983.0, + "step": 21655 + }, + { + "epoch": 2.754865793156087, + "grad_norm": 2.983492374420166, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8954928517341614, + "num_tokens": 840517744.0, + "step": 21656 + }, + { + "epoch": 2.7549930034346777, + "grad_norm": 3.1335368156433105, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9021524786949158, + "num_tokens": 840553129.0, + "step": 21657 + }, + { + "epoch": 2.755120213713268, + "grad_norm": 2.7827322483062744, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8956307172775269, + "num_tokens": 840595009.0, + "step": 21658 + }, + { + "epoch": 2.7552474239918583, + "grad_norm": 2.679605007171631, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9126905202865601, + "num_tokens": 840635199.0, + "step": 21659 + }, + { + "epoch": 2.7553746342704493, + "grad_norm": 2.9102630615234375, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.8996944427490234, + "num_tokens": 840675863.0, + "step": 21660 + }, + { + "epoch": 2.7555018445490393, + "grad_norm": 2.8649322986602783, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9009007215499878, + "num_tokens": 840715053.0, + "step": 21661 + }, + { + "epoch": 2.7556290548276303, + "grad_norm": 2.7580878734588623, + "learning_rate": 1e-06, + "loss": 0.234, + "mean_token_accuracy": 0.9145961999893188, + "num_tokens": 840752557.0, + "step": 21662 + }, + { + "epoch": 2.7557562651062204, + "grad_norm": 2.773395299911499, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9108134508132935, + "num_tokens": 840790745.0, + "step": 21663 + }, + { + "epoch": 2.755883475384811, + "grad_norm": 2.7924818992614746, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.9021369218826294, + "num_tokens": 840834734.0, + "step": 21664 + }, + { + "epoch": 2.7560106856634015, + "grad_norm": 2.6554207801818848, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9090188145637512, + "num_tokens": 840877417.0, + "step": 21665 + }, + { + "epoch": 2.756137895941992, + "grad_norm": 2.6541807651519775, + "learning_rate": 1e-06, + "loss": 0.2258, + "mean_token_accuracy": 0.9127677083015442, + "num_tokens": 840915482.0, + "step": 21666 + }, + { + "epoch": 2.7562651062205825, + "grad_norm": 2.926487922668457, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9023576378822327, + "num_tokens": 840956286.0, + "step": 21667 + }, + { + "epoch": 2.756392316499173, + "grad_norm": 2.6535305976867676, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.9001240730285645, + "num_tokens": 840998221.0, + "step": 21668 + }, + { + "epoch": 2.7565195267777636, + "grad_norm": 2.910266160964966, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.905703604221344, + "num_tokens": 841035756.0, + "step": 21669 + }, + { + "epoch": 2.756646737056354, + "grad_norm": 3.0796356201171875, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9090245962142944, + "num_tokens": 841069716.0, + "step": 21670 + }, + { + "epoch": 2.7567739473349446, + "grad_norm": 2.8855063915252686, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8983986377716064, + "num_tokens": 841110989.0, + "step": 21671 + }, + { + "epoch": 2.756901157613535, + "grad_norm": 2.9335098266601562, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9110054969787598, + "num_tokens": 841143234.0, + "step": 21672 + }, + { + "epoch": 2.7570283678921257, + "grad_norm": 2.6881675720214844, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9123122096061707, + "num_tokens": 841182958.0, + "step": 21673 + }, + { + "epoch": 2.757155578170716, + "grad_norm": 2.980064868927002, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9000645875930786, + "num_tokens": 841219315.0, + "step": 21674 + }, + { + "epoch": 2.7572827884493067, + "grad_norm": 2.8989098072052, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9020752310752869, + "num_tokens": 841261631.0, + "step": 21675 + }, + { + "epoch": 2.7574099987278973, + "grad_norm": 2.8795008659362793, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9042285680770874, + "num_tokens": 841298391.0, + "step": 21676 + }, + { + "epoch": 2.757537209006488, + "grad_norm": 2.9134552478790283, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9086884260177612, + "num_tokens": 841337866.0, + "step": 21677 + }, + { + "epoch": 2.7576644192850783, + "grad_norm": 2.8781650066375732, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9011222720146179, + "num_tokens": 841377565.0, + "step": 21678 + }, + { + "epoch": 2.757791629563669, + "grad_norm": 2.732170820236206, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9070439338684082, + "num_tokens": 841420022.0, + "step": 21679 + }, + { + "epoch": 2.7579188398422594, + "grad_norm": 2.9369356632232666, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9064595699310303, + "num_tokens": 841458652.0, + "step": 21680 + }, + { + "epoch": 2.75804605012085, + "grad_norm": 2.7847647666931152, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8938563466072083, + "num_tokens": 841502667.0, + "step": 21681 + }, + { + "epoch": 2.7581732603994404, + "grad_norm": 2.936149835586548, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9022767543792725, + "num_tokens": 841540086.0, + "step": 21682 + }, + { + "epoch": 2.758300470678031, + "grad_norm": 2.801262855529785, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9038583636283875, + "num_tokens": 841581745.0, + "step": 21683 + }, + { + "epoch": 2.758427680956621, + "grad_norm": 2.82765793800354, + "learning_rate": 1e-06, + "loss": 0.2392, + "mean_token_accuracy": 0.9099277257919312, + "num_tokens": 841620624.0, + "step": 21684 + }, + { + "epoch": 2.758554891235212, + "grad_norm": 2.8253374099731445, + "learning_rate": 1e-06, + "loss": 0.2303, + "mean_token_accuracy": 0.9141061305999756, + "num_tokens": 841658545.0, + "step": 21685 + }, + { + "epoch": 2.758682101513802, + "grad_norm": 3.077728748321533, + "learning_rate": 1e-06, + "loss": 0.2416, + "mean_token_accuracy": 0.9098613262176514, + "num_tokens": 841689026.0, + "step": 21686 + }, + { + "epoch": 2.758809311792393, + "grad_norm": 3.1848607063293457, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8983930349349976, + "num_tokens": 841723797.0, + "step": 21687 + }, + { + "epoch": 2.758936522070983, + "grad_norm": 3.0464212894439697, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9008229970932007, + "num_tokens": 841758663.0, + "step": 21688 + }, + { + "epoch": 2.7590637323495737, + "grad_norm": 3.092813014984131, + "learning_rate": 1e-06, + "loss": 0.303, + "mean_token_accuracy": 0.8922384977340698, + "num_tokens": 841799082.0, + "step": 21689 + }, + { + "epoch": 2.7591909426281642, + "grad_norm": 2.807448387145996, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9079451560974121, + "num_tokens": 841836179.0, + "step": 21690 + }, + { + "epoch": 2.7593181529067548, + "grad_norm": 3.0646731853485107, + "learning_rate": 1e-06, + "loss": 0.2735, + "mean_token_accuracy": 0.9008063077926636, + "num_tokens": 841871667.0, + "step": 21691 + }, + { + "epoch": 2.7594453631853453, + "grad_norm": 2.962679386138916, + "learning_rate": 1e-06, + "loss": 0.2508, + "mean_token_accuracy": 0.9086644649505615, + "num_tokens": 841906261.0, + "step": 21692 + }, + { + "epoch": 2.759572573463936, + "grad_norm": 2.713120460510254, + "learning_rate": 1e-06, + "loss": 0.2372, + "mean_token_accuracy": 0.9126294851303101, + "num_tokens": 841945962.0, + "step": 21693 + }, + { + "epoch": 2.7596997837425263, + "grad_norm": 2.680047035217285, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9096229076385498, + "num_tokens": 841987997.0, + "step": 21694 + }, + { + "epoch": 2.759826994021117, + "grad_norm": 2.8848533630371094, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9068496823310852, + "num_tokens": 842026894.0, + "step": 21695 + }, + { + "epoch": 2.7599542042997074, + "grad_norm": 3.1436946392059326, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9040687084197998, + "num_tokens": 842059863.0, + "step": 21696 + }, + { + "epoch": 2.760081414578298, + "grad_norm": 3.3704657554626465, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9067294001579285, + "num_tokens": 842090309.0, + "step": 21697 + }, + { + "epoch": 2.7602086248568884, + "grad_norm": 3.187361240386963, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9045357704162598, + "num_tokens": 842127146.0, + "step": 21698 + }, + { + "epoch": 2.760335835135479, + "grad_norm": 2.5609776973724365, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.910964846611023, + "num_tokens": 842167238.0, + "step": 21699 + }, + { + "epoch": 2.7604630454140695, + "grad_norm": 2.5629477500915527, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9081199169158936, + "num_tokens": 842211768.0, + "step": 21700 + }, + { + "epoch": 2.76059025569266, + "grad_norm": 3.01495623588562, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9030259251594543, + "num_tokens": 842249662.0, + "step": 21701 + }, + { + "epoch": 2.7607174659712506, + "grad_norm": 2.721971035003662, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9114229083061218, + "num_tokens": 842290613.0, + "step": 21702 + }, + { + "epoch": 2.760844676249841, + "grad_norm": 3.1745529174804688, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8972209692001343, + "num_tokens": 842328355.0, + "step": 21703 + }, + { + "epoch": 2.7609718865284316, + "grad_norm": 2.97064471244812, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9021317362785339, + "num_tokens": 842369024.0, + "step": 21704 + }, + { + "epoch": 2.761099096807022, + "grad_norm": 2.7701542377471924, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9062296152114868, + "num_tokens": 842412559.0, + "step": 21705 + }, + { + "epoch": 2.7612263070856127, + "grad_norm": 2.894474983215332, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9071862697601318, + "num_tokens": 842448757.0, + "step": 21706 + }, + { + "epoch": 2.7613535173642028, + "grad_norm": 2.5556559562683105, + "learning_rate": 1e-06, + "loss": 0.2068, + "mean_token_accuracy": 0.9215198159217834, + "num_tokens": 842487486.0, + "step": 21707 + }, + { + "epoch": 2.7614807276427937, + "grad_norm": 2.8655998706817627, + "learning_rate": 1e-06, + "loss": 0.2805, + "mean_token_accuracy": 0.8960254192352295, + "num_tokens": 842529360.0, + "step": 21708 + }, + { + "epoch": 2.761607937921384, + "grad_norm": 2.829181671142578, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9093856811523438, + "num_tokens": 842567054.0, + "step": 21709 + }, + { + "epoch": 2.761735148199975, + "grad_norm": 2.8395702838897705, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8988173007965088, + "num_tokens": 842610689.0, + "step": 21710 + }, + { + "epoch": 2.761862358478565, + "grad_norm": 2.6987175941467285, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9074730277061462, + "num_tokens": 842651832.0, + "step": 21711 + }, + { + "epoch": 2.761989568757156, + "grad_norm": 2.861154794692993, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.909072995185852, + "num_tokens": 842688049.0, + "step": 21712 + }, + { + "epoch": 2.762116779035746, + "grad_norm": 2.791978597640991, + "learning_rate": 1e-06, + "loss": 0.2201, + "mean_token_accuracy": 0.9177960157394409, + "num_tokens": 842725396.0, + "step": 21713 + }, + { + "epoch": 2.7622439893143365, + "grad_norm": 2.845842123031616, + "learning_rate": 1e-06, + "loss": 0.2351, + "mean_token_accuracy": 0.9130042791366577, + "num_tokens": 842764250.0, + "step": 21714 + }, + { + "epoch": 2.762371199592927, + "grad_norm": 2.7370054721832275, + "learning_rate": 1e-06, + "loss": 0.2921, + "mean_token_accuracy": 0.892400324344635, + "num_tokens": 842808356.0, + "step": 21715 + }, + { + "epoch": 2.7624984098715175, + "grad_norm": 3.036719560623169, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9081413745880127, + "num_tokens": 842841597.0, + "step": 21716 + }, + { + "epoch": 2.762625620150108, + "grad_norm": 2.8627030849456787, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9049068689346313, + "num_tokens": 842880403.0, + "step": 21717 + }, + { + "epoch": 2.7627528304286986, + "grad_norm": 2.6398792266845703, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.896886944770813, + "num_tokens": 842923064.0, + "step": 21718 + }, + { + "epoch": 2.762880040707289, + "grad_norm": 2.7435286045074463, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9019337892532349, + "num_tokens": 842965042.0, + "step": 21719 + }, + { + "epoch": 2.7630072509858796, + "grad_norm": 2.9070537090301514, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9123409390449524, + "num_tokens": 843003029.0, + "step": 21720 + }, + { + "epoch": 2.76313446126447, + "grad_norm": 2.8171353340148926, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9006527662277222, + "num_tokens": 843040967.0, + "step": 21721 + }, + { + "epoch": 2.7632616715430607, + "grad_norm": 2.8333537578582764, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8947468400001526, + "num_tokens": 843086303.0, + "step": 21722 + }, + { + "epoch": 2.763388881821651, + "grad_norm": 2.8718409538269043, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8963884115219116, + "num_tokens": 843123768.0, + "step": 21723 + }, + { + "epoch": 2.7635160921002417, + "grad_norm": 3.126107931137085, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.8966740369796753, + "num_tokens": 843159838.0, + "step": 21724 + }, + { + "epoch": 2.7636433023788323, + "grad_norm": 2.542065382003784, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.905289351940155, + "num_tokens": 843206310.0, + "step": 21725 + }, + { + "epoch": 2.763770512657423, + "grad_norm": 2.9953343868255615, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.902658224105835, + "num_tokens": 843244485.0, + "step": 21726 + }, + { + "epoch": 2.7638977229360133, + "grad_norm": 2.84804630279541, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.9116874933242798, + "num_tokens": 843282282.0, + "step": 21727 + }, + { + "epoch": 2.764024933214604, + "grad_norm": 3.1068811416625977, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8973040580749512, + "num_tokens": 843317057.0, + "step": 21728 + }, + { + "epoch": 2.7641521434931944, + "grad_norm": 2.7663207054138184, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.891933798789978, + "num_tokens": 843364301.0, + "step": 21729 + }, + { + "epoch": 2.764279353771785, + "grad_norm": 2.910980701446533, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.897687554359436, + "num_tokens": 843405389.0, + "step": 21730 + }, + { + "epoch": 2.7644065640503754, + "grad_norm": 2.9404826164245605, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9107617139816284, + "num_tokens": 843440380.0, + "step": 21731 + }, + { + "epoch": 2.7645337743289655, + "grad_norm": 2.8844990730285645, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9057192206382751, + "num_tokens": 843476994.0, + "step": 21732 + }, + { + "epoch": 2.7646609846075565, + "grad_norm": 3.263509750366211, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8925984501838684, + "num_tokens": 843507941.0, + "step": 21733 + }, + { + "epoch": 2.7647881948861466, + "grad_norm": 2.8345706462860107, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9036992192268372, + "num_tokens": 843547626.0, + "step": 21734 + }, + { + "epoch": 2.7649154051647375, + "grad_norm": 2.835909128189087, + "learning_rate": 1e-06, + "loss": 0.2229, + "mean_token_accuracy": 0.9160979986190796, + "num_tokens": 843582644.0, + "step": 21735 + }, + { + "epoch": 2.7650426154433276, + "grad_norm": 2.7719027996063232, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9037899374961853, + "num_tokens": 843624974.0, + "step": 21736 + }, + { + "epoch": 2.7651698257219186, + "grad_norm": 2.7580790519714355, + "learning_rate": 1e-06, + "loss": 0.2957, + "mean_token_accuracy": 0.8915019035339355, + "num_tokens": 843670896.0, + "step": 21737 + }, + { + "epoch": 2.7652970360005087, + "grad_norm": 2.896672487258911, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8971846103668213, + "num_tokens": 843711112.0, + "step": 21738 + }, + { + "epoch": 2.765424246279099, + "grad_norm": 2.910170793533325, + "learning_rate": 1e-06, + "loss": 0.2205, + "mean_token_accuracy": 0.9183686375617981, + "num_tokens": 843744188.0, + "step": 21739 + }, + { + "epoch": 2.7655514565576897, + "grad_norm": 2.7673094272613525, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9011648893356323, + "num_tokens": 843787559.0, + "step": 21740 + }, + { + "epoch": 2.7656786668362803, + "grad_norm": 3.239868640899658, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8920835256576538, + "num_tokens": 843825409.0, + "step": 21741 + }, + { + "epoch": 2.765805877114871, + "grad_norm": 2.660682201385498, + "learning_rate": 1e-06, + "loss": 0.2285, + "mean_token_accuracy": 0.9148606061935425, + "num_tokens": 843863874.0, + "step": 21742 + }, + { + "epoch": 2.7659330873934613, + "grad_norm": 2.9926466941833496, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9007616639137268, + "num_tokens": 843900235.0, + "step": 21743 + }, + { + "epoch": 2.766060297672052, + "grad_norm": 3.0309689044952393, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8932156562805176, + "num_tokens": 843939807.0, + "step": 21744 + }, + { + "epoch": 2.7661875079506424, + "grad_norm": 2.755359649658203, + "learning_rate": 1e-06, + "loss": 0.2295, + "mean_token_accuracy": 0.9140671491622925, + "num_tokens": 843977831.0, + "step": 21745 + }, + { + "epoch": 2.766314718229233, + "grad_norm": 2.6874642372131348, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9098823070526123, + "num_tokens": 844021245.0, + "step": 21746 + }, + { + "epoch": 2.7664419285078234, + "grad_norm": 2.9101881980895996, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9078786969184875, + "num_tokens": 844058491.0, + "step": 21747 + }, + { + "epoch": 2.766569138786414, + "grad_norm": 2.938476800918579, + "learning_rate": 1e-06, + "loss": 0.2848, + "mean_token_accuracy": 0.8955636620521545, + "num_tokens": 844099983.0, + "step": 21748 + }, + { + "epoch": 2.7666963490650045, + "grad_norm": 2.5522232055664062, + "learning_rate": 1e-06, + "loss": 0.2119, + "mean_token_accuracy": 0.9192233085632324, + "num_tokens": 844137805.0, + "step": 21749 + }, + { + "epoch": 2.766823559343595, + "grad_norm": 2.804647207260132, + "learning_rate": 1e-06, + "loss": 0.2917, + "mean_token_accuracy": 0.8916971683502197, + "num_tokens": 844181982.0, + "step": 21750 + }, + { + "epoch": 2.7669507696221856, + "grad_norm": 2.769871234893799, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9055460095405579, + "num_tokens": 844219809.0, + "step": 21751 + }, + { + "epoch": 2.767077979900776, + "grad_norm": 2.791853666305542, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9131230711936951, + "num_tokens": 844259377.0, + "step": 21752 + }, + { + "epoch": 2.7672051901793666, + "grad_norm": 2.847513437271118, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.8990199565887451, + "num_tokens": 844297862.0, + "step": 21753 + }, + { + "epoch": 2.767332400457957, + "grad_norm": 2.381300210952759, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9067494869232178, + "num_tokens": 844347816.0, + "step": 21754 + }, + { + "epoch": 2.7674596107365477, + "grad_norm": 2.90411639213562, + "learning_rate": 1e-06, + "loss": 0.235, + "mean_token_accuracy": 0.9074994325637817, + "num_tokens": 844382187.0, + "step": 21755 + }, + { + "epoch": 2.767586821015138, + "grad_norm": 2.754470109939575, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8969919085502625, + "num_tokens": 844426074.0, + "step": 21756 + }, + { + "epoch": 2.7677140312937283, + "grad_norm": 2.619856119155884, + "learning_rate": 1e-06, + "loss": 0.2391, + "mean_token_accuracy": 0.9125208258628845, + "num_tokens": 844466945.0, + "step": 21757 + }, + { + "epoch": 2.7678412415723193, + "grad_norm": 2.659013271331787, + "learning_rate": 1e-06, + "loss": 0.2368, + "mean_token_accuracy": 0.9110264778137207, + "num_tokens": 844507236.0, + "step": 21758 + }, + { + "epoch": 2.7679684518509093, + "grad_norm": 2.675034761428833, + "learning_rate": 1e-06, + "loss": 0.2296, + "mean_token_accuracy": 0.9157418608665466, + "num_tokens": 844544655.0, + "step": 21759 + }, + { + "epoch": 2.7680956621295003, + "grad_norm": 2.7764334678649902, + "learning_rate": 1e-06, + "loss": 0.2784, + "mean_token_accuracy": 0.8956211805343628, + "num_tokens": 844588648.0, + "step": 21760 + }, + { + "epoch": 2.7682228724080904, + "grad_norm": 2.7855327129364014, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9045045971870422, + "num_tokens": 844630571.0, + "step": 21761 + }, + { + "epoch": 2.768350082686681, + "grad_norm": 2.6928961277008057, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.8974324464797974, + "num_tokens": 844678108.0, + "step": 21762 + }, + { + "epoch": 2.7684772929652715, + "grad_norm": 2.960639476776123, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8973661661148071, + "num_tokens": 844715419.0, + "step": 21763 + }, + { + "epoch": 2.768604503243862, + "grad_norm": 2.8028206825256348, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9012715220451355, + "num_tokens": 844757025.0, + "step": 21764 + }, + { + "epoch": 2.7687317135224525, + "grad_norm": 2.791771173477173, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.90626060962677, + "num_tokens": 844794471.0, + "step": 21765 + }, + { + "epoch": 2.768858923801043, + "grad_norm": 3.01770281791687, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9131567478179932, + "num_tokens": 844826818.0, + "step": 21766 + }, + { + "epoch": 2.7689861340796336, + "grad_norm": 2.8503317832946777, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9119353294372559, + "num_tokens": 844862928.0, + "step": 21767 + }, + { + "epoch": 2.769113344358224, + "grad_norm": 2.857045888900757, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9070240259170532, + "num_tokens": 844900909.0, + "step": 21768 + }, + { + "epoch": 2.7692405546368146, + "grad_norm": 2.9638078212738037, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9083117246627808, + "num_tokens": 844939020.0, + "step": 21769 + }, + { + "epoch": 2.769367764915405, + "grad_norm": 3.011197090148926, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9144759774208069, + "num_tokens": 844971439.0, + "step": 21770 + }, + { + "epoch": 2.7694949751939957, + "grad_norm": 3.1838953495025635, + "learning_rate": 1e-06, + "loss": 0.3075, + "mean_token_accuracy": 0.8867251873016357, + "num_tokens": 845011031.0, + "step": 21771 + }, + { + "epoch": 2.769622185472586, + "grad_norm": 2.822420120239258, + "learning_rate": 1e-06, + "loss": 0.2446, + "mean_token_accuracy": 0.911239504814148, + "num_tokens": 845049724.0, + "step": 21772 + }, + { + "epoch": 2.7697493957511767, + "grad_norm": 2.6390879154205322, + "learning_rate": 1e-06, + "loss": 0.2426, + "mean_token_accuracy": 0.9099305868148804, + "num_tokens": 845090131.0, + "step": 21773 + }, + { + "epoch": 2.7698766060297673, + "grad_norm": 2.6692516803741455, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9016567468643188, + "num_tokens": 845135162.0, + "step": 21774 + }, + { + "epoch": 2.770003816308358, + "grad_norm": 2.4682469367980957, + "learning_rate": 1e-06, + "loss": 0.2348, + "mean_token_accuracy": 0.9139411449432373, + "num_tokens": 845180331.0, + "step": 21775 + }, + { + "epoch": 2.7701310265869483, + "grad_norm": 2.9034640789031982, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9032020568847656, + "num_tokens": 845219908.0, + "step": 21776 + }, + { + "epoch": 2.770258236865539, + "grad_norm": 3.0347018241882324, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8934099674224854, + "num_tokens": 845258465.0, + "step": 21777 + }, + { + "epoch": 2.7703854471441294, + "grad_norm": 2.95578670501709, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9028483629226685, + "num_tokens": 845294849.0, + "step": 21778 + }, + { + "epoch": 2.77051265742272, + "grad_norm": 2.872695207595825, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.8992822170257568, + "num_tokens": 845332992.0, + "step": 21779 + }, + { + "epoch": 2.77063986770131, + "grad_norm": 3.0067081451416016, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8938469886779785, + "num_tokens": 845371263.0, + "step": 21780 + }, + { + "epoch": 2.770767077979901, + "grad_norm": 2.864844560623169, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.9089218378067017, + "num_tokens": 845408895.0, + "step": 21781 + }, + { + "epoch": 2.770894288258491, + "grad_norm": 3.119204521179199, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9095175266265869, + "num_tokens": 845439856.0, + "step": 21782 + }, + { + "epoch": 2.771021498537082, + "grad_norm": 2.853804349899292, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9046766757965088, + "num_tokens": 845475359.0, + "step": 21783 + }, + { + "epoch": 2.771148708815672, + "grad_norm": 2.958650588989258, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8998004198074341, + "num_tokens": 845512023.0, + "step": 21784 + }, + { + "epoch": 2.771275919094263, + "grad_norm": 2.7717816829681396, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9115160703659058, + "num_tokens": 845550427.0, + "step": 21785 + }, + { + "epoch": 2.771403129372853, + "grad_norm": 2.7650651931762695, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9073594808578491, + "num_tokens": 845589217.0, + "step": 21786 + }, + { + "epoch": 2.7715303396514437, + "grad_norm": 2.5382843017578125, + "learning_rate": 1e-06, + "loss": 0.2253, + "mean_token_accuracy": 0.9152908325195312, + "num_tokens": 845630763.0, + "step": 21787 + }, + { + "epoch": 2.771657549930034, + "grad_norm": 2.9865944385528564, + "learning_rate": 1e-06, + "loss": 0.2769, + "mean_token_accuracy": 0.8997801542282104, + "num_tokens": 845671481.0, + "step": 21788 + }, + { + "epoch": 2.7717847602086247, + "grad_norm": 2.8073973655700684, + "learning_rate": 1e-06, + "loss": 0.2351, + "mean_token_accuracy": 0.910530686378479, + "num_tokens": 845708152.0, + "step": 21789 + }, + { + "epoch": 2.7719119704872153, + "grad_norm": 3.8568053245544434, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9058497548103333, + "num_tokens": 845745286.0, + "step": 21790 + }, + { + "epoch": 2.772039180765806, + "grad_norm": 3.1529417037963867, + "learning_rate": 1e-06, + "loss": 0.2778, + "mean_token_accuracy": 0.8969437479972839, + "num_tokens": 845780105.0, + "step": 21791 + }, + { + "epoch": 2.7721663910443963, + "grad_norm": 3.0780153274536133, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9043866395950317, + "num_tokens": 845812805.0, + "step": 21792 + }, + { + "epoch": 2.772293601322987, + "grad_norm": 2.7060866355895996, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9039326906204224, + "num_tokens": 845857170.0, + "step": 21793 + }, + { + "epoch": 2.7724208116015774, + "grad_norm": 2.8401169776916504, + "learning_rate": 1e-06, + "loss": 0.2835, + "mean_token_accuracy": 0.8982132077217102, + "num_tokens": 845902614.0, + "step": 21794 + }, + { + "epoch": 2.772548021880168, + "grad_norm": 3.130375862121582, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9050494432449341, + "num_tokens": 845940233.0, + "step": 21795 + }, + { + "epoch": 2.7726752321587584, + "grad_norm": 2.8285369873046875, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9043530225753784, + "num_tokens": 845976788.0, + "step": 21796 + }, + { + "epoch": 2.772802442437349, + "grad_norm": 2.876281976699829, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9049060940742493, + "num_tokens": 846016005.0, + "step": 21797 + }, + { + "epoch": 2.7729296527159395, + "grad_norm": 2.9041290283203125, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9070355892181396, + "num_tokens": 846054666.0, + "step": 21798 + }, + { + "epoch": 2.77305686299453, + "grad_norm": 3.047666311264038, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8911837339401245, + "num_tokens": 846095334.0, + "step": 21799 + }, + { + "epoch": 2.7731840732731206, + "grad_norm": 2.812854528427124, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.901086688041687, + "num_tokens": 846140924.0, + "step": 21800 + }, + { + "epoch": 2.773311283551711, + "grad_norm": 3.2980751991271973, + "learning_rate": 1e-06, + "loss": 0.3219, + "mean_token_accuracy": 0.8846786022186279, + "num_tokens": 846175294.0, + "step": 21801 + }, + { + "epoch": 2.7734384938303016, + "grad_norm": 3.096264123916626, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9037787914276123, + "num_tokens": 846211631.0, + "step": 21802 + }, + { + "epoch": 2.773565704108892, + "grad_norm": 2.7289698123931885, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9055861234664917, + "num_tokens": 846254949.0, + "step": 21803 + }, + { + "epoch": 2.7736929143874827, + "grad_norm": 3.0349209308624268, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.893395185470581, + "num_tokens": 846295094.0, + "step": 21804 + }, + { + "epoch": 2.7738201246660728, + "grad_norm": 2.860384464263916, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9008344411849976, + "num_tokens": 846333177.0, + "step": 21805 + }, + { + "epoch": 2.7739473349446637, + "grad_norm": 2.780061721801758, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.8994438648223877, + "num_tokens": 846373950.0, + "step": 21806 + }, + { + "epoch": 2.774074545223254, + "grad_norm": 2.96201753616333, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.9084850549697876, + "num_tokens": 846408307.0, + "step": 21807 + }, + { + "epoch": 2.774201755501845, + "grad_norm": 2.583179235458374, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9046883583068848, + "num_tokens": 846451262.0, + "step": 21808 + }, + { + "epoch": 2.774328965780435, + "grad_norm": 2.8104970455169678, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9076991677284241, + "num_tokens": 846491357.0, + "step": 21809 + }, + { + "epoch": 2.774456176059026, + "grad_norm": 2.6854007244110107, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9034978151321411, + "num_tokens": 846532079.0, + "step": 21810 + }, + { + "epoch": 2.774583386337616, + "grad_norm": 2.6188032627105713, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9048765897750854, + "num_tokens": 846577771.0, + "step": 21811 + }, + { + "epoch": 2.7747105966162064, + "grad_norm": 3.374105930328369, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9034467935562134, + "num_tokens": 846606800.0, + "step": 21812 + }, + { + "epoch": 2.774837806894797, + "grad_norm": 2.6174373626708984, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9035539627075195, + "num_tokens": 846652726.0, + "step": 21813 + }, + { + "epoch": 2.7749650171733875, + "grad_norm": 3.018561363220215, + "learning_rate": 1e-06, + "loss": 0.2849, + "mean_token_accuracy": 0.8956757187843323, + "num_tokens": 846694823.0, + "step": 21814 + }, + { + "epoch": 2.775092227451978, + "grad_norm": 2.8794589042663574, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8970438241958618, + "num_tokens": 846734945.0, + "step": 21815 + }, + { + "epoch": 2.7752194377305686, + "grad_norm": 3.0983479022979736, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.8989620208740234, + "num_tokens": 846772653.0, + "step": 21816 + }, + { + "epoch": 2.775346648009159, + "grad_norm": 2.7186710834503174, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.9062312245368958, + "num_tokens": 846814267.0, + "step": 21817 + }, + { + "epoch": 2.7754738582877496, + "grad_norm": 2.9885571002960205, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9132565259933472, + "num_tokens": 846849578.0, + "step": 21818 + }, + { + "epoch": 2.77560106856634, + "grad_norm": 2.9430911540985107, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9018394947052002, + "num_tokens": 846889347.0, + "step": 21819 + }, + { + "epoch": 2.7757282788449307, + "grad_norm": 2.636617660522461, + "learning_rate": 1e-06, + "loss": 0.2245, + "mean_token_accuracy": 0.9149346947669983, + "num_tokens": 846931570.0, + "step": 21820 + }, + { + "epoch": 2.775855489123521, + "grad_norm": 3.0970420837402344, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9095495939254761, + "num_tokens": 846962150.0, + "step": 21821 + }, + { + "epoch": 2.7759826994021117, + "grad_norm": 2.6064395904541016, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9016327857971191, + "num_tokens": 847006445.0, + "step": 21822 + }, + { + "epoch": 2.7761099096807023, + "grad_norm": 2.7336957454681396, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9041652679443359, + "num_tokens": 847050966.0, + "step": 21823 + }, + { + "epoch": 2.776237119959293, + "grad_norm": 2.8475887775421143, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.9003544449806213, + "num_tokens": 847090604.0, + "step": 21824 + }, + { + "epoch": 2.7763643302378833, + "grad_norm": 2.811887741088867, + "learning_rate": 1e-06, + "loss": 0.22, + "mean_token_accuracy": 0.917452335357666, + "num_tokens": 847127202.0, + "step": 21825 + }, + { + "epoch": 2.776491540516474, + "grad_norm": 2.79325795173645, + "learning_rate": 1e-06, + "loss": 0.2234, + "mean_token_accuracy": 0.9150717258453369, + "num_tokens": 847163256.0, + "step": 21826 + }, + { + "epoch": 2.7766187507950644, + "grad_norm": 2.9508819580078125, + "learning_rate": 1e-06, + "loss": 0.2968, + "mean_token_accuracy": 0.8903102874755859, + "num_tokens": 847203171.0, + "step": 21827 + }, + { + "epoch": 2.776745961073655, + "grad_norm": 3.106776475906372, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9046623706817627, + "num_tokens": 847235675.0, + "step": 21828 + }, + { + "epoch": 2.7768731713522454, + "grad_norm": 2.974780559539795, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9045363664627075, + "num_tokens": 847271643.0, + "step": 21829 + }, + { + "epoch": 2.7770003816308355, + "grad_norm": 3.0161638259887695, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.8966643810272217, + "num_tokens": 847309667.0, + "step": 21830 + }, + { + "epoch": 2.7771275919094265, + "grad_norm": 2.698812484741211, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9059935808181763, + "num_tokens": 847351119.0, + "step": 21831 + }, + { + "epoch": 2.7772548021880166, + "grad_norm": 3.1204328536987305, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9058603048324585, + "num_tokens": 847385659.0, + "step": 21832 + }, + { + "epoch": 2.7773820124666075, + "grad_norm": 2.6860897541046143, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8957028388977051, + "num_tokens": 847431484.0, + "step": 21833 + }, + { + "epoch": 2.7775092227451976, + "grad_norm": 2.8135483264923096, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9064512848854065, + "num_tokens": 847470609.0, + "step": 21834 + }, + { + "epoch": 2.7776364330237886, + "grad_norm": 2.9755797386169434, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9091007709503174, + "num_tokens": 847504295.0, + "step": 21835 + }, + { + "epoch": 2.7777636433023787, + "grad_norm": 3.0262451171875, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9056552648544312, + "num_tokens": 847541907.0, + "step": 21836 + }, + { + "epoch": 2.777890853580969, + "grad_norm": 2.9181814193725586, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9073199033737183, + "num_tokens": 847578531.0, + "step": 21837 + }, + { + "epoch": 2.7780180638595597, + "grad_norm": 2.9215033054351807, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9043428897857666, + "num_tokens": 847613694.0, + "step": 21838 + }, + { + "epoch": 2.7781452741381503, + "grad_norm": 2.9855639934539795, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8962728977203369, + "num_tokens": 847654707.0, + "step": 21839 + }, + { + "epoch": 2.778272484416741, + "grad_norm": 2.8427557945251465, + "learning_rate": 1e-06, + "loss": 0.228, + "mean_token_accuracy": 0.9147710204124451, + "num_tokens": 847687878.0, + "step": 21840 + }, + { + "epoch": 2.7783996946953313, + "grad_norm": 2.8826839923858643, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9006356000900269, + "num_tokens": 847730103.0, + "step": 21841 + }, + { + "epoch": 2.778526904973922, + "grad_norm": 2.9161746501922607, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.8980663418769836, + "num_tokens": 847769695.0, + "step": 21842 + }, + { + "epoch": 2.7786541152525124, + "grad_norm": 2.7920587062835693, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9043611288070679, + "num_tokens": 847809673.0, + "step": 21843 + }, + { + "epoch": 2.778781325531103, + "grad_norm": 2.6699869632720947, + "learning_rate": 1e-06, + "loss": 0.2366, + "mean_token_accuracy": 0.9112746715545654, + "num_tokens": 847849023.0, + "step": 21844 + }, + { + "epoch": 2.7789085358096934, + "grad_norm": 2.817117691040039, + "learning_rate": 1e-06, + "loss": 0.2439, + "mean_token_accuracy": 0.9099134802818298, + "num_tokens": 847886009.0, + "step": 21845 + }, + { + "epoch": 2.779035746088284, + "grad_norm": 2.706317186355591, + "learning_rate": 1e-06, + "loss": 0.2825, + "mean_token_accuracy": 0.897039532661438, + "num_tokens": 847931853.0, + "step": 21846 + }, + { + "epoch": 2.7791629563668745, + "grad_norm": 2.751019239425659, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9012101292610168, + "num_tokens": 847974716.0, + "step": 21847 + }, + { + "epoch": 2.779290166645465, + "grad_norm": 2.8787143230438232, + "learning_rate": 1e-06, + "loss": 0.2355, + "mean_token_accuracy": 0.9145193099975586, + "num_tokens": 848012244.0, + "step": 21848 + }, + { + "epoch": 2.7794173769240555, + "grad_norm": 3.107064962387085, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9013140201568604, + "num_tokens": 848049696.0, + "step": 21849 + }, + { + "epoch": 2.779544587202646, + "grad_norm": 2.954887628555298, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9033281207084656, + "num_tokens": 848087640.0, + "step": 21850 + }, + { + "epoch": 2.7796717974812366, + "grad_norm": 3.015650510787964, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9047924876213074, + "num_tokens": 848126420.0, + "step": 21851 + }, + { + "epoch": 2.779799007759827, + "grad_norm": 2.9247846603393555, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8991612195968628, + "num_tokens": 848162098.0, + "step": 21852 + }, + { + "epoch": 2.7799262180384177, + "grad_norm": 3.003715753555298, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9024258852005005, + "num_tokens": 848196528.0, + "step": 21853 + }, + { + "epoch": 2.780053428317008, + "grad_norm": 2.9283971786499023, + "learning_rate": 1e-06, + "loss": 0.2383, + "mean_token_accuracy": 0.9112077951431274, + "num_tokens": 848227621.0, + "step": 21854 + }, + { + "epoch": 2.7801806385955983, + "grad_norm": 2.713470935821533, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9123870134353638, + "num_tokens": 848267711.0, + "step": 21855 + }, + { + "epoch": 2.7803078488741892, + "grad_norm": 2.7245097160339355, + "learning_rate": 1e-06, + "loss": 0.2214, + "mean_token_accuracy": 0.9173585176467896, + "num_tokens": 848305661.0, + "step": 21856 + }, + { + "epoch": 2.7804350591527793, + "grad_norm": 2.8166160583496094, + "learning_rate": 1e-06, + "loss": 0.3124, + "mean_token_accuracy": 0.8853225111961365, + "num_tokens": 848349556.0, + "step": 21857 + }, + { + "epoch": 2.7805622694313703, + "grad_norm": 2.8527989387512207, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9071117639541626, + "num_tokens": 848385442.0, + "step": 21858 + }, + { + "epoch": 2.7806894797099604, + "grad_norm": 2.732607126235962, + "learning_rate": 1e-06, + "loss": 0.2437, + "mean_token_accuracy": 0.9092633724212646, + "num_tokens": 848425010.0, + "step": 21859 + }, + { + "epoch": 2.780816689988551, + "grad_norm": 2.69477915763855, + "learning_rate": 1e-06, + "loss": 0.2118, + "mean_token_accuracy": 0.9211362600326538, + "num_tokens": 848462534.0, + "step": 21860 + }, + { + "epoch": 2.7809439002671414, + "grad_norm": 2.918281078338623, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8953105211257935, + "num_tokens": 848499751.0, + "step": 21861 + }, + { + "epoch": 2.781071110545732, + "grad_norm": 2.7984871864318848, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9160352349281311, + "num_tokens": 848536731.0, + "step": 21862 + }, + { + "epoch": 2.7811983208243225, + "grad_norm": 2.69836163520813, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.9127805233001709, + "num_tokens": 848577595.0, + "step": 21863 + }, + { + "epoch": 2.781325531102913, + "grad_norm": 2.8073158264160156, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9015772342681885, + "num_tokens": 848617517.0, + "step": 21864 + }, + { + "epoch": 2.7814527413815036, + "grad_norm": 2.8419320583343506, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.907863974571228, + "num_tokens": 848656353.0, + "step": 21865 + }, + { + "epoch": 2.781579951660094, + "grad_norm": 2.7114315032958984, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8976354002952576, + "num_tokens": 848698664.0, + "step": 21866 + }, + { + "epoch": 2.7817071619386846, + "grad_norm": 2.8459408283233643, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.902015209197998, + "num_tokens": 848743008.0, + "step": 21867 + }, + { + "epoch": 2.781834372217275, + "grad_norm": 3.039933681488037, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9044607877731323, + "num_tokens": 848776749.0, + "step": 21868 + }, + { + "epoch": 2.7819615824958657, + "grad_norm": 2.6657025814056396, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9046660661697388, + "num_tokens": 848819465.0, + "step": 21869 + }, + { + "epoch": 2.782088792774456, + "grad_norm": 2.920689582824707, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9026660919189453, + "num_tokens": 848858201.0, + "step": 21870 + }, + { + "epoch": 2.7822160030530467, + "grad_norm": 2.921811103820801, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.8991070985794067, + "num_tokens": 848896959.0, + "step": 21871 + }, + { + "epoch": 2.7823432133316373, + "grad_norm": 2.734712600708008, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9032144546508789, + "num_tokens": 848938625.0, + "step": 21872 + }, + { + "epoch": 2.782470423610228, + "grad_norm": 2.4798195362091064, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9108193516731262, + "num_tokens": 848981856.0, + "step": 21873 + }, + { + "epoch": 2.7825976338888183, + "grad_norm": 2.6513009071350098, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9076972603797913, + "num_tokens": 849021444.0, + "step": 21874 + }, + { + "epoch": 2.782724844167409, + "grad_norm": 3.217665195465088, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8948373198509216, + "num_tokens": 849057884.0, + "step": 21875 + }, + { + "epoch": 2.7828520544459994, + "grad_norm": 2.836667060852051, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.897993803024292, + "num_tokens": 849099971.0, + "step": 21876 + }, + { + "epoch": 2.78297926472459, + "grad_norm": 3.169544219970703, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8921707272529602, + "num_tokens": 849139559.0, + "step": 21877 + }, + { + "epoch": 2.78310647500318, + "grad_norm": 2.935006618499756, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9070581197738647, + "num_tokens": 849177468.0, + "step": 21878 + }, + { + "epoch": 2.783233685281771, + "grad_norm": 2.9687650203704834, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.8962198495864868, + "num_tokens": 849216922.0, + "step": 21879 + }, + { + "epoch": 2.783360895560361, + "grad_norm": 2.8296778202056885, + "learning_rate": 1e-06, + "loss": 0.2317, + "mean_token_accuracy": 0.9137281775474548, + "num_tokens": 849251059.0, + "step": 21880 + }, + { + "epoch": 2.783488105838952, + "grad_norm": 3.1489462852478027, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8983919620513916, + "num_tokens": 849287656.0, + "step": 21881 + }, + { + "epoch": 2.783615316117542, + "grad_norm": 2.80251145362854, + "learning_rate": 1e-06, + "loss": 0.2247, + "mean_token_accuracy": 0.9154402017593384, + "num_tokens": 849324109.0, + "step": 21882 + }, + { + "epoch": 2.783742526396133, + "grad_norm": 2.7135279178619385, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9105702042579651, + "num_tokens": 849366444.0, + "step": 21883 + }, + { + "epoch": 2.783869736674723, + "grad_norm": 3.0554401874542236, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9075422286987305, + "num_tokens": 849398331.0, + "step": 21884 + }, + { + "epoch": 2.7839969469533137, + "grad_norm": 2.5420963764190674, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.9137805104255676, + "num_tokens": 849441008.0, + "step": 21885 + }, + { + "epoch": 2.784124157231904, + "grad_norm": 2.8241262435913086, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9041993618011475, + "num_tokens": 849483014.0, + "step": 21886 + }, + { + "epoch": 2.7842513675104947, + "grad_norm": 2.743178606033325, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9108784198760986, + "num_tokens": 849525286.0, + "step": 21887 + }, + { + "epoch": 2.7843785777890853, + "grad_norm": 2.663386583328247, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9057985544204712, + "num_tokens": 849569056.0, + "step": 21888 + }, + { + "epoch": 2.784505788067676, + "grad_norm": 3.1868157386779785, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.898040235042572, + "num_tokens": 849604810.0, + "step": 21889 + }, + { + "epoch": 2.7846329983462663, + "grad_norm": 2.8223867416381836, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.8996570706367493, + "num_tokens": 849645742.0, + "step": 21890 + }, + { + "epoch": 2.784760208624857, + "grad_norm": 2.739805221557617, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9051235914230347, + "num_tokens": 849686142.0, + "step": 21891 + }, + { + "epoch": 2.7848874189034474, + "grad_norm": 2.7843050956726074, + "learning_rate": 1e-06, + "loss": 0.2487, + "mean_token_accuracy": 0.9074179530143738, + "num_tokens": 849724333.0, + "step": 21892 + }, + { + "epoch": 2.785014629182038, + "grad_norm": 2.8987441062927246, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.902736485004425, + "num_tokens": 849762912.0, + "step": 21893 + }, + { + "epoch": 2.7851418394606284, + "grad_norm": 2.8736016750335693, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9130164384841919, + "num_tokens": 849797697.0, + "step": 21894 + }, + { + "epoch": 2.785269049739219, + "grad_norm": 2.7356417179107666, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9098606109619141, + "num_tokens": 849836245.0, + "step": 21895 + }, + { + "epoch": 2.7853962600178095, + "grad_norm": 3.154552698135376, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.906071662902832, + "num_tokens": 849867389.0, + "step": 21896 + }, + { + "epoch": 2.7855234702964, + "grad_norm": 3.034034013748169, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.9020455479621887, + "num_tokens": 849903552.0, + "step": 21897 + }, + { + "epoch": 2.7856506805749905, + "grad_norm": 2.7564516067504883, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9055696129798889, + "num_tokens": 849943351.0, + "step": 21898 + }, + { + "epoch": 2.785777890853581, + "grad_norm": 3.016378402709961, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.9014825224876404, + "num_tokens": 849979303.0, + "step": 21899 + }, + { + "epoch": 2.7859051011321716, + "grad_norm": 2.8903050422668457, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9017153382301331, + "num_tokens": 850023096.0, + "step": 21900 + }, + { + "epoch": 2.786032311410762, + "grad_norm": 3.145498275756836, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9095569252967834, + "num_tokens": 850060424.0, + "step": 21901 + }, + { + "epoch": 2.7861595216893527, + "grad_norm": 2.756857395172119, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.90147864818573, + "num_tokens": 850106684.0, + "step": 21902 + }, + { + "epoch": 2.7862867319679427, + "grad_norm": 2.930718183517456, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9143482446670532, + "num_tokens": 850141200.0, + "step": 21903 + }, + { + "epoch": 2.7864139422465337, + "grad_norm": 2.790978193283081, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8994138240814209, + "num_tokens": 850181013.0, + "step": 21904 + }, + { + "epoch": 2.786541152525124, + "grad_norm": 2.8357675075531006, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.8984900712966919, + "num_tokens": 850221671.0, + "step": 21905 + }, + { + "epoch": 2.7866683628037148, + "grad_norm": 2.78224778175354, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9046691656112671, + "num_tokens": 850264646.0, + "step": 21906 + }, + { + "epoch": 2.786795573082305, + "grad_norm": 2.7346813678741455, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9034788012504578, + "num_tokens": 850307997.0, + "step": 21907 + }, + { + "epoch": 2.786922783360896, + "grad_norm": 2.7895352840423584, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9064116477966309, + "num_tokens": 850349134.0, + "step": 21908 + }, + { + "epoch": 2.787049993639486, + "grad_norm": 2.880253791809082, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9068464040756226, + "num_tokens": 850390905.0, + "step": 21909 + }, + { + "epoch": 2.7871772039180764, + "grad_norm": 2.881927490234375, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9016856551170349, + "num_tokens": 850432264.0, + "step": 21910 + }, + { + "epoch": 2.787304414196667, + "grad_norm": 2.907895565032959, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8964029550552368, + "num_tokens": 850470110.0, + "step": 21911 + }, + { + "epoch": 2.7874316244752575, + "grad_norm": 2.7285373210906982, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9102680683135986, + "num_tokens": 850509471.0, + "step": 21912 + }, + { + "epoch": 2.787558834753848, + "grad_norm": 2.6676392555236816, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9075958728790283, + "num_tokens": 850551434.0, + "step": 21913 + }, + { + "epoch": 2.7876860450324386, + "grad_norm": 2.6463170051574707, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8975386023521423, + "num_tokens": 850596047.0, + "step": 21914 + }, + { + "epoch": 2.787813255311029, + "grad_norm": 2.800351619720459, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9065858721733093, + "num_tokens": 850637084.0, + "step": 21915 + }, + { + "epoch": 2.7879404655896196, + "grad_norm": 2.887895107269287, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9081885814666748, + "num_tokens": 850672036.0, + "step": 21916 + }, + { + "epoch": 2.78806767586821, + "grad_norm": 2.962477445602417, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.909875750541687, + "num_tokens": 850705742.0, + "step": 21917 + }, + { + "epoch": 2.7881948861468007, + "grad_norm": 3.3515892028808594, + "learning_rate": 1e-06, + "loss": 0.3345, + "mean_token_accuracy": 0.8822914958000183, + "num_tokens": 850742276.0, + "step": 21918 + }, + { + "epoch": 2.788322096425391, + "grad_norm": 2.7914280891418457, + "learning_rate": 1e-06, + "loss": 0.3, + "mean_token_accuracy": 0.8921293020248413, + "num_tokens": 850786454.0, + "step": 21919 + }, + { + "epoch": 2.7884493067039817, + "grad_norm": 2.6110920906066895, + "learning_rate": 1e-06, + "loss": 0.2201, + "mean_token_accuracy": 0.9210336208343506, + "num_tokens": 850826050.0, + "step": 21920 + }, + { + "epoch": 2.7885765169825723, + "grad_norm": 3.0740652084350586, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.8994669914245605, + "num_tokens": 850866122.0, + "step": 21921 + }, + { + "epoch": 2.788703727261163, + "grad_norm": 2.7117578983306885, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9143951535224915, + "num_tokens": 850904888.0, + "step": 21922 + }, + { + "epoch": 2.7888309375397533, + "grad_norm": 2.8408398628234863, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9052286148071289, + "num_tokens": 850945895.0, + "step": 21923 + }, + { + "epoch": 2.788958147818344, + "grad_norm": 2.9341893196105957, + "learning_rate": 1e-06, + "loss": 0.2991, + "mean_token_accuracy": 0.8928312063217163, + "num_tokens": 850989512.0, + "step": 21924 + }, + { + "epoch": 2.7890853580969344, + "grad_norm": 2.6463799476623535, + "learning_rate": 1e-06, + "loss": 0.2288, + "mean_token_accuracy": 0.9149647951126099, + "num_tokens": 851030456.0, + "step": 21925 + }, + { + "epoch": 2.789212568375525, + "grad_norm": 2.7046945095062256, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9050775170326233, + "num_tokens": 851074761.0, + "step": 21926 + }, + { + "epoch": 2.7893397786541154, + "grad_norm": 3.1981308460235596, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9120467305183411, + "num_tokens": 851107243.0, + "step": 21927 + }, + { + "epoch": 2.7894669889327055, + "grad_norm": 2.9434823989868164, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9066100120544434, + "num_tokens": 851147056.0, + "step": 21928 + }, + { + "epoch": 2.7895941992112965, + "grad_norm": 3.325981855392456, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9024898409843445, + "num_tokens": 851176121.0, + "step": 21929 + }, + { + "epoch": 2.7897214094898866, + "grad_norm": 3.063676357269287, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9025853872299194, + "num_tokens": 851208392.0, + "step": 21930 + }, + { + "epoch": 2.7898486197684775, + "grad_norm": 2.5657761096954346, + "learning_rate": 1e-06, + "loss": 0.2301, + "mean_token_accuracy": 0.9153158664703369, + "num_tokens": 851252163.0, + "step": 21931 + }, + { + "epoch": 2.7899758300470676, + "grad_norm": 2.9231934547424316, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9001295566558838, + "num_tokens": 851290007.0, + "step": 21932 + }, + { + "epoch": 2.7901030403256586, + "grad_norm": 3.411747932434082, + "learning_rate": 1e-06, + "loss": 0.2345, + "mean_token_accuracy": 0.9091964960098267, + "num_tokens": 851318221.0, + "step": 21933 + }, + { + "epoch": 2.7902302506042487, + "grad_norm": 3.1522412300109863, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.9021873474121094, + "num_tokens": 851355104.0, + "step": 21934 + }, + { + "epoch": 2.790357460882839, + "grad_norm": 3.223799467086792, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8927208781242371, + "num_tokens": 851392918.0, + "step": 21935 + }, + { + "epoch": 2.7904846711614297, + "grad_norm": 2.833251953125, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.908121645450592, + "num_tokens": 851432172.0, + "step": 21936 + }, + { + "epoch": 2.7906118814400203, + "grad_norm": 2.5207571983337402, + "learning_rate": 1e-06, + "loss": 0.206, + "mean_token_accuracy": 0.9231446385383606, + "num_tokens": 851475074.0, + "step": 21937 + }, + { + "epoch": 2.790739091718611, + "grad_norm": 2.93830943107605, + "learning_rate": 1e-06, + "loss": 0.2413, + "mean_token_accuracy": 0.9125381112098694, + "num_tokens": 851510198.0, + "step": 21938 + }, + { + "epoch": 2.7908663019972013, + "grad_norm": 2.991950750350952, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8950873017311096, + "num_tokens": 851551720.0, + "step": 21939 + }, + { + "epoch": 2.790993512275792, + "grad_norm": 2.954826593399048, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9060057401657104, + "num_tokens": 851587033.0, + "step": 21940 + }, + { + "epoch": 2.7911207225543824, + "grad_norm": 3.2645812034606934, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9043576121330261, + "num_tokens": 851619320.0, + "step": 21941 + }, + { + "epoch": 2.791247932832973, + "grad_norm": 2.7486467361450195, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9064234495162964, + "num_tokens": 851662228.0, + "step": 21942 + }, + { + "epoch": 2.7913751431115634, + "grad_norm": 2.7983365058898926, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9006229639053345, + "num_tokens": 851702826.0, + "step": 21943 + }, + { + "epoch": 2.791502353390154, + "grad_norm": 3.0016608238220215, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.909632682800293, + "num_tokens": 851736720.0, + "step": 21944 + }, + { + "epoch": 2.7916295636687445, + "grad_norm": 2.816770553588867, + "learning_rate": 1e-06, + "loss": 0.2939, + "mean_token_accuracy": 0.8928420543670654, + "num_tokens": 851780284.0, + "step": 21945 + }, + { + "epoch": 2.791756773947335, + "grad_norm": 3.0017292499542236, + "learning_rate": 1e-06, + "loss": 0.2268, + "mean_token_accuracy": 0.9182548522949219, + "num_tokens": 851818587.0, + "step": 21946 + }, + { + "epoch": 2.7918839842259255, + "grad_norm": 2.8900938034057617, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9084399938583374, + "num_tokens": 851854626.0, + "step": 21947 + }, + { + "epoch": 2.792011194504516, + "grad_norm": 3.128336191177368, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9013529419898987, + "num_tokens": 851890733.0, + "step": 21948 + }, + { + "epoch": 2.7921384047831066, + "grad_norm": 2.681710720062256, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8994996547698975, + "num_tokens": 851933360.0, + "step": 21949 + }, + { + "epoch": 2.792265615061697, + "grad_norm": 2.6948235034942627, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9074045419692993, + "num_tokens": 851976988.0, + "step": 21950 + }, + { + "epoch": 2.7923928253402877, + "grad_norm": 2.6603174209594727, + "learning_rate": 1e-06, + "loss": 0.2669, + "mean_token_accuracy": 0.9020481705665588, + "num_tokens": 852020208.0, + "step": 21951 + }, + { + "epoch": 2.792520035618878, + "grad_norm": 2.9705519676208496, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8968896865844727, + "num_tokens": 852056973.0, + "step": 21952 + }, + { + "epoch": 2.7926472458974683, + "grad_norm": 2.9889984130859375, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9026138186454773, + "num_tokens": 852094157.0, + "step": 21953 + }, + { + "epoch": 2.7927744561760592, + "grad_norm": 2.802243947982788, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9091719388961792, + "num_tokens": 852133168.0, + "step": 21954 + }, + { + "epoch": 2.7929016664546493, + "grad_norm": 2.7563388347625732, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9104365706443787, + "num_tokens": 852174153.0, + "step": 21955 + }, + { + "epoch": 2.7930288767332403, + "grad_norm": 3.0455265045166016, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9007885456085205, + "num_tokens": 852208375.0, + "step": 21956 + }, + { + "epoch": 2.7931560870118304, + "grad_norm": 2.5380451679229736, + "learning_rate": 1e-06, + "loss": 0.2369, + "mean_token_accuracy": 0.9127424955368042, + "num_tokens": 852250147.0, + "step": 21957 + }, + { + "epoch": 2.793283297290421, + "grad_norm": 2.8723630905151367, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9051940441131592, + "num_tokens": 852288610.0, + "step": 21958 + }, + { + "epoch": 2.7934105075690114, + "grad_norm": 2.716484308242798, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9026750326156616, + "num_tokens": 852331914.0, + "step": 21959 + }, + { + "epoch": 2.793537717847602, + "grad_norm": 3.051736354827881, + "learning_rate": 1e-06, + "loss": 0.2489, + "mean_token_accuracy": 0.9082655906677246, + "num_tokens": 852367392.0, + "step": 21960 + }, + { + "epoch": 2.7936649281261925, + "grad_norm": 3.6619718074798584, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8966193199157715, + "num_tokens": 852397776.0, + "step": 21961 + }, + { + "epoch": 2.793792138404783, + "grad_norm": 3.223466157913208, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8954213857650757, + "num_tokens": 852438355.0, + "step": 21962 + }, + { + "epoch": 2.7939193486833735, + "grad_norm": 2.7058444023132324, + "learning_rate": 1e-06, + "loss": 0.2371, + "mean_token_accuracy": 0.9117238521575928, + "num_tokens": 852476541.0, + "step": 21963 + }, + { + "epoch": 2.794046558961964, + "grad_norm": 3.3300728797912598, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8970901966094971, + "num_tokens": 852507354.0, + "step": 21964 + }, + { + "epoch": 2.7941737692405546, + "grad_norm": 3.075115442276001, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.8978323340415955, + "num_tokens": 852542443.0, + "step": 21965 + }, + { + "epoch": 2.794300979519145, + "grad_norm": 2.827432632446289, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.8983829021453857, + "num_tokens": 852583807.0, + "step": 21966 + }, + { + "epoch": 2.7944281897977357, + "grad_norm": 2.6023592948913574, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9112522602081299, + "num_tokens": 852625558.0, + "step": 21967 + }, + { + "epoch": 2.794555400076326, + "grad_norm": 2.566256523132324, + "learning_rate": 1e-06, + "loss": 0.2347, + "mean_token_accuracy": 0.9133843183517456, + "num_tokens": 852666614.0, + "step": 21968 + }, + { + "epoch": 2.7946826103549167, + "grad_norm": 2.993058443069458, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9028269052505493, + "num_tokens": 852705776.0, + "step": 21969 + }, + { + "epoch": 2.7948098206335072, + "grad_norm": 2.7408905029296875, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9106142520904541, + "num_tokens": 852744129.0, + "step": 21970 + }, + { + "epoch": 2.7949370309120978, + "grad_norm": 2.6008570194244385, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9040666818618774, + "num_tokens": 852788415.0, + "step": 21971 + }, + { + "epoch": 2.7950642411906883, + "grad_norm": 3.3361423015594482, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8952336311340332, + "num_tokens": 852821337.0, + "step": 21972 + }, + { + "epoch": 2.795191451469279, + "grad_norm": 2.650564193725586, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9044426083564758, + "num_tokens": 852865936.0, + "step": 21973 + }, + { + "epoch": 2.7953186617478694, + "grad_norm": 3.080894947052002, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8960837125778198, + "num_tokens": 852906651.0, + "step": 21974 + }, + { + "epoch": 2.79544587202646, + "grad_norm": 2.97623610496521, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9065554141998291, + "num_tokens": 852945530.0, + "step": 21975 + }, + { + "epoch": 2.79557308230505, + "grad_norm": 2.7855348587036133, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9117651581764221, + "num_tokens": 852984117.0, + "step": 21976 + }, + { + "epoch": 2.795700292583641, + "grad_norm": 2.9953835010528564, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9024330973625183, + "num_tokens": 853020262.0, + "step": 21977 + }, + { + "epoch": 2.795827502862231, + "grad_norm": 3.1660304069519043, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.9012630581855774, + "num_tokens": 853057218.0, + "step": 21978 + }, + { + "epoch": 2.795954713140822, + "grad_norm": 2.583385467529297, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9135075211524963, + "num_tokens": 853096325.0, + "step": 21979 + }, + { + "epoch": 2.796081923419412, + "grad_norm": 2.571364402770996, + "learning_rate": 1e-06, + "loss": 0.2342, + "mean_token_accuracy": 0.9129588603973389, + "num_tokens": 853137432.0, + "step": 21980 + }, + { + "epoch": 2.796209133698003, + "grad_norm": 3.1236038208007812, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9053655862808228, + "num_tokens": 853175491.0, + "step": 21981 + }, + { + "epoch": 2.796336343976593, + "grad_norm": 2.8232662677764893, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8979303240776062, + "num_tokens": 853216347.0, + "step": 21982 + }, + { + "epoch": 2.7964635542551837, + "grad_norm": 3.0513081550598145, + "learning_rate": 1e-06, + "loss": 0.2837, + "mean_token_accuracy": 0.8987524509429932, + "num_tokens": 853254840.0, + "step": 21983 + }, + { + "epoch": 2.796590764533774, + "grad_norm": 2.805781841278076, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9056304693222046, + "num_tokens": 853294479.0, + "step": 21984 + }, + { + "epoch": 2.7967179748123647, + "grad_norm": 2.682880401611328, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.912068247795105, + "num_tokens": 853333086.0, + "step": 21985 + }, + { + "epoch": 2.7968451850909553, + "grad_norm": 3.3600664138793945, + "learning_rate": 1e-06, + "loss": 0.2946, + "mean_token_accuracy": 0.8929743766784668, + "num_tokens": 853366766.0, + "step": 21986 + }, + { + "epoch": 2.796972395369546, + "grad_norm": 2.8413875102996826, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9025405049324036, + "num_tokens": 853406953.0, + "step": 21987 + }, + { + "epoch": 2.7970996056481363, + "grad_norm": 2.887802839279175, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9005074501037598, + "num_tokens": 853448292.0, + "step": 21988 + }, + { + "epoch": 2.797226815926727, + "grad_norm": 3.024345874786377, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9004561901092529, + "num_tokens": 853484922.0, + "step": 21989 + }, + { + "epoch": 2.7973540262053174, + "grad_norm": 3.0002658367156982, + "learning_rate": 1e-06, + "loss": 0.2975, + "mean_token_accuracy": 0.8931072354316711, + "num_tokens": 853524885.0, + "step": 21990 + }, + { + "epoch": 2.797481236483908, + "grad_norm": 2.79750657081604, + "learning_rate": 1e-06, + "loss": 0.2306, + "mean_token_accuracy": 0.9126988053321838, + "num_tokens": 853563938.0, + "step": 21991 + }, + { + "epoch": 2.7976084467624984, + "grad_norm": 2.6243348121643066, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9071279764175415, + "num_tokens": 853609412.0, + "step": 21992 + }, + { + "epoch": 2.797735657041089, + "grad_norm": 2.787266254425049, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9048025608062744, + "num_tokens": 853649863.0, + "step": 21993 + }, + { + "epoch": 2.7978628673196795, + "grad_norm": 2.7695841789245605, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9028207063674927, + "num_tokens": 853691913.0, + "step": 21994 + }, + { + "epoch": 2.79799007759827, + "grad_norm": 2.543504238128662, + "learning_rate": 1e-06, + "loss": 0.2245, + "mean_token_accuracy": 0.9157304763793945, + "num_tokens": 853732629.0, + "step": 21995 + }, + { + "epoch": 2.7981172878768605, + "grad_norm": 3.046560287475586, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9072034955024719, + "num_tokens": 853766611.0, + "step": 21996 + }, + { + "epoch": 2.798244498155451, + "grad_norm": 2.8057351112365723, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9052506685256958, + "num_tokens": 853804825.0, + "step": 21997 + }, + { + "epoch": 2.7983717084340416, + "grad_norm": 2.8568007946014404, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9111247062683105, + "num_tokens": 853839461.0, + "step": 21998 + }, + { + "epoch": 2.798498918712632, + "grad_norm": 2.7845349311828613, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.900486946105957, + "num_tokens": 853881038.0, + "step": 21999 + }, + { + "epoch": 2.7986261289912227, + "grad_norm": 3.023297071456909, + "learning_rate": 1e-06, + "loss": 0.2886, + "mean_token_accuracy": 0.8921784162521362, + "num_tokens": 853921347.0, + "step": 22000 + }, + { + "epoch": 2.7987533392698127, + "grad_norm": 3.0457842350006104, + "learning_rate": 1e-06, + "loss": 0.3074, + "mean_token_accuracy": 0.8856617212295532, + "num_tokens": 853957605.0, + "step": 22001 + }, + { + "epoch": 2.7988805495484037, + "grad_norm": 2.7379183769226074, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9059497117996216, + "num_tokens": 853998089.0, + "step": 22002 + }, + { + "epoch": 2.799007759826994, + "grad_norm": 2.634737491607666, + "learning_rate": 1e-06, + "loss": 0.2781, + "mean_token_accuracy": 0.8985583782196045, + "num_tokens": 854044779.0, + "step": 22003 + }, + { + "epoch": 2.7991349701055848, + "grad_norm": 2.6148431301116943, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9113247394561768, + "num_tokens": 854083485.0, + "step": 22004 + }, + { + "epoch": 2.799262180384175, + "grad_norm": 2.9517147541046143, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.9004380702972412, + "num_tokens": 854119706.0, + "step": 22005 + }, + { + "epoch": 2.799389390662766, + "grad_norm": 2.7777323722839355, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.900558352470398, + "num_tokens": 854158936.0, + "step": 22006 + }, + { + "epoch": 2.799516600941356, + "grad_norm": 2.6923344135284424, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.8982219696044922, + "num_tokens": 854200413.0, + "step": 22007 + }, + { + "epoch": 2.7996438112199464, + "grad_norm": 2.7666850090026855, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9090992212295532, + "num_tokens": 854238335.0, + "step": 22008 + }, + { + "epoch": 2.799771021498537, + "grad_norm": 3.0141234397888184, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9059282541275024, + "num_tokens": 854274113.0, + "step": 22009 + }, + { + "epoch": 2.7998982317771275, + "grad_norm": 2.9150493144989014, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9011743664741516, + "num_tokens": 854313308.0, + "step": 22010 + }, + { + "epoch": 2.800025442055718, + "grad_norm": 2.9069674015045166, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9053425788879395, + "num_tokens": 854349441.0, + "step": 22011 + }, + { + "epoch": 2.8001526523343085, + "grad_norm": 2.94101619720459, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9022761583328247, + "num_tokens": 854385288.0, + "step": 22012 + }, + { + "epoch": 2.800279862612899, + "grad_norm": 2.9031436443328857, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9089198112487793, + "num_tokens": 854421257.0, + "step": 22013 + }, + { + "epoch": 2.8004070728914896, + "grad_norm": 2.8008673191070557, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9034717082977295, + "num_tokens": 854465840.0, + "step": 22014 + }, + { + "epoch": 2.80053428317008, + "grad_norm": 2.831637144088745, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.9050760269165039, + "num_tokens": 854504840.0, + "step": 22015 + }, + { + "epoch": 2.8006614934486707, + "grad_norm": 3.0017924308776855, + "learning_rate": 1e-06, + "loss": 0.2885, + "mean_token_accuracy": 0.8948720097541809, + "num_tokens": 854544182.0, + "step": 22016 + }, + { + "epoch": 2.800788703727261, + "grad_norm": 2.989623546600342, + "learning_rate": 1e-06, + "loss": 0.2875, + "mean_token_accuracy": 0.8922654390335083, + "num_tokens": 854581545.0, + "step": 22017 + }, + { + "epoch": 2.8009159140058517, + "grad_norm": 2.7919821739196777, + "learning_rate": 1e-06, + "loss": 0.231, + "mean_token_accuracy": 0.9134116768836975, + "num_tokens": 854617129.0, + "step": 22018 + }, + { + "epoch": 2.8010431242844422, + "grad_norm": 2.7750234603881836, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9069116115570068, + "num_tokens": 854661246.0, + "step": 22019 + }, + { + "epoch": 2.8011703345630328, + "grad_norm": 3.0246057510375977, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.894600510597229, + "num_tokens": 854698742.0, + "step": 22020 + }, + { + "epoch": 2.8012975448416233, + "grad_norm": 2.727815628051758, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9074169397354126, + "num_tokens": 854742329.0, + "step": 22021 + }, + { + "epoch": 2.801424755120214, + "grad_norm": 2.8819680213928223, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9062041640281677, + "num_tokens": 854780289.0, + "step": 22022 + }, + { + "epoch": 2.8015519653988044, + "grad_norm": 2.8378264904022217, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.897200882434845, + "num_tokens": 854822274.0, + "step": 22023 + }, + { + "epoch": 2.801679175677395, + "grad_norm": 2.8798375129699707, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.89653480052948, + "num_tokens": 854863307.0, + "step": 22024 + }, + { + "epoch": 2.8018063859559854, + "grad_norm": 2.8398349285125732, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9015201330184937, + "num_tokens": 854900728.0, + "step": 22025 + }, + { + "epoch": 2.8019335962345755, + "grad_norm": 2.7286782264709473, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9105033874511719, + "num_tokens": 854940933.0, + "step": 22026 + }, + { + "epoch": 2.8020608065131665, + "grad_norm": 2.8924474716186523, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9039163589477539, + "num_tokens": 854976463.0, + "step": 22027 + }, + { + "epoch": 2.8021880167917566, + "grad_norm": 2.6733193397521973, + "learning_rate": 1e-06, + "loss": 0.2283, + "mean_token_accuracy": 0.9156535863876343, + "num_tokens": 855016604.0, + "step": 22028 + }, + { + "epoch": 2.8023152270703475, + "grad_norm": 2.660602331161499, + "learning_rate": 1e-06, + "loss": 0.2398, + "mean_token_accuracy": 0.9113166332244873, + "num_tokens": 855057317.0, + "step": 22029 + }, + { + "epoch": 2.8024424373489376, + "grad_norm": 3.1351511478424072, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9011730551719666, + "num_tokens": 855091349.0, + "step": 22030 + }, + { + "epoch": 2.8025696476275286, + "grad_norm": 2.825232982635498, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.90541672706604, + "num_tokens": 855130696.0, + "step": 22031 + }, + { + "epoch": 2.8026968579061187, + "grad_norm": 2.7064359188079834, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9121150374412537, + "num_tokens": 855170172.0, + "step": 22032 + }, + { + "epoch": 2.802824068184709, + "grad_norm": 3.0673739910125732, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.8989665508270264, + "num_tokens": 855201775.0, + "step": 22033 + }, + { + "epoch": 2.8029512784632997, + "grad_norm": 2.5802204608917236, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9118306040763855, + "num_tokens": 855246317.0, + "step": 22034 + }, + { + "epoch": 2.8030784887418903, + "grad_norm": 3.079915761947632, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9030405879020691, + "num_tokens": 855283174.0, + "step": 22035 + }, + { + "epoch": 2.803205699020481, + "grad_norm": 2.7637341022491455, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9092473983764648, + "num_tokens": 855323618.0, + "step": 22036 + }, + { + "epoch": 2.8033329092990713, + "grad_norm": 2.9550344944000244, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9061715006828308, + "num_tokens": 855358575.0, + "step": 22037 + }, + { + "epoch": 2.803460119577662, + "grad_norm": 2.7221171855926514, + "learning_rate": 1e-06, + "loss": 0.2212, + "mean_token_accuracy": 0.9165585041046143, + "num_tokens": 855397567.0, + "step": 22038 + }, + { + "epoch": 2.8035873298562524, + "grad_norm": 2.8334484100341797, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8959888219833374, + "num_tokens": 855441468.0, + "step": 22039 + }, + { + "epoch": 2.803714540134843, + "grad_norm": 2.885429620742798, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9095333218574524, + "num_tokens": 855479606.0, + "step": 22040 + }, + { + "epoch": 2.8038417504134334, + "grad_norm": 2.6802899837493896, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.900738000869751, + "num_tokens": 855524198.0, + "step": 22041 + }, + { + "epoch": 2.803968960692024, + "grad_norm": 2.8964734077453613, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.901367723941803, + "num_tokens": 855562755.0, + "step": 22042 + }, + { + "epoch": 2.8040961709706145, + "grad_norm": 2.826669931411743, + "learning_rate": 1e-06, + "loss": 0.2993, + "mean_token_accuracy": 0.8919205069541931, + "num_tokens": 855604708.0, + "step": 22043 + }, + { + "epoch": 2.804223381249205, + "grad_norm": 3.0268170833587646, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.909119725227356, + "num_tokens": 855643118.0, + "step": 22044 + }, + { + "epoch": 2.8043505915277955, + "grad_norm": 2.675503730773926, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9028599858283997, + "num_tokens": 855686105.0, + "step": 22045 + }, + { + "epoch": 2.804477801806386, + "grad_norm": 3.044585704803467, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9008721113204956, + "num_tokens": 855723031.0, + "step": 22046 + }, + { + "epoch": 2.8046050120849766, + "grad_norm": 2.8462655544281006, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9033280611038208, + "num_tokens": 855758803.0, + "step": 22047 + }, + { + "epoch": 2.804732222363567, + "grad_norm": 2.927668333053589, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9079849123954773, + "num_tokens": 855793681.0, + "step": 22048 + }, + { + "epoch": 2.8048594326421576, + "grad_norm": 2.8705103397369385, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9080357551574707, + "num_tokens": 855828190.0, + "step": 22049 + }, + { + "epoch": 2.804986642920748, + "grad_norm": 2.817748785018921, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.894913375377655, + "num_tokens": 855870685.0, + "step": 22050 + }, + { + "epoch": 2.8051138531993383, + "grad_norm": 2.676259756088257, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9105402231216431, + "num_tokens": 855908881.0, + "step": 22051 + }, + { + "epoch": 2.8052410634779292, + "grad_norm": 2.9781534671783447, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9033423662185669, + "num_tokens": 855945251.0, + "step": 22052 + }, + { + "epoch": 2.8053682737565193, + "grad_norm": 2.7939887046813965, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8970690965652466, + "num_tokens": 855989328.0, + "step": 22053 + }, + { + "epoch": 2.8054954840351103, + "grad_norm": 2.989133596420288, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.905677080154419, + "num_tokens": 856026712.0, + "step": 22054 + }, + { + "epoch": 2.8056226943137004, + "grad_norm": 3.041090250015259, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8956620097160339, + "num_tokens": 856063493.0, + "step": 22055 + }, + { + "epoch": 2.805749904592291, + "grad_norm": 3.2554550170898438, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9060194492340088, + "num_tokens": 856093011.0, + "step": 22056 + }, + { + "epoch": 2.8058771148708814, + "grad_norm": 2.705897808074951, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9015294313430786, + "num_tokens": 856139407.0, + "step": 22057 + }, + { + "epoch": 2.806004325149472, + "grad_norm": 2.8862173557281494, + "learning_rate": 1e-06, + "loss": 0.2889, + "mean_token_accuracy": 0.8922520875930786, + "num_tokens": 856181040.0, + "step": 22058 + }, + { + "epoch": 2.8061315354280625, + "grad_norm": 3.0471315383911133, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8970808982849121, + "num_tokens": 856215291.0, + "step": 22059 + }, + { + "epoch": 2.806258745706653, + "grad_norm": 2.811704397201538, + "learning_rate": 1e-06, + "loss": 0.2523, + "mean_token_accuracy": 0.9039801359176636, + "num_tokens": 856255617.0, + "step": 22060 + }, + { + "epoch": 2.8063859559852435, + "grad_norm": 2.8242557048797607, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.8986563682556152, + "num_tokens": 856296090.0, + "step": 22061 + }, + { + "epoch": 2.806513166263834, + "grad_norm": 2.823946237564087, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9001213312149048, + "num_tokens": 856337336.0, + "step": 22062 + }, + { + "epoch": 2.8066403765424246, + "grad_norm": 2.8968164920806885, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.8990048766136169, + "num_tokens": 856376082.0, + "step": 22063 + }, + { + "epoch": 2.806767586821015, + "grad_norm": 3.018305778503418, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9054521322250366, + "num_tokens": 856413933.0, + "step": 22064 + }, + { + "epoch": 2.8068947970996057, + "grad_norm": 2.803264856338501, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9017873406410217, + "num_tokens": 856451924.0, + "step": 22065 + }, + { + "epoch": 2.807022007378196, + "grad_norm": 2.639296054840088, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9080684185028076, + "num_tokens": 856496995.0, + "step": 22066 + }, + { + "epoch": 2.8071492176567867, + "grad_norm": 2.8291616439819336, + "learning_rate": 1e-06, + "loss": 0.2411, + "mean_token_accuracy": 0.9073688983917236, + "num_tokens": 856533825.0, + "step": 22067 + }, + { + "epoch": 2.8072764279353772, + "grad_norm": 2.989468574523926, + "learning_rate": 1e-06, + "loss": 0.2435, + "mean_token_accuracy": 0.9072136282920837, + "num_tokens": 856570940.0, + "step": 22068 + }, + { + "epoch": 2.8074036382139678, + "grad_norm": 2.983407497406006, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9027344584465027, + "num_tokens": 856608127.0, + "step": 22069 + }, + { + "epoch": 2.8075308484925583, + "grad_norm": 2.6604669094085693, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.9013388156890869, + "num_tokens": 856652967.0, + "step": 22070 + }, + { + "epoch": 2.807658058771149, + "grad_norm": 2.6529572010040283, + "learning_rate": 1e-06, + "loss": 0.2337, + "mean_token_accuracy": 0.9131412506103516, + "num_tokens": 856689239.0, + "step": 22071 + }, + { + "epoch": 2.8077852690497394, + "grad_norm": 3.144490957260132, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9058048129081726, + "num_tokens": 856719351.0, + "step": 22072 + }, + { + "epoch": 2.80791247932833, + "grad_norm": 2.8681180477142334, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9056200981140137, + "num_tokens": 856755999.0, + "step": 22073 + }, + { + "epoch": 2.80803968960692, + "grad_norm": 2.869643211364746, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9034349918365479, + "num_tokens": 856796043.0, + "step": 22074 + }, + { + "epoch": 2.808166899885511, + "grad_norm": 2.8815910816192627, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9014175534248352, + "num_tokens": 856835417.0, + "step": 22075 + }, + { + "epoch": 2.808294110164101, + "grad_norm": 2.9250292778015137, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.8999745845794678, + "num_tokens": 856874287.0, + "step": 22076 + }, + { + "epoch": 2.808421320442692, + "grad_norm": 3.4101929664611816, + "learning_rate": 1e-06, + "loss": 0.3289, + "mean_token_accuracy": 0.882610559463501, + "num_tokens": 856909456.0, + "step": 22077 + }, + { + "epoch": 2.808548530721282, + "grad_norm": 2.878417491912842, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9071239829063416, + "num_tokens": 856947675.0, + "step": 22078 + }, + { + "epoch": 2.808675740999873, + "grad_norm": 3.090914487838745, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9076789617538452, + "num_tokens": 856981131.0, + "step": 22079 + }, + { + "epoch": 2.808802951278463, + "grad_norm": 2.98995041847229, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.9029090404510498, + "num_tokens": 857017936.0, + "step": 22080 + }, + { + "epoch": 2.8089301615570537, + "grad_norm": 2.991185426712036, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9110002517700195, + "num_tokens": 857051616.0, + "step": 22081 + }, + { + "epoch": 2.809057371835644, + "grad_norm": 3.1166915893554688, + "learning_rate": 1e-06, + "loss": 0.3053, + "mean_token_accuracy": 0.8927122354507446, + "num_tokens": 857090626.0, + "step": 22082 + }, + { + "epoch": 2.8091845821142347, + "grad_norm": 2.9457743167877197, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9055740237236023, + "num_tokens": 857126603.0, + "step": 22083 + }, + { + "epoch": 2.8093117923928252, + "grad_norm": 2.715672492980957, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9105141162872314, + "num_tokens": 857164056.0, + "step": 22084 + }, + { + "epoch": 2.8094390026714158, + "grad_norm": 2.963745355606079, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.9033106565475464, + "num_tokens": 857203030.0, + "step": 22085 + }, + { + "epoch": 2.8095662129500063, + "grad_norm": 3.274127960205078, + "learning_rate": 1e-06, + "loss": 0.2706, + "mean_token_accuracy": 0.8996986150741577, + "num_tokens": 857233022.0, + "step": 22086 + }, + { + "epoch": 2.809693423228597, + "grad_norm": 2.717028856277466, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9065676331520081, + "num_tokens": 857276327.0, + "step": 22087 + }, + { + "epoch": 2.8098206335071874, + "grad_norm": 3.0127925872802734, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8952867984771729, + "num_tokens": 857317531.0, + "step": 22088 + }, + { + "epoch": 2.809947843785778, + "grad_norm": 2.658802032470703, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9064655303955078, + "num_tokens": 857359149.0, + "step": 22089 + }, + { + "epoch": 2.8100750540643684, + "grad_norm": 3.198996067047119, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9041261076927185, + "num_tokens": 857392655.0, + "step": 22090 + }, + { + "epoch": 2.810202264342959, + "grad_norm": 2.7652690410614014, + "learning_rate": 1e-06, + "loss": 0.2207, + "mean_token_accuracy": 0.9158502817153931, + "num_tokens": 857429600.0, + "step": 22091 + }, + { + "epoch": 2.8103294746215495, + "grad_norm": 2.8740718364715576, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9056873321533203, + "num_tokens": 857466203.0, + "step": 22092 + }, + { + "epoch": 2.81045668490014, + "grad_norm": 2.666900634765625, + "learning_rate": 1e-06, + "loss": 0.2345, + "mean_token_accuracy": 0.9144992828369141, + "num_tokens": 857508258.0, + "step": 22093 + }, + { + "epoch": 2.8105838951787305, + "grad_norm": 2.8076367378234863, + "learning_rate": 1e-06, + "loss": 0.2248, + "mean_token_accuracy": 0.9144172072410583, + "num_tokens": 857542323.0, + "step": 22094 + }, + { + "epoch": 2.810711105457321, + "grad_norm": 2.6989543437957764, + "learning_rate": 1e-06, + "loss": 0.2136, + "mean_token_accuracy": 0.9191097021102905, + "num_tokens": 857578716.0, + "step": 22095 + }, + { + "epoch": 2.8108383157359116, + "grad_norm": 2.707211494445801, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9103060960769653, + "num_tokens": 857622317.0, + "step": 22096 + }, + { + "epoch": 2.810965526014502, + "grad_norm": 2.8363490104675293, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8965377807617188, + "num_tokens": 857666917.0, + "step": 22097 + }, + { + "epoch": 2.8110927362930926, + "grad_norm": 2.9424588680267334, + "learning_rate": 1e-06, + "loss": 0.3177, + "mean_token_accuracy": 0.8875279426574707, + "num_tokens": 857718452.0, + "step": 22098 + }, + { + "epoch": 2.8112199465716827, + "grad_norm": 2.969045639038086, + "learning_rate": 1e-06, + "loss": 0.2411, + "mean_token_accuracy": 0.9114250540733337, + "num_tokens": 857749709.0, + "step": 22099 + }, + { + "epoch": 2.8113471568502737, + "grad_norm": 2.947315216064453, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9062897562980652, + "num_tokens": 857784842.0, + "step": 22100 + }, + { + "epoch": 2.811474367128864, + "grad_norm": 2.8095855712890625, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9018782377243042, + "num_tokens": 857826994.0, + "step": 22101 + }, + { + "epoch": 2.8116015774074548, + "grad_norm": 2.7974820137023926, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9049067497253418, + "num_tokens": 857867352.0, + "step": 22102 + }, + { + "epoch": 2.811728787686045, + "grad_norm": 2.91082501411438, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8951621055603027, + "num_tokens": 857904003.0, + "step": 22103 + }, + { + "epoch": 2.811855997964636, + "grad_norm": 3.1608433723449707, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.8992784023284912, + "num_tokens": 857942430.0, + "step": 22104 + }, + { + "epoch": 2.811983208243226, + "grad_norm": 2.794163227081299, + "learning_rate": 1e-06, + "loss": 0.2405, + "mean_token_accuracy": 0.9111377596855164, + "num_tokens": 857978410.0, + "step": 22105 + }, + { + "epoch": 2.8121104185218164, + "grad_norm": 2.827069044113159, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.896759033203125, + "num_tokens": 858021340.0, + "step": 22106 + }, + { + "epoch": 2.812237628800407, + "grad_norm": 2.727428913116455, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9099017381668091, + "num_tokens": 858059780.0, + "step": 22107 + }, + { + "epoch": 2.8123648390789975, + "grad_norm": 2.825112819671631, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8988299369812012, + "num_tokens": 858101286.0, + "step": 22108 + }, + { + "epoch": 2.812492049357588, + "grad_norm": 3.003682851791382, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9038355350494385, + "num_tokens": 858137923.0, + "step": 22109 + }, + { + "epoch": 2.8126192596361785, + "grad_norm": 2.768444776535034, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9010876417160034, + "num_tokens": 858182175.0, + "step": 22110 + }, + { + "epoch": 2.812746469914769, + "grad_norm": 2.8483283519744873, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9076757431030273, + "num_tokens": 858218953.0, + "step": 22111 + }, + { + "epoch": 2.8128736801933596, + "grad_norm": 2.801753282546997, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8989720344543457, + "num_tokens": 858258092.0, + "step": 22112 + }, + { + "epoch": 2.81300089047195, + "grad_norm": 2.7588887214660645, + "learning_rate": 1e-06, + "loss": 0.2353, + "mean_token_accuracy": 0.9124287366867065, + "num_tokens": 858297442.0, + "step": 22113 + }, + { + "epoch": 2.8131281007505406, + "grad_norm": 2.773214101791382, + "learning_rate": 1e-06, + "loss": 0.2729, + "mean_token_accuracy": 0.9000795483589172, + "num_tokens": 858340636.0, + "step": 22114 + }, + { + "epoch": 2.813255311029131, + "grad_norm": 2.9124622344970703, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9137287735939026, + "num_tokens": 858376116.0, + "step": 22115 + }, + { + "epoch": 2.8133825213077217, + "grad_norm": 2.8770577907562256, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9069450497627258, + "num_tokens": 858412898.0, + "step": 22116 + }, + { + "epoch": 2.8135097315863122, + "grad_norm": 2.8501293659210205, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9030351638793945, + "num_tokens": 858451146.0, + "step": 22117 + }, + { + "epoch": 2.8136369418649028, + "grad_norm": 2.5681090354919434, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9020366072654724, + "num_tokens": 858494460.0, + "step": 22118 + }, + { + "epoch": 2.8137641521434933, + "grad_norm": 2.8037819862365723, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9037752151489258, + "num_tokens": 858533579.0, + "step": 22119 + }, + { + "epoch": 2.813891362422084, + "grad_norm": 2.7903692722320557, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.914595365524292, + "num_tokens": 858572623.0, + "step": 22120 + }, + { + "epoch": 2.8140185727006743, + "grad_norm": 2.6798529624938965, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9113867282867432, + "num_tokens": 858612530.0, + "step": 22121 + }, + { + "epoch": 2.814145782979265, + "grad_norm": 2.986675262451172, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9040486216545105, + "num_tokens": 858648440.0, + "step": 22122 + }, + { + "epoch": 2.8142729932578554, + "grad_norm": 2.926841974258423, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.8991455435752869, + "num_tokens": 858689505.0, + "step": 22123 + }, + { + "epoch": 2.8144002035364455, + "grad_norm": 3.0188655853271484, + "learning_rate": 1e-06, + "loss": 0.2511, + "mean_token_accuracy": 0.9072609543800354, + "num_tokens": 858724267.0, + "step": 22124 + }, + { + "epoch": 2.8145274138150365, + "grad_norm": 2.789337158203125, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9074710607528687, + "num_tokens": 858761989.0, + "step": 22125 + }, + { + "epoch": 2.8146546240936265, + "grad_norm": 2.706156015396118, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9099594950675964, + "num_tokens": 858799861.0, + "step": 22126 + }, + { + "epoch": 2.8147818343722175, + "grad_norm": 2.802701473236084, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9038118720054626, + "num_tokens": 858842909.0, + "step": 22127 + }, + { + "epoch": 2.8149090446508076, + "grad_norm": 2.688986301422119, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9087019562721252, + "num_tokens": 858883068.0, + "step": 22128 + }, + { + "epoch": 2.815036254929398, + "grad_norm": 2.9143130779266357, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9085810780525208, + "num_tokens": 858923566.0, + "step": 22129 + }, + { + "epoch": 2.8151634652079887, + "grad_norm": 2.892512321472168, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9042561650276184, + "num_tokens": 858960163.0, + "step": 22130 + }, + { + "epoch": 2.815290675486579, + "grad_norm": 3.224180221557617, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8952728509902954, + "num_tokens": 858994092.0, + "step": 22131 + }, + { + "epoch": 2.8154178857651697, + "grad_norm": 2.9795260429382324, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.897925615310669, + "num_tokens": 859035654.0, + "step": 22132 + }, + { + "epoch": 2.8155450960437602, + "grad_norm": 3.05330491065979, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.907116174697876, + "num_tokens": 859069357.0, + "step": 22133 + }, + { + "epoch": 2.8156723063223508, + "grad_norm": 2.9546561241149902, + "learning_rate": 1e-06, + "loss": 0.2819, + "mean_token_accuracy": 0.8978734016418457, + "num_tokens": 859114132.0, + "step": 22134 + }, + { + "epoch": 2.8157995166009413, + "grad_norm": 2.7702226638793945, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9083172082901001, + "num_tokens": 859152425.0, + "step": 22135 + }, + { + "epoch": 2.815926726879532, + "grad_norm": 3.000601291656494, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9097201824188232, + "num_tokens": 859186972.0, + "step": 22136 + }, + { + "epoch": 2.8160539371581224, + "grad_norm": 3.1465418338775635, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.9005576968193054, + "num_tokens": 859219760.0, + "step": 22137 + }, + { + "epoch": 2.816181147436713, + "grad_norm": 2.75349760055542, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8945008516311646, + "num_tokens": 859264165.0, + "step": 22138 + }, + { + "epoch": 2.8163083577153034, + "grad_norm": 2.609286308288574, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9096320867538452, + "num_tokens": 859305047.0, + "step": 22139 + }, + { + "epoch": 2.816435567993894, + "grad_norm": 2.928252696990967, + "learning_rate": 1e-06, + "loss": 0.272, + "mean_token_accuracy": 0.9011356234550476, + "num_tokens": 859346310.0, + "step": 22140 + }, + { + "epoch": 2.8165627782724845, + "grad_norm": 2.95622181892395, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9119073152542114, + "num_tokens": 859385575.0, + "step": 22141 + }, + { + "epoch": 2.816689988551075, + "grad_norm": 2.8759028911590576, + "learning_rate": 1e-06, + "loss": 0.2445, + "mean_token_accuracy": 0.908764123916626, + "num_tokens": 859420520.0, + "step": 22142 + }, + { + "epoch": 2.8168171988296655, + "grad_norm": 2.887477397918701, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9128590822219849, + "num_tokens": 859456184.0, + "step": 22143 + }, + { + "epoch": 2.816944409108256, + "grad_norm": 2.5634377002716064, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9080760478973389, + "num_tokens": 859500893.0, + "step": 22144 + }, + { + "epoch": 2.8170716193868466, + "grad_norm": 2.7981364727020264, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9101929664611816, + "num_tokens": 859538592.0, + "step": 22145 + }, + { + "epoch": 2.817198829665437, + "grad_norm": 2.7255446910858154, + "learning_rate": 1e-06, + "loss": 0.268, + "mean_token_accuracy": 0.9010649919509888, + "num_tokens": 859580585.0, + "step": 22146 + }, + { + "epoch": 2.8173260399440276, + "grad_norm": 2.9489972591400146, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9033602476119995, + "num_tokens": 859617573.0, + "step": 22147 + }, + { + "epoch": 2.817453250222618, + "grad_norm": 2.713231325149536, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9128332138061523, + "num_tokens": 859661089.0, + "step": 22148 + }, + { + "epoch": 2.8175804605012083, + "grad_norm": 3.1729297637939453, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8931124210357666, + "num_tokens": 859698797.0, + "step": 22149 + }, + { + "epoch": 2.8177076707797992, + "grad_norm": 2.66549015045166, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9067921042442322, + "num_tokens": 859740359.0, + "step": 22150 + }, + { + "epoch": 2.8178348810583893, + "grad_norm": 2.6679422855377197, + "learning_rate": 1e-06, + "loss": 0.2392, + "mean_token_accuracy": 0.911365807056427, + "num_tokens": 859780453.0, + "step": 22151 + }, + { + "epoch": 2.8179620913369803, + "grad_norm": 2.573559522628784, + "learning_rate": 1e-06, + "loss": 0.2174, + "mean_token_accuracy": 0.9199194312095642, + "num_tokens": 859819793.0, + "step": 22152 + }, + { + "epoch": 2.8180893016155704, + "grad_norm": 2.880934715270996, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9050464034080505, + "num_tokens": 859859620.0, + "step": 22153 + }, + { + "epoch": 2.818216511894161, + "grad_norm": 3.064030170440674, + "learning_rate": 1e-06, + "loss": 0.31, + "mean_token_accuracy": 0.8879127502441406, + "num_tokens": 859899521.0, + "step": 22154 + }, + { + "epoch": 2.8183437221727514, + "grad_norm": 2.7462682723999023, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9043489694595337, + "num_tokens": 859941776.0, + "step": 22155 + }, + { + "epoch": 2.818470932451342, + "grad_norm": 2.819734811782837, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9072719216346741, + "num_tokens": 859981587.0, + "step": 22156 + }, + { + "epoch": 2.8185981427299325, + "grad_norm": 2.771339178085327, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9050697684288025, + "num_tokens": 860022040.0, + "step": 22157 + }, + { + "epoch": 2.818725353008523, + "grad_norm": 2.7103211879730225, + "learning_rate": 1e-06, + "loss": 0.2242, + "mean_token_accuracy": 0.9168068766593933, + "num_tokens": 860060446.0, + "step": 22158 + }, + { + "epoch": 2.8188525632871135, + "grad_norm": 2.9244070053100586, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9069161415100098, + "num_tokens": 860099292.0, + "step": 22159 + }, + { + "epoch": 2.818979773565704, + "grad_norm": 2.882974147796631, + "learning_rate": 1e-06, + "loss": 0.2635, + "mean_token_accuracy": 0.90168696641922, + "num_tokens": 860135937.0, + "step": 22160 + }, + { + "epoch": 2.8191069838442946, + "grad_norm": 3.0003914833068848, + "learning_rate": 1e-06, + "loss": 0.3079, + "mean_token_accuracy": 0.8920851349830627, + "num_tokens": 860178467.0, + "step": 22161 + }, + { + "epoch": 2.819234194122885, + "grad_norm": 2.8384358882904053, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8970372676849365, + "num_tokens": 860220123.0, + "step": 22162 + }, + { + "epoch": 2.8193614044014756, + "grad_norm": 2.887763261795044, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9040277600288391, + "num_tokens": 860259378.0, + "step": 22163 + }, + { + "epoch": 2.819488614680066, + "grad_norm": 2.694841146469116, + "learning_rate": 1e-06, + "loss": 0.2256, + "mean_token_accuracy": 0.9145427346229553, + "num_tokens": 860297197.0, + "step": 22164 + }, + { + "epoch": 2.8196158249586567, + "grad_norm": 2.8753621578216553, + "learning_rate": 1e-06, + "loss": 0.2327, + "mean_token_accuracy": 0.9125543832778931, + "num_tokens": 860331913.0, + "step": 22165 + }, + { + "epoch": 2.8197430352372472, + "grad_norm": 3.266289710998535, + "learning_rate": 1e-06, + "loss": 0.3052, + "mean_token_accuracy": 0.8882691860198975, + "num_tokens": 860367784.0, + "step": 22166 + }, + { + "epoch": 2.8198702455158378, + "grad_norm": 2.807772159576416, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.909034013748169, + "num_tokens": 860406258.0, + "step": 22167 + }, + { + "epoch": 2.8199974557944283, + "grad_norm": 3.234558343887329, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8967247009277344, + "num_tokens": 860442428.0, + "step": 22168 + }, + { + "epoch": 2.820124666073019, + "grad_norm": 2.88946533203125, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9033706188201904, + "num_tokens": 860481700.0, + "step": 22169 + }, + { + "epoch": 2.8202518763516093, + "grad_norm": 2.8077733516693115, + "learning_rate": 1e-06, + "loss": 0.2739, + "mean_token_accuracy": 0.8988593816757202, + "num_tokens": 860523996.0, + "step": 22170 + }, + { + "epoch": 2.8203790866302, + "grad_norm": 3.0311100482940674, + "learning_rate": 1e-06, + "loss": 0.3032, + "mean_token_accuracy": 0.8903298377990723, + "num_tokens": 860562762.0, + "step": 22171 + }, + { + "epoch": 2.82050629690879, + "grad_norm": 2.8946945667266846, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.90664142370224, + "num_tokens": 860598431.0, + "step": 22172 + }, + { + "epoch": 2.820633507187381, + "grad_norm": 3.0280449390411377, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8934112787246704, + "num_tokens": 860639696.0, + "step": 22173 + }, + { + "epoch": 2.820760717465971, + "grad_norm": 2.724724531173706, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9071753025054932, + "num_tokens": 860682759.0, + "step": 22174 + }, + { + "epoch": 2.820887927744562, + "grad_norm": 3.252469539642334, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9063069820404053, + "num_tokens": 860710862.0, + "step": 22175 + }, + { + "epoch": 2.821015138023152, + "grad_norm": 2.870180606842041, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9032339453697205, + "num_tokens": 860752538.0, + "step": 22176 + }, + { + "epoch": 2.821142348301743, + "grad_norm": 2.847721815109253, + "learning_rate": 1e-06, + "loss": 0.2927, + "mean_token_accuracy": 0.8962962627410889, + "num_tokens": 860794692.0, + "step": 22177 + }, + { + "epoch": 2.821269558580333, + "grad_norm": 2.8595736026763916, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9045025110244751, + "num_tokens": 860834364.0, + "step": 22178 + }, + { + "epoch": 2.8213967688589237, + "grad_norm": 2.628338575363159, + "learning_rate": 1e-06, + "loss": 0.2313, + "mean_token_accuracy": 0.9094521999359131, + "num_tokens": 860874866.0, + "step": 22179 + }, + { + "epoch": 2.821523979137514, + "grad_norm": 2.901512622833252, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9093308448791504, + "num_tokens": 860910621.0, + "step": 22180 + }, + { + "epoch": 2.8216511894161047, + "grad_norm": 3.002248764038086, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8933879137039185, + "num_tokens": 860952125.0, + "step": 22181 + }, + { + "epoch": 2.8217783996946952, + "grad_norm": 2.8284568786621094, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9090789556503296, + "num_tokens": 860989211.0, + "step": 22182 + }, + { + "epoch": 2.8219056099732858, + "grad_norm": 3.091646909713745, + "learning_rate": 1e-06, + "loss": 0.284, + "mean_token_accuracy": 0.8984912633895874, + "num_tokens": 861026821.0, + "step": 22183 + }, + { + "epoch": 2.8220328202518763, + "grad_norm": 2.9818778038024902, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.905218243598938, + "num_tokens": 861061208.0, + "step": 22184 + }, + { + "epoch": 2.822160030530467, + "grad_norm": 3.1094164848327637, + "learning_rate": 1e-06, + "loss": 0.2876, + "mean_token_accuracy": 0.8960564732551575, + "num_tokens": 861097912.0, + "step": 22185 + }, + { + "epoch": 2.8222872408090574, + "grad_norm": 3.117267370223999, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8949562311172485, + "num_tokens": 861133184.0, + "step": 22186 + }, + { + "epoch": 2.822414451087648, + "grad_norm": 2.946603298187256, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9105597734451294, + "num_tokens": 861168191.0, + "step": 22187 + }, + { + "epoch": 2.8225416613662384, + "grad_norm": 3.133120536804199, + "learning_rate": 1e-06, + "loss": 0.2559, + "mean_token_accuracy": 0.9062986373901367, + "num_tokens": 861203124.0, + "step": 22188 + }, + { + "epoch": 2.822668871644829, + "grad_norm": 2.8319520950317383, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9084094762802124, + "num_tokens": 861240952.0, + "step": 22189 + }, + { + "epoch": 2.8227960819234195, + "grad_norm": 2.73500657081604, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9066464900970459, + "num_tokens": 861280998.0, + "step": 22190 + }, + { + "epoch": 2.82292329220201, + "grad_norm": 2.869004726409912, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9068031907081604, + "num_tokens": 861321773.0, + "step": 22191 + }, + { + "epoch": 2.8230505024806005, + "grad_norm": 2.8976082801818848, + "learning_rate": 1e-06, + "loss": 0.2793, + "mean_token_accuracy": 0.8976932764053345, + "num_tokens": 861360603.0, + "step": 22192 + }, + { + "epoch": 2.823177712759191, + "grad_norm": 2.9690439701080322, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9055227041244507, + "num_tokens": 861397309.0, + "step": 22193 + }, + { + "epoch": 2.8233049230377816, + "grad_norm": 2.6289756298065186, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9030225276947021, + "num_tokens": 861436619.0, + "step": 22194 + }, + { + "epoch": 2.823432133316372, + "grad_norm": 2.6115589141845703, + "learning_rate": 1e-06, + "loss": 0.2277, + "mean_token_accuracy": 0.9125217199325562, + "num_tokens": 861478416.0, + "step": 22195 + }, + { + "epoch": 2.8235593435949626, + "grad_norm": 2.785571813583374, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.904890775680542, + "num_tokens": 861518857.0, + "step": 22196 + }, + { + "epoch": 2.8236865538735527, + "grad_norm": 2.7173328399658203, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.900990903377533, + "num_tokens": 861561547.0, + "step": 22197 + }, + { + "epoch": 2.8238137641521437, + "grad_norm": 2.773563861846924, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9135036468505859, + "num_tokens": 861600559.0, + "step": 22198 + }, + { + "epoch": 2.8239409744307338, + "grad_norm": 2.7929790019989014, + "learning_rate": 1e-06, + "loss": 0.2334, + "mean_token_accuracy": 0.9128870964050293, + "num_tokens": 861638690.0, + "step": 22199 + }, + { + "epoch": 2.8240681847093247, + "grad_norm": 2.7734737396240234, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9101744890213013, + "num_tokens": 861676897.0, + "step": 22200 + }, + { + "epoch": 2.824195394987915, + "grad_norm": 3.0866432189941406, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.8978870511054993, + "num_tokens": 861714672.0, + "step": 22201 + }, + { + "epoch": 2.824322605266506, + "grad_norm": 3.0643112659454346, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9035325646400452, + "num_tokens": 861749901.0, + "step": 22202 + }, + { + "epoch": 2.824449815545096, + "grad_norm": 2.98022723197937, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8966730833053589, + "num_tokens": 861789190.0, + "step": 22203 + }, + { + "epoch": 2.8245770258236864, + "grad_norm": 2.960130453109741, + "learning_rate": 1e-06, + "loss": 0.2884, + "mean_token_accuracy": 0.8929172158241272, + "num_tokens": 861830034.0, + "step": 22204 + }, + { + "epoch": 2.824704236102277, + "grad_norm": 3.028690814971924, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9015129804611206, + "num_tokens": 861869110.0, + "step": 22205 + }, + { + "epoch": 2.8248314463808675, + "grad_norm": 3.1150729656219482, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9015830755233765, + "num_tokens": 861901019.0, + "step": 22206 + }, + { + "epoch": 2.824958656659458, + "grad_norm": 2.9068522453308105, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9043922424316406, + "num_tokens": 861936868.0, + "step": 22207 + }, + { + "epoch": 2.8250858669380485, + "grad_norm": 3.4606826305389404, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.9016724228858948, + "num_tokens": 861967474.0, + "step": 22208 + }, + { + "epoch": 2.825213077216639, + "grad_norm": 3.09822154045105, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9039583206176758, + "num_tokens": 862000158.0, + "step": 22209 + }, + { + "epoch": 2.8253402874952296, + "grad_norm": 3.18827486038208, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.907384991645813, + "num_tokens": 862031725.0, + "step": 22210 + }, + { + "epoch": 2.82546749777382, + "grad_norm": 2.911567449569702, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.8993275761604309, + "num_tokens": 862071541.0, + "step": 22211 + }, + { + "epoch": 2.8255947080524106, + "grad_norm": 3.009202241897583, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.89772629737854, + "num_tokens": 862109723.0, + "step": 22212 + }, + { + "epoch": 2.825721918331001, + "grad_norm": 3.3554818630218506, + "learning_rate": 1e-06, + "loss": 0.2954, + "mean_token_accuracy": 0.8946651220321655, + "num_tokens": 862141486.0, + "step": 22213 + }, + { + "epoch": 2.8258491286095917, + "grad_norm": 2.9870195388793945, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9052700400352478, + "num_tokens": 862177956.0, + "step": 22214 + }, + { + "epoch": 2.8259763388881822, + "grad_norm": 3.070225715637207, + "learning_rate": 1e-06, + "loss": 0.2759, + "mean_token_accuracy": 0.8965758681297302, + "num_tokens": 862215139.0, + "step": 22215 + }, + { + "epoch": 2.8261035491667728, + "grad_norm": 3.1466474533081055, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8975870013237, + "num_tokens": 862249924.0, + "step": 22216 + }, + { + "epoch": 2.8262307594453633, + "grad_norm": 3.030078172683716, + "learning_rate": 1e-06, + "loss": 0.2792, + "mean_token_accuracy": 0.8948623538017273, + "num_tokens": 862286291.0, + "step": 22217 + }, + { + "epoch": 2.826357969723954, + "grad_norm": 2.8822805881500244, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.9069065451622009, + "num_tokens": 862323344.0, + "step": 22218 + }, + { + "epoch": 2.8264851800025443, + "grad_norm": 3.0348148345947266, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8950576782226562, + "num_tokens": 862361543.0, + "step": 22219 + }, + { + "epoch": 2.826612390281135, + "grad_norm": 2.89723801612854, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9012317657470703, + "num_tokens": 862398717.0, + "step": 22220 + }, + { + "epoch": 2.8267396005597254, + "grad_norm": 3.137416362762451, + "learning_rate": 1e-06, + "loss": 0.2899, + "mean_token_accuracy": 0.8951708674430847, + "num_tokens": 862433644.0, + "step": 22221 + }, + { + "epoch": 2.8268668108383155, + "grad_norm": 2.859285831451416, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8988534808158875, + "num_tokens": 862472811.0, + "step": 22222 + }, + { + "epoch": 2.8269940211169065, + "grad_norm": 2.863759756088257, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9050697088241577, + "num_tokens": 862511847.0, + "step": 22223 + }, + { + "epoch": 2.8271212313954965, + "grad_norm": 2.7130308151245117, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.9067915081977844, + "num_tokens": 862556206.0, + "step": 22224 + }, + { + "epoch": 2.8272484416740875, + "grad_norm": 3.005190372467041, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9012875556945801, + "num_tokens": 862594994.0, + "step": 22225 + }, + { + "epoch": 2.8273756519526776, + "grad_norm": 2.854910373687744, + "learning_rate": 1e-06, + "loss": 0.2779, + "mean_token_accuracy": 0.9006921648979187, + "num_tokens": 862633635.0, + "step": 22226 + }, + { + "epoch": 2.827502862231268, + "grad_norm": 3.091362714767456, + "learning_rate": 1e-06, + "loss": 0.2955, + "mean_token_accuracy": 0.8893976211547852, + "num_tokens": 862670107.0, + "step": 22227 + }, + { + "epoch": 2.8276300725098586, + "grad_norm": 2.5189707279205322, + "learning_rate": 1e-06, + "loss": 0.2207, + "mean_token_accuracy": 0.917263925075531, + "num_tokens": 862713866.0, + "step": 22228 + }, + { + "epoch": 2.827757282788449, + "grad_norm": 3.164309501647949, + "learning_rate": 1e-06, + "loss": 0.3067, + "mean_token_accuracy": 0.8863318562507629, + "num_tokens": 862748885.0, + "step": 22229 + }, + { + "epoch": 2.8278844930670397, + "grad_norm": 2.9503254890441895, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8945300579071045, + "num_tokens": 862786330.0, + "step": 22230 + }, + { + "epoch": 2.8280117033456302, + "grad_norm": 2.823354959487915, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.902553915977478, + "num_tokens": 862829130.0, + "step": 22231 + }, + { + "epoch": 2.8281389136242208, + "grad_norm": 3.0321149826049805, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9042184352874756, + "num_tokens": 862861902.0, + "step": 22232 + }, + { + "epoch": 2.8282661239028113, + "grad_norm": 2.9406678676605225, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9008464217185974, + "num_tokens": 862901409.0, + "step": 22233 + }, + { + "epoch": 2.828393334181402, + "grad_norm": 2.9645628929138184, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.8999836444854736, + "num_tokens": 862942042.0, + "step": 22234 + }, + { + "epoch": 2.8285205444599923, + "grad_norm": 2.6667020320892334, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.8997809886932373, + "num_tokens": 862987934.0, + "step": 22235 + }, + { + "epoch": 2.828647754738583, + "grad_norm": 3.091688394546509, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.903136134147644, + "num_tokens": 863023282.0, + "step": 22236 + }, + { + "epoch": 2.8287749650171734, + "grad_norm": 2.793281078338623, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9063477516174316, + "num_tokens": 863062166.0, + "step": 22237 + }, + { + "epoch": 2.828902175295764, + "grad_norm": 2.980910062789917, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8934680223464966, + "num_tokens": 863100544.0, + "step": 22238 + }, + { + "epoch": 2.8290293855743545, + "grad_norm": 3.0120766162872314, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.9019463062286377, + "num_tokens": 863139451.0, + "step": 22239 + }, + { + "epoch": 2.829156595852945, + "grad_norm": 3.130981683731079, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.910451352596283, + "num_tokens": 863172730.0, + "step": 22240 + }, + { + "epoch": 2.8292838061315355, + "grad_norm": 3.071352005004883, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.9090948104858398, + "num_tokens": 863206727.0, + "step": 22241 + }, + { + "epoch": 2.829411016410126, + "grad_norm": 2.8241095542907715, + "learning_rate": 1e-06, + "loss": 0.2625, + "mean_token_accuracy": 0.9019174575805664, + "num_tokens": 863247061.0, + "step": 22242 + }, + { + "epoch": 2.8295382266887166, + "grad_norm": 2.8701393604278564, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8976321220397949, + "num_tokens": 863288193.0, + "step": 22243 + }, + { + "epoch": 2.829665436967307, + "grad_norm": 3.0546443462371826, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.8978204727172852, + "num_tokens": 863323750.0, + "step": 22244 + }, + { + "epoch": 2.8297926472458976, + "grad_norm": 3.5394601821899414, + "learning_rate": 1e-06, + "loss": 0.3409, + "mean_token_accuracy": 0.8774479031562805, + "num_tokens": 863357503.0, + "step": 22245 + }, + { + "epoch": 2.829919857524488, + "grad_norm": 2.730938673019409, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9022359848022461, + "num_tokens": 863400268.0, + "step": 22246 + }, + { + "epoch": 2.8300470678030782, + "grad_norm": 2.8232321739196777, + "learning_rate": 1e-06, + "loss": 0.2302, + "mean_token_accuracy": 0.9113638401031494, + "num_tokens": 863440090.0, + "step": 22247 + }, + { + "epoch": 2.830174278081669, + "grad_norm": 3.1001648902893066, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8993622064590454, + "num_tokens": 863472437.0, + "step": 22248 + }, + { + "epoch": 2.8303014883602593, + "grad_norm": 2.6935081481933594, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9038102626800537, + "num_tokens": 863512936.0, + "step": 22249 + }, + { + "epoch": 2.8304286986388503, + "grad_norm": 2.8117783069610596, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.897272527217865, + "num_tokens": 863551680.0, + "step": 22250 + }, + { + "epoch": 2.8305559089174404, + "grad_norm": 2.787508726119995, + "learning_rate": 1e-06, + "loss": 0.2408, + "mean_token_accuracy": 0.9109412431716919, + "num_tokens": 863587739.0, + "step": 22251 + }, + { + "epoch": 2.830683119196031, + "grad_norm": 2.949873447418213, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9098435640335083, + "num_tokens": 863622054.0, + "step": 22252 + }, + { + "epoch": 2.8308103294746214, + "grad_norm": 2.920173406600952, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9019781351089478, + "num_tokens": 863659414.0, + "step": 22253 + }, + { + "epoch": 2.830937539753212, + "grad_norm": 3.399414539337158, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.8957861661911011, + "num_tokens": 863690501.0, + "step": 22254 + }, + { + "epoch": 2.8310647500318025, + "grad_norm": 2.946932077407837, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9000864624977112, + "num_tokens": 863725381.0, + "step": 22255 + }, + { + "epoch": 2.831191960310393, + "grad_norm": 2.6021182537078857, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9025908708572388, + "num_tokens": 863769116.0, + "step": 22256 + }, + { + "epoch": 2.8313191705889835, + "grad_norm": 2.8254339694976807, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.8985015749931335, + "num_tokens": 863810525.0, + "step": 22257 + }, + { + "epoch": 2.831446380867574, + "grad_norm": 2.753911018371582, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.9062386751174927, + "num_tokens": 863852562.0, + "step": 22258 + }, + { + "epoch": 2.8315735911461646, + "grad_norm": 3.0493111610412598, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8937708735466003, + "num_tokens": 863894589.0, + "step": 22259 + }, + { + "epoch": 2.831700801424755, + "grad_norm": 2.7895777225494385, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9007153511047363, + "num_tokens": 863937788.0, + "step": 22260 + }, + { + "epoch": 2.8318280117033456, + "grad_norm": 2.804654359817505, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9066767692565918, + "num_tokens": 863977284.0, + "step": 22261 + }, + { + "epoch": 2.831955221981936, + "grad_norm": 2.637127161026001, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9107231497764587, + "num_tokens": 864017745.0, + "step": 22262 + }, + { + "epoch": 2.8320824322605267, + "grad_norm": 3.039592742919922, + "learning_rate": 1e-06, + "loss": 0.2347, + "mean_token_accuracy": 0.9116123914718628, + "num_tokens": 864050202.0, + "step": 22263 + }, + { + "epoch": 2.8322096425391172, + "grad_norm": 2.880896806716919, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.9068928360939026, + "num_tokens": 864086192.0, + "step": 22264 + }, + { + "epoch": 2.8323368528177078, + "grad_norm": 2.930604934692383, + "learning_rate": 1e-06, + "loss": 0.3006, + "mean_token_accuracy": 0.8917840123176575, + "num_tokens": 864129347.0, + "step": 22265 + }, + { + "epoch": 2.8324640630962983, + "grad_norm": 2.8347342014312744, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.8952882289886475, + "num_tokens": 864169018.0, + "step": 22266 + }, + { + "epoch": 2.832591273374889, + "grad_norm": 3.007711887359619, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9032056331634521, + "num_tokens": 864206948.0, + "step": 22267 + }, + { + "epoch": 2.8327184836534793, + "grad_norm": 2.746248960494995, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9006969928741455, + "num_tokens": 864248834.0, + "step": 22268 + }, + { + "epoch": 2.83284569393207, + "grad_norm": 2.8667731285095215, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.9025683999061584, + "num_tokens": 864289975.0, + "step": 22269 + }, + { + "epoch": 2.83297290421066, + "grad_norm": 2.857825517654419, + "learning_rate": 1e-06, + "loss": 0.2351, + "mean_token_accuracy": 0.9126619696617126, + "num_tokens": 864328085.0, + "step": 22270 + }, + { + "epoch": 2.833100114489251, + "grad_norm": 3.1138741970062256, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.8939244747161865, + "num_tokens": 864362543.0, + "step": 22271 + }, + { + "epoch": 2.833227324767841, + "grad_norm": 2.926393985748291, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9091563820838928, + "num_tokens": 864398427.0, + "step": 22272 + }, + { + "epoch": 2.833354535046432, + "grad_norm": 2.797205686569214, + "learning_rate": 1e-06, + "loss": 0.216, + "mean_token_accuracy": 0.9206719398498535, + "num_tokens": 864433498.0, + "step": 22273 + }, + { + "epoch": 2.833481745325022, + "grad_norm": 3.240886688232422, + "learning_rate": 1e-06, + "loss": 0.247, + "mean_token_accuracy": 0.9082887768745422, + "num_tokens": 864466404.0, + "step": 22274 + }, + { + "epoch": 2.833608955603613, + "grad_norm": 2.8600027561187744, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9010267853736877, + "num_tokens": 864505495.0, + "step": 22275 + }, + { + "epoch": 2.833736165882203, + "grad_norm": 2.6506054401397705, + "learning_rate": 1e-06, + "loss": 0.2256, + "mean_token_accuracy": 0.9177910685539246, + "num_tokens": 864545165.0, + "step": 22276 + }, + { + "epoch": 2.8338633761607936, + "grad_norm": 2.669053554534912, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9137235283851624, + "num_tokens": 864583357.0, + "step": 22277 + }, + { + "epoch": 2.833990586439384, + "grad_norm": 2.739032030105591, + "learning_rate": 1e-06, + "loss": 0.2646, + "mean_token_accuracy": 0.9013373255729675, + "num_tokens": 864624089.0, + "step": 22278 + }, + { + "epoch": 2.8341177967179747, + "grad_norm": 2.7648961544036865, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9116616249084473, + "num_tokens": 864663372.0, + "step": 22279 + }, + { + "epoch": 2.8342450069965652, + "grad_norm": 2.7263617515563965, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9000743627548218, + "num_tokens": 864706218.0, + "step": 22280 + }, + { + "epoch": 2.8343722172751558, + "grad_norm": 2.773516893386841, + "learning_rate": 1e-06, + "loss": 0.2181, + "mean_token_accuracy": 0.9171743392944336, + "num_tokens": 864741344.0, + "step": 22281 + }, + { + "epoch": 2.8344994275537463, + "grad_norm": 2.947753429412842, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8964877724647522, + "num_tokens": 864779977.0, + "step": 22282 + }, + { + "epoch": 2.834626637832337, + "grad_norm": 2.9600460529327393, + "learning_rate": 1e-06, + "loss": 0.2374, + "mean_token_accuracy": 0.9113900065422058, + "num_tokens": 864814302.0, + "step": 22283 + }, + { + "epoch": 2.8347538481109273, + "grad_norm": 2.690661668777466, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9019149541854858, + "num_tokens": 864861624.0, + "step": 22284 + }, + { + "epoch": 2.834881058389518, + "grad_norm": 3.0173585414886475, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.899969220161438, + "num_tokens": 864899218.0, + "step": 22285 + }, + { + "epoch": 2.8350082686681084, + "grad_norm": 2.913808822631836, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9061269760131836, + "num_tokens": 864936754.0, + "step": 22286 + }, + { + "epoch": 2.835135478946699, + "grad_norm": 2.7912652492523193, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9039251804351807, + "num_tokens": 864979417.0, + "step": 22287 + }, + { + "epoch": 2.8352626892252895, + "grad_norm": 2.8509693145751953, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9008245468139648, + "num_tokens": 865021755.0, + "step": 22288 + }, + { + "epoch": 2.83538989950388, + "grad_norm": 3.0768280029296875, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9029637575149536, + "num_tokens": 865056253.0, + "step": 22289 + }, + { + "epoch": 2.8355171097824705, + "grad_norm": 2.814093828201294, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8992038369178772, + "num_tokens": 865100696.0, + "step": 22290 + }, + { + "epoch": 2.835644320061061, + "grad_norm": 3.1204447746276855, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.9022641181945801, + "num_tokens": 865133849.0, + "step": 22291 + }, + { + "epoch": 2.8357715303396516, + "grad_norm": 2.7088630199432373, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9109415411949158, + "num_tokens": 865177172.0, + "step": 22292 + }, + { + "epoch": 2.835898740618242, + "grad_norm": 3.0961461067199707, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.903429388999939, + "num_tokens": 865211501.0, + "step": 22293 + }, + { + "epoch": 2.8360259508968326, + "grad_norm": 2.7270963191986084, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9051579236984253, + "num_tokens": 865252278.0, + "step": 22294 + }, + { + "epoch": 2.8361531611754227, + "grad_norm": 3.346148729324341, + "learning_rate": 1e-06, + "loss": 0.2999, + "mean_token_accuracy": 0.8869023323059082, + "num_tokens": 865288107.0, + "step": 22295 + }, + { + "epoch": 2.8362803714540137, + "grad_norm": 2.9935081005096436, + "learning_rate": 1e-06, + "loss": 0.2936, + "mean_token_accuracy": 0.8919697403907776, + "num_tokens": 865331165.0, + "step": 22296 + }, + { + "epoch": 2.8364075817326038, + "grad_norm": 2.7158963680267334, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9015967845916748, + "num_tokens": 865370316.0, + "step": 22297 + }, + { + "epoch": 2.8365347920111947, + "grad_norm": 2.9957635402679443, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9072726368904114, + "num_tokens": 865407839.0, + "step": 22298 + }, + { + "epoch": 2.836662002289785, + "grad_norm": 2.869497776031494, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9028049111366272, + "num_tokens": 865449422.0, + "step": 22299 + }, + { + "epoch": 2.836789212568376, + "grad_norm": 2.8156309127807617, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9067867994308472, + "num_tokens": 865489439.0, + "step": 22300 + }, + { + "epoch": 2.836916422846966, + "grad_norm": 3.0338945388793945, + "learning_rate": 1e-06, + "loss": 0.3001, + "mean_token_accuracy": 0.8897374272346497, + "num_tokens": 865530050.0, + "step": 22301 + }, + { + "epoch": 2.8370436331255564, + "grad_norm": 2.812330961227417, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9062326550483704, + "num_tokens": 865567414.0, + "step": 22302 + }, + { + "epoch": 2.837170843404147, + "grad_norm": 2.8670525550842285, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9031837582588196, + "num_tokens": 865605123.0, + "step": 22303 + }, + { + "epoch": 2.8372980536827375, + "grad_norm": 3.1542885303497314, + "learning_rate": 1e-06, + "loss": 0.2541, + "mean_token_accuracy": 0.9065653085708618, + "num_tokens": 865647895.0, + "step": 22304 + }, + { + "epoch": 2.837425263961328, + "grad_norm": 2.8415615558624268, + "learning_rate": 1e-06, + "loss": 0.2348, + "mean_token_accuracy": 0.9119901657104492, + "num_tokens": 865683810.0, + "step": 22305 + }, + { + "epoch": 2.8375524742399185, + "grad_norm": 3.0531821250915527, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8908248543739319, + "num_tokens": 865724736.0, + "step": 22306 + }, + { + "epoch": 2.837679684518509, + "grad_norm": 2.9163436889648438, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.9125966429710388, + "num_tokens": 865758257.0, + "step": 22307 + }, + { + "epoch": 2.8378068947970996, + "grad_norm": 2.869230270385742, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.904720664024353, + "num_tokens": 865798191.0, + "step": 22308 + }, + { + "epoch": 2.83793410507569, + "grad_norm": 2.996951103210449, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9074051380157471, + "num_tokens": 865836858.0, + "step": 22309 + }, + { + "epoch": 2.8380613153542806, + "grad_norm": 2.803748607635498, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9057007431983948, + "num_tokens": 865873976.0, + "step": 22310 + }, + { + "epoch": 2.838188525632871, + "grad_norm": 2.8134565353393555, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9009011387825012, + "num_tokens": 865916073.0, + "step": 22311 + }, + { + "epoch": 2.8383157359114617, + "grad_norm": 2.4815428256988525, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9054283499717712, + "num_tokens": 865964917.0, + "step": 22312 + }, + { + "epoch": 2.838442946190052, + "grad_norm": 2.5544183254241943, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9077338576316833, + "num_tokens": 866010556.0, + "step": 22313 + }, + { + "epoch": 2.8385701564686427, + "grad_norm": 2.6102540493011475, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9109642505645752, + "num_tokens": 866050249.0, + "step": 22314 + }, + { + "epoch": 2.8386973667472333, + "grad_norm": 2.858614921569824, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9067988991737366, + "num_tokens": 866090251.0, + "step": 22315 + }, + { + "epoch": 2.838824577025824, + "grad_norm": 2.661818265914917, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9065858125686646, + "num_tokens": 866135811.0, + "step": 22316 + }, + { + "epoch": 2.8389517873044143, + "grad_norm": 3.0070502758026123, + "learning_rate": 1e-06, + "loss": 0.2881, + "mean_token_accuracy": 0.8968702554702759, + "num_tokens": 866175614.0, + "step": 22317 + }, + { + "epoch": 2.839078997583005, + "grad_norm": 2.7616474628448486, + "learning_rate": 1e-06, + "loss": 0.2441, + "mean_token_accuracy": 0.9089914560317993, + "num_tokens": 866215451.0, + "step": 22318 + }, + { + "epoch": 2.8392062078615954, + "grad_norm": 2.830181837081909, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9008510708808899, + "num_tokens": 866259948.0, + "step": 22319 + }, + { + "epoch": 2.8393334181401855, + "grad_norm": 2.907623291015625, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9096217155456543, + "num_tokens": 866295843.0, + "step": 22320 + }, + { + "epoch": 2.8394606284187764, + "grad_norm": 3.0254387855529785, + "learning_rate": 1e-06, + "loss": 0.2321, + "mean_token_accuracy": 0.9147506356239319, + "num_tokens": 866329017.0, + "step": 22321 + }, + { + "epoch": 2.8395878386973665, + "grad_norm": 2.9485650062561035, + "learning_rate": 1e-06, + "loss": 0.2592, + "mean_token_accuracy": 0.9054558873176575, + "num_tokens": 866366593.0, + "step": 22322 + }, + { + "epoch": 2.8397150489759575, + "grad_norm": 2.7840545177459717, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.8999373912811279, + "num_tokens": 866406996.0, + "step": 22323 + }, + { + "epoch": 2.8398422592545476, + "grad_norm": 2.8866219520568848, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8949372172355652, + "num_tokens": 866450111.0, + "step": 22324 + }, + { + "epoch": 2.839969469533138, + "grad_norm": 2.8119115829467773, + "learning_rate": 1e-06, + "loss": 0.2871, + "mean_token_accuracy": 0.8958567380905151, + "num_tokens": 866493146.0, + "step": 22325 + }, + { + "epoch": 2.8400966798117286, + "grad_norm": 2.9885458946228027, + "learning_rate": 1e-06, + "loss": 0.253, + "mean_token_accuracy": 0.9045581817626953, + "num_tokens": 866528849.0, + "step": 22326 + }, + { + "epoch": 2.840223890090319, + "grad_norm": 2.8972513675689697, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.8995078802108765, + "num_tokens": 866567817.0, + "step": 22327 + }, + { + "epoch": 2.8403511003689097, + "grad_norm": 2.884383201599121, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8954625129699707, + "num_tokens": 866607578.0, + "step": 22328 + }, + { + "epoch": 2.8404783106475002, + "grad_norm": 2.825538158416748, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9076492786407471, + "num_tokens": 866644149.0, + "step": 22329 + }, + { + "epoch": 2.8406055209260908, + "grad_norm": 2.806450843811035, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8947890400886536, + "num_tokens": 866686842.0, + "step": 22330 + }, + { + "epoch": 2.8407327312046813, + "grad_norm": 2.807262897491455, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9020392894744873, + "num_tokens": 866726239.0, + "step": 22331 + }, + { + "epoch": 2.840859941483272, + "grad_norm": 2.791083812713623, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9056164026260376, + "num_tokens": 866767723.0, + "step": 22332 + }, + { + "epoch": 2.8409871517618623, + "grad_norm": 2.829258680343628, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8942441344261169, + "num_tokens": 866812848.0, + "step": 22333 + }, + { + "epoch": 2.841114362040453, + "grad_norm": 2.6040163040161133, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9090265035629272, + "num_tokens": 866853790.0, + "step": 22334 + }, + { + "epoch": 2.8412415723190434, + "grad_norm": 3.1651899814605713, + "learning_rate": 1e-06, + "loss": 0.2634, + "mean_token_accuracy": 0.9026408195495605, + "num_tokens": 866887530.0, + "step": 22335 + }, + { + "epoch": 2.841368782597634, + "grad_norm": 2.8538143634796143, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9112797975540161, + "num_tokens": 866926976.0, + "step": 22336 + }, + { + "epoch": 2.8414959928762245, + "grad_norm": 2.8854682445526123, + "learning_rate": 1e-06, + "loss": 0.2269, + "mean_token_accuracy": 0.9142259955406189, + "num_tokens": 866961850.0, + "step": 22337 + }, + { + "epoch": 2.841623203154815, + "grad_norm": 2.8583691120147705, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9032777547836304, + "num_tokens": 866997956.0, + "step": 22338 + }, + { + "epoch": 2.8417504134334055, + "grad_norm": 2.569401979446411, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9136477708816528, + "num_tokens": 867040386.0, + "step": 22339 + }, + { + "epoch": 2.841877623711996, + "grad_norm": 2.932108163833618, + "learning_rate": 1e-06, + "loss": 0.2141, + "mean_token_accuracy": 0.9202378392219543, + "num_tokens": 867075690.0, + "step": 22340 + }, + { + "epoch": 2.8420048339905866, + "grad_norm": 3.0130162239074707, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.900209903717041, + "num_tokens": 867111781.0, + "step": 22341 + }, + { + "epoch": 2.842132044269177, + "grad_norm": 2.7840285301208496, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9021657109260559, + "num_tokens": 867156327.0, + "step": 22342 + }, + { + "epoch": 2.8422592545477676, + "grad_norm": 3.26827335357666, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8994145393371582, + "num_tokens": 867190879.0, + "step": 22343 + }, + { + "epoch": 2.842386464826358, + "grad_norm": 2.7559938430786133, + "learning_rate": 1e-06, + "loss": 0.2791, + "mean_token_accuracy": 0.8955927491188049, + "num_tokens": 867233071.0, + "step": 22344 + }, + { + "epoch": 2.8425136751049482, + "grad_norm": 2.7856762409210205, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9109553098678589, + "num_tokens": 867269292.0, + "step": 22345 + }, + { + "epoch": 2.842640885383539, + "grad_norm": 2.9272775650024414, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9048030972480774, + "num_tokens": 867305127.0, + "step": 22346 + }, + { + "epoch": 2.8427680956621293, + "grad_norm": 2.8455495834350586, + "learning_rate": 1e-06, + "loss": 0.2471, + "mean_token_accuracy": 0.908448338508606, + "num_tokens": 867342375.0, + "step": 22347 + }, + { + "epoch": 2.8428953059407203, + "grad_norm": 2.9200708866119385, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.8947030901908875, + "num_tokens": 867381187.0, + "step": 22348 + }, + { + "epoch": 2.8430225162193103, + "grad_norm": 2.563197612762451, + "learning_rate": 1e-06, + "loss": 0.2283, + "mean_token_accuracy": 0.9162641763687134, + "num_tokens": 867425142.0, + "step": 22349 + }, + { + "epoch": 2.843149726497901, + "grad_norm": 2.8741345405578613, + "learning_rate": 1e-06, + "loss": 0.277, + "mean_token_accuracy": 0.8980963230133057, + "num_tokens": 867466715.0, + "step": 22350 + }, + { + "epoch": 2.8432769367764914, + "grad_norm": 2.7926759719848633, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8958853483200073, + "num_tokens": 867507807.0, + "step": 22351 + }, + { + "epoch": 2.843404147055082, + "grad_norm": 3.006544828414917, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9046772122383118, + "num_tokens": 867542042.0, + "step": 22352 + }, + { + "epoch": 2.8435313573336725, + "grad_norm": 2.9009463787078857, + "learning_rate": 1e-06, + "loss": 0.2672, + "mean_token_accuracy": 0.9026275277137756, + "num_tokens": 867580320.0, + "step": 22353 + }, + { + "epoch": 2.843658567612263, + "grad_norm": 2.861114978790283, + "learning_rate": 1e-06, + "loss": 0.2286, + "mean_token_accuracy": 0.9125906229019165, + "num_tokens": 867615273.0, + "step": 22354 + }, + { + "epoch": 2.8437857778908535, + "grad_norm": 3.210797071456909, + "learning_rate": 1e-06, + "loss": 0.2411, + "mean_token_accuracy": 0.910856306552887, + "num_tokens": 867646418.0, + "step": 22355 + }, + { + "epoch": 2.843912988169444, + "grad_norm": 2.565765619277954, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.910871684551239, + "num_tokens": 867687967.0, + "step": 22356 + }, + { + "epoch": 2.8440401984480346, + "grad_norm": 3.0084071159362793, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.9019616842269897, + "num_tokens": 867727498.0, + "step": 22357 + }, + { + "epoch": 2.844167408726625, + "grad_norm": 2.9747579097747803, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8948228359222412, + "num_tokens": 867766609.0, + "step": 22358 + }, + { + "epoch": 2.8442946190052156, + "grad_norm": 2.811938762664795, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9040094614028931, + "num_tokens": 867806937.0, + "step": 22359 + }, + { + "epoch": 2.844421829283806, + "grad_norm": 2.813720941543579, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9071162939071655, + "num_tokens": 867843176.0, + "step": 22360 + }, + { + "epoch": 2.8445490395623967, + "grad_norm": 2.76375675201416, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9030288457870483, + "num_tokens": 867886658.0, + "step": 22361 + }, + { + "epoch": 2.844676249840987, + "grad_norm": 2.572248935699463, + "learning_rate": 1e-06, + "loss": 0.2255, + "mean_token_accuracy": 0.9148924350738525, + "num_tokens": 867927767.0, + "step": 22362 + }, + { + "epoch": 2.8448034601195777, + "grad_norm": 2.8720405101776123, + "learning_rate": 1e-06, + "loss": 0.2307, + "mean_token_accuracy": 0.9144209623336792, + "num_tokens": 867964347.0, + "step": 22363 + }, + { + "epoch": 2.8449306703981683, + "grad_norm": 3.0137369632720947, + "learning_rate": 1e-06, + "loss": 0.3326, + "mean_token_accuracy": 0.8788062334060669, + "num_tokens": 868009035.0, + "step": 22364 + }, + { + "epoch": 2.845057880676759, + "grad_norm": 2.8265299797058105, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.8990379571914673, + "num_tokens": 868052377.0, + "step": 22365 + }, + { + "epoch": 2.8451850909553493, + "grad_norm": 2.751176118850708, + "learning_rate": 1e-06, + "loss": 0.2308, + "mean_token_accuracy": 0.914306640625, + "num_tokens": 868089910.0, + "step": 22366 + }, + { + "epoch": 2.84531230123394, + "grad_norm": 2.8269412517547607, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.9021024107933044, + "num_tokens": 868129463.0, + "step": 22367 + }, + { + "epoch": 2.84543951151253, + "grad_norm": 3.2394795417785645, + "learning_rate": 1e-06, + "loss": 0.266, + "mean_token_accuracy": 0.9006195664405823, + "num_tokens": 868161627.0, + "step": 22368 + }, + { + "epoch": 2.845566721791121, + "grad_norm": 2.756584882736206, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9046173691749573, + "num_tokens": 868201302.0, + "step": 22369 + }, + { + "epoch": 2.845693932069711, + "grad_norm": 2.7037994861602783, + "learning_rate": 1e-06, + "loss": 0.2365, + "mean_token_accuracy": 0.9137458801269531, + "num_tokens": 868242618.0, + "step": 22370 + }, + { + "epoch": 2.845821142348302, + "grad_norm": 2.829171657562256, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9083516001701355, + "num_tokens": 868282370.0, + "step": 22371 + }, + { + "epoch": 2.845948352626892, + "grad_norm": 2.684988498687744, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9040190577507019, + "num_tokens": 868322102.0, + "step": 22372 + }, + { + "epoch": 2.846075562905483, + "grad_norm": 2.9299237728118896, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9093004465103149, + "num_tokens": 868356095.0, + "step": 22373 + }, + { + "epoch": 2.846202773184073, + "grad_norm": 2.9501609802246094, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.8997063636779785, + "num_tokens": 868394219.0, + "step": 22374 + }, + { + "epoch": 2.8463299834626636, + "grad_norm": 3.0456109046936035, + "learning_rate": 1e-06, + "loss": 0.2565, + "mean_token_accuracy": 0.903603196144104, + "num_tokens": 868428394.0, + "step": 22375 + }, + { + "epoch": 2.846457193741254, + "grad_norm": 2.8715457916259766, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9049464464187622, + "num_tokens": 868466839.0, + "step": 22376 + }, + { + "epoch": 2.8465844040198447, + "grad_norm": 2.7420732975006104, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.8926382064819336, + "num_tokens": 868511628.0, + "step": 22377 + }, + { + "epoch": 2.8467116142984352, + "grad_norm": 2.829716682434082, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.9155286550521851, + "num_tokens": 868548034.0, + "step": 22378 + }, + { + "epoch": 2.8468388245770258, + "grad_norm": 2.8519535064697266, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9042765498161316, + "num_tokens": 868586345.0, + "step": 22379 + }, + { + "epoch": 2.8469660348556163, + "grad_norm": 2.9300692081451416, + "learning_rate": 1e-06, + "loss": 0.2833, + "mean_token_accuracy": 0.8949105143547058, + "num_tokens": 868626522.0, + "step": 22380 + }, + { + "epoch": 2.847093245134207, + "grad_norm": 3.314511775970459, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9073269367218018, + "num_tokens": 868658532.0, + "step": 22381 + }, + { + "epoch": 2.8472204554127973, + "grad_norm": 3.116096019744873, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9037749171257019, + "num_tokens": 868695501.0, + "step": 22382 + }, + { + "epoch": 2.847347665691388, + "grad_norm": 2.7654411792755127, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9038482904434204, + "num_tokens": 868735793.0, + "step": 22383 + }, + { + "epoch": 2.8474748759699784, + "grad_norm": 2.958017349243164, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8964084982872009, + "num_tokens": 868773086.0, + "step": 22384 + }, + { + "epoch": 2.847602086248569, + "grad_norm": 2.929192066192627, + "learning_rate": 1e-06, + "loss": 0.2869, + "mean_token_accuracy": 0.8970550298690796, + "num_tokens": 868815357.0, + "step": 22385 + }, + { + "epoch": 2.8477292965271594, + "grad_norm": 2.9414350986480713, + "learning_rate": 1e-06, + "loss": 0.2356, + "mean_token_accuracy": 0.9143897294998169, + "num_tokens": 868853271.0, + "step": 22386 + }, + { + "epoch": 2.84785650680575, + "grad_norm": 2.891272783279419, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.9137901663780212, + "num_tokens": 868885478.0, + "step": 22387 + }, + { + "epoch": 2.8479837170843405, + "grad_norm": 2.650935173034668, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9053873419761658, + "num_tokens": 868927031.0, + "step": 22388 + }, + { + "epoch": 2.848110927362931, + "grad_norm": 2.8946828842163086, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.894300639629364, + "num_tokens": 868967740.0, + "step": 22389 + }, + { + "epoch": 2.8482381376415216, + "grad_norm": 2.925511598587036, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.9017157554626465, + "num_tokens": 869008361.0, + "step": 22390 + }, + { + "epoch": 2.848365347920112, + "grad_norm": 2.9351251125335693, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9122664928436279, + "num_tokens": 869041706.0, + "step": 22391 + }, + { + "epoch": 2.8484925581987026, + "grad_norm": 2.7794647216796875, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9070852994918823, + "num_tokens": 869080389.0, + "step": 22392 + }, + { + "epoch": 2.8486197684772927, + "grad_norm": 2.930126428604126, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8978151082992554, + "num_tokens": 869119585.0, + "step": 22393 + }, + { + "epoch": 2.8487469787558837, + "grad_norm": 2.932224750518799, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9087300300598145, + "num_tokens": 869155346.0, + "step": 22394 + }, + { + "epoch": 2.8488741890344738, + "grad_norm": 2.867608070373535, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9087802171707153, + "num_tokens": 869192169.0, + "step": 22395 + }, + { + "epoch": 2.8490013993130647, + "grad_norm": 2.9887495040893555, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9036301970481873, + "num_tokens": 869228330.0, + "step": 22396 + }, + { + "epoch": 2.849128609591655, + "grad_norm": 2.549072265625, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9095450639724731, + "num_tokens": 869274761.0, + "step": 22397 + }, + { + "epoch": 2.849255819870246, + "grad_norm": 2.8871121406555176, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8967156410217285, + "num_tokens": 869314577.0, + "step": 22398 + }, + { + "epoch": 2.849383030148836, + "grad_norm": 2.902772903442383, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8941044211387634, + "num_tokens": 869354685.0, + "step": 22399 + }, + { + "epoch": 2.8495102404274264, + "grad_norm": 3.1384871006011963, + "learning_rate": 1e-06, + "loss": 0.2724, + "mean_token_accuracy": 0.8992096185684204, + "num_tokens": 869391007.0, + "step": 22400 + }, + { + "epoch": 2.849637450706017, + "grad_norm": 2.7984511852264404, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9076367616653442, + "num_tokens": 869432082.0, + "step": 22401 + }, + { + "epoch": 2.8497646609846075, + "grad_norm": 2.855067253112793, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9006835222244263, + "num_tokens": 869471121.0, + "step": 22402 + }, + { + "epoch": 2.849891871263198, + "grad_norm": 2.9704504013061523, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8966410160064697, + "num_tokens": 869508628.0, + "step": 22403 + }, + { + "epoch": 2.8500190815417885, + "grad_norm": 2.9102773666381836, + "learning_rate": 1e-06, + "loss": 0.2513, + "mean_token_accuracy": 0.9043680429458618, + "num_tokens": 869545907.0, + "step": 22404 + }, + { + "epoch": 2.850146291820379, + "grad_norm": 2.938633918762207, + "learning_rate": 1e-06, + "loss": 0.244, + "mean_token_accuracy": 0.9077847599983215, + "num_tokens": 869579418.0, + "step": 22405 + }, + { + "epoch": 2.8502735020989696, + "grad_norm": 2.778815984725952, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9058074951171875, + "num_tokens": 869621401.0, + "step": 22406 + }, + { + "epoch": 2.85040071237756, + "grad_norm": 2.8720314502716064, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.904761552810669, + "num_tokens": 869661414.0, + "step": 22407 + }, + { + "epoch": 2.8505279226561506, + "grad_norm": 3.2014927864074707, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8962956666946411, + "num_tokens": 869695499.0, + "step": 22408 + }, + { + "epoch": 2.850655132934741, + "grad_norm": 2.7135210037231445, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9134507179260254, + "num_tokens": 869737593.0, + "step": 22409 + }, + { + "epoch": 2.8507823432133317, + "grad_norm": 2.663914918899536, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9064936637878418, + "num_tokens": 869777207.0, + "step": 22410 + }, + { + "epoch": 2.850909553491922, + "grad_norm": 2.9637386798858643, + "learning_rate": 1e-06, + "loss": 0.2877, + "mean_token_accuracy": 0.8956186771392822, + "num_tokens": 869815502.0, + "step": 22411 + }, + { + "epoch": 2.8510367637705127, + "grad_norm": 2.862431764602661, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9058008193969727, + "num_tokens": 869851464.0, + "step": 22412 + }, + { + "epoch": 2.8511639740491033, + "grad_norm": 2.668569803237915, + "learning_rate": 1e-06, + "loss": 0.2435, + "mean_token_accuracy": 0.9095931053161621, + "num_tokens": 869894625.0, + "step": 22413 + }, + { + "epoch": 2.851291184327694, + "grad_norm": 3.0774736404418945, + "learning_rate": 1e-06, + "loss": 0.2937, + "mean_token_accuracy": 0.8923027515411377, + "num_tokens": 869932108.0, + "step": 22414 + }, + { + "epoch": 2.8514183946062843, + "grad_norm": 2.825493335723877, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9109742641448975, + "num_tokens": 869971346.0, + "step": 22415 + }, + { + "epoch": 2.851545604884875, + "grad_norm": 3.0421714782714844, + "learning_rate": 1e-06, + "loss": 0.2807, + "mean_token_accuracy": 0.8957250118255615, + "num_tokens": 870010828.0, + "step": 22416 + }, + { + "epoch": 2.8516728151634654, + "grad_norm": 3.0607807636260986, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9079461097717285, + "num_tokens": 870045322.0, + "step": 22417 + }, + { + "epoch": 2.8518000254420555, + "grad_norm": 2.7082746028900146, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9063342809677124, + "num_tokens": 870084643.0, + "step": 22418 + }, + { + "epoch": 2.8519272357206464, + "grad_norm": 3.242534637451172, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8970001935958862, + "num_tokens": 870119055.0, + "step": 22419 + }, + { + "epoch": 2.8520544459992365, + "grad_norm": 2.7366929054260254, + "learning_rate": 1e-06, + "loss": 0.2761, + "mean_token_accuracy": 0.8982783555984497, + "num_tokens": 870162662.0, + "step": 22420 + }, + { + "epoch": 2.8521816562778275, + "grad_norm": 2.9875118732452393, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9060533046722412, + "num_tokens": 870196368.0, + "step": 22421 + }, + { + "epoch": 2.8523088665564176, + "grad_norm": 2.920041799545288, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9031883478164673, + "num_tokens": 870236028.0, + "step": 22422 + }, + { + "epoch": 2.852436076835008, + "grad_norm": 2.6499619483947754, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9047209024429321, + "num_tokens": 870276433.0, + "step": 22423 + }, + { + "epoch": 2.8525632871135986, + "grad_norm": 2.6716761589050293, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.9013267159461975, + "num_tokens": 870321853.0, + "step": 22424 + }, + { + "epoch": 2.852690497392189, + "grad_norm": 2.95629620552063, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.908486545085907, + "num_tokens": 870359285.0, + "step": 22425 + }, + { + "epoch": 2.8528177076707797, + "grad_norm": 2.8503143787384033, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9054617881774902, + "num_tokens": 870397397.0, + "step": 22426 + }, + { + "epoch": 2.85294491794937, + "grad_norm": 2.839324474334717, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9055690765380859, + "num_tokens": 870437542.0, + "step": 22427 + }, + { + "epoch": 2.8530721282279607, + "grad_norm": 2.5848448276519775, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.908909797668457, + "num_tokens": 870481361.0, + "step": 22428 + }, + { + "epoch": 2.8531993385065513, + "grad_norm": 2.8394620418548584, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.8977869749069214, + "num_tokens": 870523530.0, + "step": 22429 + }, + { + "epoch": 2.853326548785142, + "grad_norm": 3.003751039505005, + "learning_rate": 1e-06, + "loss": 0.2948, + "mean_token_accuracy": 0.891717791557312, + "num_tokens": 870562068.0, + "step": 22430 + }, + { + "epoch": 2.8534537590637323, + "grad_norm": 2.6889920234680176, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9026024341583252, + "num_tokens": 870600541.0, + "step": 22431 + }, + { + "epoch": 2.853580969342323, + "grad_norm": 2.8705642223358154, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8955491185188293, + "num_tokens": 870639253.0, + "step": 22432 + }, + { + "epoch": 2.8537081796209134, + "grad_norm": 2.9217801094055176, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9018336534500122, + "num_tokens": 870678657.0, + "step": 22433 + }, + { + "epoch": 2.853835389899504, + "grad_norm": 2.909929037094116, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8938380479812622, + "num_tokens": 870720340.0, + "step": 22434 + }, + { + "epoch": 2.8539626001780944, + "grad_norm": 3.2692391872406006, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8907696008682251, + "num_tokens": 870758538.0, + "step": 22435 + }, + { + "epoch": 2.854089810456685, + "grad_norm": 3.178248167037964, + "learning_rate": 1e-06, + "loss": 0.2741, + "mean_token_accuracy": 0.8986207246780396, + "num_tokens": 870793250.0, + "step": 22436 + }, + { + "epoch": 2.8542170207352755, + "grad_norm": 2.9629154205322266, + "learning_rate": 1e-06, + "loss": 0.2655, + "mean_token_accuracy": 0.9018795490264893, + "num_tokens": 870831443.0, + "step": 22437 + }, + { + "epoch": 2.854344231013866, + "grad_norm": 2.716221332550049, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.8974874019622803, + "num_tokens": 870874383.0, + "step": 22438 + }, + { + "epoch": 2.8544714412924566, + "grad_norm": 2.8279967308044434, + "learning_rate": 1e-06, + "loss": 0.2777, + "mean_token_accuracy": 0.8977786302566528, + "num_tokens": 870921361.0, + "step": 22439 + }, + { + "epoch": 2.854598651571047, + "grad_norm": 2.9611992835998535, + "learning_rate": 1e-06, + "loss": 0.2274, + "mean_token_accuracy": 0.9146103858947754, + "num_tokens": 870955823.0, + "step": 22440 + }, + { + "epoch": 2.8547258618496376, + "grad_norm": 3.389249324798584, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9015412330627441, + "num_tokens": 870986391.0, + "step": 22441 + }, + { + "epoch": 2.854853072128228, + "grad_norm": 2.5271875858306885, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.9128341674804688, + "num_tokens": 871031418.0, + "step": 22442 + }, + { + "epoch": 2.8549802824068182, + "grad_norm": 2.828803300857544, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.9100245237350464, + "num_tokens": 871072981.0, + "step": 22443 + }, + { + "epoch": 2.855107492685409, + "grad_norm": 2.7950923442840576, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9094579219818115, + "num_tokens": 871111561.0, + "step": 22444 + }, + { + "epoch": 2.8552347029639993, + "grad_norm": 2.860912322998047, + "learning_rate": 1e-06, + "loss": 0.2547, + "mean_token_accuracy": 0.904936671257019, + "num_tokens": 871149964.0, + "step": 22445 + }, + { + "epoch": 2.8553619132425903, + "grad_norm": 2.872784376144409, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8989579677581787, + "num_tokens": 871190159.0, + "step": 22446 + }, + { + "epoch": 2.8554891235211803, + "grad_norm": 2.8670575618743896, + "learning_rate": 1e-06, + "loss": 0.2477, + "mean_token_accuracy": 0.9089641571044922, + "num_tokens": 871228568.0, + "step": 22447 + }, + { + "epoch": 2.855616333799771, + "grad_norm": 2.959934949874878, + "learning_rate": 1e-06, + "loss": 0.2393, + "mean_token_accuracy": 0.9110990762710571, + "num_tokens": 871264718.0, + "step": 22448 + }, + { + "epoch": 2.8557435440783614, + "grad_norm": 2.8025922775268555, + "learning_rate": 1e-06, + "loss": 0.2668, + "mean_token_accuracy": 0.902906596660614, + "num_tokens": 871304980.0, + "step": 22449 + }, + { + "epoch": 2.855870754356952, + "grad_norm": 2.8726730346679688, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9012891054153442, + "num_tokens": 871343674.0, + "step": 22450 + }, + { + "epoch": 2.8559979646355425, + "grad_norm": 2.987640857696533, + "learning_rate": 1e-06, + "loss": 0.2487, + "mean_token_accuracy": 0.9084637761116028, + "num_tokens": 871378657.0, + "step": 22451 + }, + { + "epoch": 2.856125174914133, + "grad_norm": 2.827208995819092, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9067046642303467, + "num_tokens": 871418423.0, + "step": 22452 + }, + { + "epoch": 2.8562523851927235, + "grad_norm": 2.731157064437866, + "learning_rate": 1e-06, + "loss": 0.2178, + "mean_token_accuracy": 0.9176636338233948, + "num_tokens": 871453795.0, + "step": 22453 + }, + { + "epoch": 2.856379595471314, + "grad_norm": 2.8348395824432373, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9041471481323242, + "num_tokens": 871491441.0, + "step": 22454 + }, + { + "epoch": 2.8565068057499046, + "grad_norm": 2.9292585849761963, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9060288667678833, + "num_tokens": 871527433.0, + "step": 22455 + }, + { + "epoch": 2.856634016028495, + "grad_norm": 3.1815743446350098, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8933911323547363, + "num_tokens": 871564362.0, + "step": 22456 + }, + { + "epoch": 2.8567612263070856, + "grad_norm": 2.8481180667877197, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9029061794281006, + "num_tokens": 871604564.0, + "step": 22457 + }, + { + "epoch": 2.856888436585676, + "grad_norm": 2.8576438426971436, + "learning_rate": 1e-06, + "loss": 0.2213, + "mean_token_accuracy": 0.9171672463417053, + "num_tokens": 871643589.0, + "step": 22458 + }, + { + "epoch": 2.8570156468642667, + "grad_norm": 3.0183358192443848, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9007658362388611, + "num_tokens": 871680611.0, + "step": 22459 + }, + { + "epoch": 2.857142857142857, + "grad_norm": 3.008924961090088, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9054555296897888, + "num_tokens": 871715790.0, + "step": 22460 + }, + { + "epoch": 2.8572700674214477, + "grad_norm": 2.6667330265045166, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9060987234115601, + "num_tokens": 871757112.0, + "step": 22461 + }, + { + "epoch": 2.8573972777000383, + "grad_norm": 2.6231024265289307, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9063576459884644, + "num_tokens": 871799932.0, + "step": 22462 + }, + { + "epoch": 2.857524487978629, + "grad_norm": 2.6092469692230225, + "learning_rate": 1e-06, + "loss": 0.2299, + "mean_token_accuracy": 0.9118512272834778, + "num_tokens": 871838189.0, + "step": 22463 + }, + { + "epoch": 2.8576516982572193, + "grad_norm": 3.082913875579834, + "learning_rate": 1e-06, + "loss": 0.3047, + "mean_token_accuracy": 0.8916968107223511, + "num_tokens": 871876244.0, + "step": 22464 + }, + { + "epoch": 2.85777890853581, + "grad_norm": 3.305098295211792, + "learning_rate": 1e-06, + "loss": 0.3055, + "mean_token_accuracy": 0.8916342854499817, + "num_tokens": 871912087.0, + "step": 22465 + }, + { + "epoch": 2.8579061188144, + "grad_norm": 2.7362687587738037, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9075368642807007, + "num_tokens": 871951051.0, + "step": 22466 + }, + { + "epoch": 2.858033329092991, + "grad_norm": 2.575589895248413, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9095388054847717, + "num_tokens": 871993679.0, + "step": 22467 + }, + { + "epoch": 2.858160539371581, + "grad_norm": 3.0039358139038086, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8996560573577881, + "num_tokens": 872032848.0, + "step": 22468 + }, + { + "epoch": 2.858287749650172, + "grad_norm": 2.8478004932403564, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9140384197235107, + "num_tokens": 872072558.0, + "step": 22469 + }, + { + "epoch": 2.858414959928762, + "grad_norm": 3.592341661453247, + "learning_rate": 1e-06, + "loss": 0.3274, + "mean_token_accuracy": 0.8797600269317627, + "num_tokens": 872104196.0, + "step": 22470 + }, + { + "epoch": 2.858542170207353, + "grad_norm": 2.8203682899475098, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8922176957130432, + "num_tokens": 872151503.0, + "step": 22471 + }, + { + "epoch": 2.858669380485943, + "grad_norm": 3.133859872817993, + "learning_rate": 1e-06, + "loss": 0.2893, + "mean_token_accuracy": 0.8953492045402527, + "num_tokens": 872188895.0, + "step": 22472 + }, + { + "epoch": 2.8587965907645336, + "grad_norm": 2.9715938568115234, + "learning_rate": 1e-06, + "loss": 0.2776, + "mean_token_accuracy": 0.8973193168640137, + "num_tokens": 872228222.0, + "step": 22473 + }, + { + "epoch": 2.858923801043124, + "grad_norm": 3.2147467136383057, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8934379816055298, + "num_tokens": 872266388.0, + "step": 22474 + }, + { + "epoch": 2.8590510113217147, + "grad_norm": 2.7910356521606445, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.8996795415878296, + "num_tokens": 872304842.0, + "step": 22475 + }, + { + "epoch": 2.859178221600305, + "grad_norm": 2.6693129539489746, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9051133394241333, + "num_tokens": 872346773.0, + "step": 22476 + }, + { + "epoch": 2.8593054318788957, + "grad_norm": 2.694735527038574, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9019991159439087, + "num_tokens": 872390028.0, + "step": 22477 + }, + { + "epoch": 2.8594326421574863, + "grad_norm": 2.6094605922698975, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.903704047203064, + "num_tokens": 872435350.0, + "step": 22478 + }, + { + "epoch": 2.859559852436077, + "grad_norm": 2.685195207595825, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9051111936569214, + "num_tokens": 872477455.0, + "step": 22479 + }, + { + "epoch": 2.8596870627146673, + "grad_norm": 2.7337749004364014, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9106521606445312, + "num_tokens": 872515848.0, + "step": 22480 + }, + { + "epoch": 2.859814272993258, + "grad_norm": 2.8211989402770996, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9104885458946228, + "num_tokens": 872549910.0, + "step": 22481 + }, + { + "epoch": 2.8599414832718484, + "grad_norm": 2.713017225265503, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9056597948074341, + "num_tokens": 872594994.0, + "step": 22482 + }, + { + "epoch": 2.860068693550439, + "grad_norm": 2.6283085346221924, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9016059637069702, + "num_tokens": 872642748.0, + "step": 22483 + }, + { + "epoch": 2.8601959038290294, + "grad_norm": 2.9868812561035156, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9030290842056274, + "num_tokens": 872679615.0, + "step": 22484 + }, + { + "epoch": 2.86032311410762, + "grad_norm": 3.140648603439331, + "learning_rate": 1e-06, + "loss": 0.3384, + "mean_token_accuracy": 0.8774704933166504, + "num_tokens": 872724223.0, + "step": 22485 + }, + { + "epoch": 2.8604503243862105, + "grad_norm": 3.103656768798828, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8989496827125549, + "num_tokens": 872760565.0, + "step": 22486 + }, + { + "epoch": 2.860577534664801, + "grad_norm": 3.0250606536865234, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.9040383100509644, + "num_tokens": 872796122.0, + "step": 22487 + }, + { + "epoch": 2.8607047449433916, + "grad_norm": 2.718153476715088, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9081083536148071, + "num_tokens": 872834723.0, + "step": 22488 + }, + { + "epoch": 2.860831955221982, + "grad_norm": 2.9040443897247314, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9047168493270874, + "num_tokens": 872873198.0, + "step": 22489 + }, + { + "epoch": 2.8609591655005726, + "grad_norm": 2.7824647426605225, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9044510722160339, + "num_tokens": 872914526.0, + "step": 22490 + }, + { + "epoch": 2.8610863757791627, + "grad_norm": 2.755131959915161, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.9154980182647705, + "num_tokens": 872951044.0, + "step": 22491 + }, + { + "epoch": 2.8612135860577537, + "grad_norm": 3.0280022621154785, + "learning_rate": 1e-06, + "loss": 0.2789, + "mean_token_accuracy": 0.8962310552597046, + "num_tokens": 872989193.0, + "step": 22492 + }, + { + "epoch": 2.8613407963363438, + "grad_norm": 2.9826455116271973, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.907169759273529, + "num_tokens": 873021931.0, + "step": 22493 + }, + { + "epoch": 2.8614680066149347, + "grad_norm": 3.0523812770843506, + "learning_rate": 1e-06, + "loss": 0.2438, + "mean_token_accuracy": 0.9105517864227295, + "num_tokens": 873054824.0, + "step": 22494 + }, + { + "epoch": 2.861595216893525, + "grad_norm": 3.018303632736206, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9062545299530029, + "num_tokens": 873090750.0, + "step": 22495 + }, + { + "epoch": 2.861722427172116, + "grad_norm": 2.9869961738586426, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9014297723770142, + "num_tokens": 873127584.0, + "step": 22496 + }, + { + "epoch": 2.861849637450706, + "grad_norm": 2.914390802383423, + "learning_rate": 1e-06, + "loss": 0.269, + "mean_token_accuracy": 0.9023024439811707, + "num_tokens": 873170540.0, + "step": 22497 + }, + { + "epoch": 2.8619768477292964, + "grad_norm": 3.069138288497925, + "learning_rate": 1e-06, + "loss": 0.2626, + "mean_token_accuracy": 0.9024680852890015, + "num_tokens": 873209603.0, + "step": 22498 + }, + { + "epoch": 2.862104058007887, + "grad_norm": 3.1674251556396484, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.8964778184890747, + "num_tokens": 873243720.0, + "step": 22499 + }, + { + "epoch": 2.8622312682864774, + "grad_norm": 2.8611433506011963, + "learning_rate": 1e-06, + "loss": 0.2271, + "mean_token_accuracy": 0.9132081270217896, + "num_tokens": 873282395.0, + "step": 22500 + }, + { + "epoch": 2.862358478565068, + "grad_norm": 2.8619415760040283, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.903841495513916, + "num_tokens": 873323847.0, + "step": 22501 + }, + { + "epoch": 2.8624856888436585, + "grad_norm": 2.830096483230591, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9024050235748291, + "num_tokens": 873363575.0, + "step": 22502 + }, + { + "epoch": 2.862612899122249, + "grad_norm": 2.552205801010132, + "learning_rate": 1e-06, + "loss": 0.2154, + "mean_token_accuracy": 0.9196797013282776, + "num_tokens": 873403331.0, + "step": 22503 + }, + { + "epoch": 2.8627401094008396, + "grad_norm": 2.802022695541382, + "learning_rate": 1e-06, + "loss": 0.2307, + "mean_token_accuracy": 0.9145945310592651, + "num_tokens": 873437280.0, + "step": 22504 + }, + { + "epoch": 2.86286731967943, + "grad_norm": 2.9366612434387207, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9054513573646545, + "num_tokens": 873476161.0, + "step": 22505 + }, + { + "epoch": 2.8629945299580206, + "grad_norm": 3.251117467880249, + "learning_rate": 1e-06, + "loss": 0.293, + "mean_token_accuracy": 0.8909804224967957, + "num_tokens": 873509783.0, + "step": 22506 + }, + { + "epoch": 2.863121740236611, + "grad_norm": 2.9134316444396973, + "learning_rate": 1e-06, + "loss": 0.2193, + "mean_token_accuracy": 0.9189653396606445, + "num_tokens": 873540270.0, + "step": 22507 + }, + { + "epoch": 2.8632489505152017, + "grad_norm": 2.729100227355957, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9062464237213135, + "num_tokens": 873581522.0, + "step": 22508 + }, + { + "epoch": 2.863376160793792, + "grad_norm": 3.022327423095703, + "learning_rate": 1e-06, + "loss": 0.2752, + "mean_token_accuracy": 0.8976375460624695, + "num_tokens": 873619364.0, + "step": 22509 + }, + { + "epoch": 2.8635033710723827, + "grad_norm": 2.7459049224853516, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9003063440322876, + "num_tokens": 873659874.0, + "step": 22510 + }, + { + "epoch": 2.8636305813509733, + "grad_norm": 2.724287748336792, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.907719075679779, + "num_tokens": 873698603.0, + "step": 22511 + }, + { + "epoch": 2.863757791629564, + "grad_norm": 2.7480852603912354, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9061773419380188, + "num_tokens": 873738221.0, + "step": 22512 + }, + { + "epoch": 2.8638850019081543, + "grad_norm": 2.7703287601470947, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.9078534841537476, + "num_tokens": 873776266.0, + "step": 22513 + }, + { + "epoch": 2.864012212186745, + "grad_norm": 2.7162492275238037, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9085004329681396, + "num_tokens": 873815170.0, + "step": 22514 + }, + { + "epoch": 2.8641394224653354, + "grad_norm": 2.8056888580322266, + "learning_rate": 1e-06, + "loss": 0.2421, + "mean_token_accuracy": 0.9093029499053955, + "num_tokens": 873853160.0, + "step": 22515 + }, + { + "epoch": 2.8642666327439255, + "grad_norm": 2.9097506999969482, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.901760458946228, + "num_tokens": 873893140.0, + "step": 22516 + }, + { + "epoch": 2.8643938430225164, + "grad_norm": 3.1702911853790283, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.901864767074585, + "num_tokens": 873928283.0, + "step": 22517 + }, + { + "epoch": 2.8645210533011065, + "grad_norm": 2.8630752563476562, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9058610200881958, + "num_tokens": 873966468.0, + "step": 22518 + }, + { + "epoch": 2.8646482635796975, + "grad_norm": 2.907855749130249, + "learning_rate": 1e-06, + "loss": 0.2106, + "mean_token_accuracy": 0.9200032949447632, + "num_tokens": 874000018.0, + "step": 22519 + }, + { + "epoch": 2.8647754738582876, + "grad_norm": 2.827544689178467, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9037516713142395, + "num_tokens": 874042082.0, + "step": 22520 + }, + { + "epoch": 2.864902684136878, + "grad_norm": 2.793003559112549, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9058278203010559, + "num_tokens": 874082839.0, + "step": 22521 + }, + { + "epoch": 2.8650298944154686, + "grad_norm": 2.866835594177246, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9066888093948364, + "num_tokens": 874118328.0, + "step": 22522 + }, + { + "epoch": 2.865157104694059, + "grad_norm": 2.625711679458618, + "learning_rate": 1e-06, + "loss": 0.2202, + "mean_token_accuracy": 0.9155703783035278, + "num_tokens": 874154201.0, + "step": 22523 + }, + { + "epoch": 2.8652843149726497, + "grad_norm": 2.801788568496704, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9056856632232666, + "num_tokens": 874192039.0, + "step": 22524 + }, + { + "epoch": 2.86541152525124, + "grad_norm": 3.11214542388916, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9031360149383545, + "num_tokens": 874230223.0, + "step": 22525 + }, + { + "epoch": 2.8655387355298307, + "grad_norm": 2.842273473739624, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9051600694656372, + "num_tokens": 874269592.0, + "step": 22526 + }, + { + "epoch": 2.8656659458084213, + "grad_norm": 3.07273268699646, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9041979312896729, + "num_tokens": 874306713.0, + "step": 22527 + }, + { + "epoch": 2.865793156087012, + "grad_norm": 2.616960287094116, + "learning_rate": 1e-06, + "loss": 0.2248, + "mean_token_accuracy": 0.9146568179130554, + "num_tokens": 874346045.0, + "step": 22528 + }, + { + "epoch": 2.8659203663656023, + "grad_norm": 3.0549466609954834, + "learning_rate": 1e-06, + "loss": 0.2422, + "mean_token_accuracy": 0.9108718633651733, + "num_tokens": 874379286.0, + "step": 22529 + }, + { + "epoch": 2.866047576644193, + "grad_norm": 3.0515153408050537, + "learning_rate": 1e-06, + "loss": 0.2998, + "mean_token_accuracy": 0.8923751711845398, + "num_tokens": 874419908.0, + "step": 22530 + }, + { + "epoch": 2.8661747869227834, + "grad_norm": 2.918735980987549, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9032485485076904, + "num_tokens": 874460153.0, + "step": 22531 + }, + { + "epoch": 2.866301997201374, + "grad_norm": 2.96631121635437, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.90211021900177, + "num_tokens": 874503950.0, + "step": 22532 + }, + { + "epoch": 2.8664292074799644, + "grad_norm": 2.949556827545166, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9045323729515076, + "num_tokens": 874542515.0, + "step": 22533 + }, + { + "epoch": 2.866556417758555, + "grad_norm": 2.987185001373291, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9088864326477051, + "num_tokens": 874575866.0, + "step": 22534 + }, + { + "epoch": 2.8666836280371455, + "grad_norm": 3.273421049118042, + "learning_rate": 1e-06, + "loss": 0.3139, + "mean_token_accuracy": 0.883696436882019, + "num_tokens": 874612720.0, + "step": 22535 + }, + { + "epoch": 2.866810838315736, + "grad_norm": 3.081470251083374, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9104806184768677, + "num_tokens": 874648477.0, + "step": 22536 + }, + { + "epoch": 2.8669380485943265, + "grad_norm": 2.5378470420837402, + "learning_rate": 1e-06, + "loss": 0.2282, + "mean_token_accuracy": 0.9149857759475708, + "num_tokens": 874692948.0, + "step": 22537 + }, + { + "epoch": 2.867065258872917, + "grad_norm": 2.875110387802124, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.9012158513069153, + "num_tokens": 874733597.0, + "step": 22538 + }, + { + "epoch": 2.8671924691515076, + "grad_norm": 3.0603811740875244, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9074900150299072, + "num_tokens": 874768412.0, + "step": 22539 + }, + { + "epoch": 2.867319679430098, + "grad_norm": 2.929982900619507, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.907860517501831, + "num_tokens": 874807348.0, + "step": 22540 + }, + { + "epoch": 2.867446889708688, + "grad_norm": 2.891188383102417, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9040113687515259, + "num_tokens": 874843960.0, + "step": 22541 + }, + { + "epoch": 2.867574099987279, + "grad_norm": 2.970338821411133, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.899640679359436, + "num_tokens": 874882000.0, + "step": 22542 + }, + { + "epoch": 2.8677013102658693, + "grad_norm": 2.676966905593872, + "learning_rate": 1e-06, + "loss": 0.2376, + "mean_token_accuracy": 0.9142662882804871, + "num_tokens": 874923461.0, + "step": 22543 + }, + { + "epoch": 2.8678285205444602, + "grad_norm": 2.903329849243164, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.900540292263031, + "num_tokens": 874961503.0, + "step": 22544 + }, + { + "epoch": 2.8679557308230503, + "grad_norm": 2.5633246898651123, + "learning_rate": 1e-06, + "loss": 0.2906, + "mean_token_accuracy": 0.892705500125885, + "num_tokens": 875007522.0, + "step": 22545 + }, + { + "epoch": 2.868082941101641, + "grad_norm": 2.9997105598449707, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.912102997303009, + "num_tokens": 875042081.0, + "step": 22546 + }, + { + "epoch": 2.8682101513802314, + "grad_norm": 2.8412702083587646, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9045079946517944, + "num_tokens": 875080709.0, + "step": 22547 + }, + { + "epoch": 2.868337361658822, + "grad_norm": 2.7448208332061768, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9046508073806763, + "num_tokens": 875121376.0, + "step": 22548 + }, + { + "epoch": 2.8684645719374124, + "grad_norm": 2.677805185317993, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9037291407585144, + "num_tokens": 875164718.0, + "step": 22549 + }, + { + "epoch": 2.868591782216003, + "grad_norm": 2.6261703968048096, + "learning_rate": 1e-06, + "loss": 0.2164, + "mean_token_accuracy": 0.9180104732513428, + "num_tokens": 875198994.0, + "step": 22550 + }, + { + "epoch": 2.8687189924945935, + "grad_norm": 2.651477336883545, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9089022874832153, + "num_tokens": 875240754.0, + "step": 22551 + }, + { + "epoch": 2.868846202773184, + "grad_norm": 2.899914264678955, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8922470808029175, + "num_tokens": 875285205.0, + "step": 22552 + }, + { + "epoch": 2.8689734130517746, + "grad_norm": 3.2504031658172607, + "learning_rate": 1e-06, + "loss": 0.2894, + "mean_token_accuracy": 0.8934434056282043, + "num_tokens": 875321220.0, + "step": 22553 + }, + { + "epoch": 2.869100623330365, + "grad_norm": 2.5945520401000977, + "learning_rate": 1e-06, + "loss": 0.2351, + "mean_token_accuracy": 0.9124270677566528, + "num_tokens": 875360309.0, + "step": 22554 + }, + { + "epoch": 2.8692278336089556, + "grad_norm": 2.907011032104492, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9050777554512024, + "num_tokens": 875395661.0, + "step": 22555 + }, + { + "epoch": 2.869355043887546, + "grad_norm": 2.8471670150756836, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.907810389995575, + "num_tokens": 875435587.0, + "step": 22556 + }, + { + "epoch": 2.8694822541661367, + "grad_norm": 2.548823118209839, + "learning_rate": 1e-06, + "loss": 0.233, + "mean_token_accuracy": 0.914008378982544, + "num_tokens": 875477402.0, + "step": 22557 + }, + { + "epoch": 2.869609464444727, + "grad_norm": 2.726716995239258, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9067303538322449, + "num_tokens": 875518186.0, + "step": 22558 + }, + { + "epoch": 2.8697366747233177, + "grad_norm": 3.012141704559326, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.908120334148407, + "num_tokens": 875554313.0, + "step": 22559 + }, + { + "epoch": 2.8698638850019083, + "grad_norm": 3.3104214668273926, + "learning_rate": 1e-06, + "loss": 0.2823, + "mean_token_accuracy": 0.8947607278823853, + "num_tokens": 875586998.0, + "step": 22560 + }, + { + "epoch": 2.869991095280499, + "grad_norm": 2.8607289791107178, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9029470682144165, + "num_tokens": 875626547.0, + "step": 22561 + }, + { + "epoch": 2.8701183055590893, + "grad_norm": 2.746873140335083, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9062647223472595, + "num_tokens": 875666528.0, + "step": 22562 + }, + { + "epoch": 2.87024551583768, + "grad_norm": 2.8165268898010254, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9041732549667358, + "num_tokens": 875709257.0, + "step": 22563 + }, + { + "epoch": 2.87037272611627, + "grad_norm": 2.769099235534668, + "learning_rate": 1e-06, + "loss": 0.2216, + "mean_token_accuracy": 0.9155542850494385, + "num_tokens": 875745103.0, + "step": 22564 + }, + { + "epoch": 2.870499936394861, + "grad_norm": 3.0246224403381348, + "learning_rate": 1e-06, + "loss": 0.2345, + "mean_token_accuracy": 0.9126625061035156, + "num_tokens": 875778554.0, + "step": 22565 + }, + { + "epoch": 2.870627146673451, + "grad_norm": 2.669869899749756, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9111803770065308, + "num_tokens": 875820177.0, + "step": 22566 + }, + { + "epoch": 2.870754356952042, + "grad_norm": 3.0678279399871826, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9045325517654419, + "num_tokens": 875856610.0, + "step": 22567 + }, + { + "epoch": 2.870881567230632, + "grad_norm": 2.915339469909668, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.910618782043457, + "num_tokens": 875891771.0, + "step": 22568 + }, + { + "epoch": 2.871008777509223, + "grad_norm": 2.826493978500366, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.8992002010345459, + "num_tokens": 875931096.0, + "step": 22569 + }, + { + "epoch": 2.871135987787813, + "grad_norm": 2.659588098526001, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9087325930595398, + "num_tokens": 875974049.0, + "step": 22570 + }, + { + "epoch": 2.8712631980664036, + "grad_norm": 2.8768365383148193, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8926006555557251, + "num_tokens": 876016021.0, + "step": 22571 + }, + { + "epoch": 2.871390408344994, + "grad_norm": 2.7899177074432373, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9109222888946533, + "num_tokens": 876054357.0, + "step": 22572 + }, + { + "epoch": 2.8715176186235847, + "grad_norm": 2.7429544925689697, + "learning_rate": 1e-06, + "loss": 0.2649, + "mean_token_accuracy": 0.9066952466964722, + "num_tokens": 876095735.0, + "step": 22573 + }, + { + "epoch": 2.871644828902175, + "grad_norm": 2.7827415466308594, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.9022908210754395, + "num_tokens": 876137018.0, + "step": 22574 + }, + { + "epoch": 2.8717720391807657, + "grad_norm": 2.7323522567749023, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9071639776229858, + "num_tokens": 876178852.0, + "step": 22575 + }, + { + "epoch": 2.8718992494593563, + "grad_norm": 2.767979383468628, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9059789180755615, + "num_tokens": 876219564.0, + "step": 22576 + }, + { + "epoch": 2.872026459737947, + "grad_norm": 2.6323182582855225, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9076008796691895, + "num_tokens": 876264409.0, + "step": 22577 + }, + { + "epoch": 2.8721536700165373, + "grad_norm": 2.7829184532165527, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9071946144104004, + "num_tokens": 876304549.0, + "step": 22578 + }, + { + "epoch": 2.872280880295128, + "grad_norm": 2.987637519836426, + "learning_rate": 1e-06, + "loss": 0.3134, + "mean_token_accuracy": 0.8875065445899963, + "num_tokens": 876352285.0, + "step": 22579 + }, + { + "epoch": 2.8724080905737184, + "grad_norm": 3.2251534461975098, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8995015621185303, + "num_tokens": 876387570.0, + "step": 22580 + }, + { + "epoch": 2.872535300852309, + "grad_norm": 2.7902143001556396, + "learning_rate": 1e-06, + "loss": 0.2392, + "mean_token_accuracy": 0.9126591682434082, + "num_tokens": 876423663.0, + "step": 22581 + }, + { + "epoch": 2.8726625111308994, + "grad_norm": 2.6979446411132812, + "learning_rate": 1e-06, + "loss": 0.2417, + "mean_token_accuracy": 0.9092512130737305, + "num_tokens": 876464132.0, + "step": 22582 + }, + { + "epoch": 2.87278972140949, + "grad_norm": 2.8503661155700684, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.8981912732124329, + "num_tokens": 876503932.0, + "step": 22583 + }, + { + "epoch": 2.8729169316880805, + "grad_norm": 2.7614083290100098, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9111093282699585, + "num_tokens": 876543823.0, + "step": 22584 + }, + { + "epoch": 2.873044141966671, + "grad_norm": 2.730210781097412, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.896644115447998, + "num_tokens": 876590243.0, + "step": 22585 + }, + { + "epoch": 2.8731713522452615, + "grad_norm": 2.8201515674591064, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9081799983978271, + "num_tokens": 876628529.0, + "step": 22586 + }, + { + "epoch": 2.873298562523852, + "grad_norm": 2.9976284503936768, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9068484306335449, + "num_tokens": 876664637.0, + "step": 22587 + }, + { + "epoch": 2.8734257728024426, + "grad_norm": 3.1433496475219727, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8950508236885071, + "num_tokens": 876699164.0, + "step": 22588 + }, + { + "epoch": 2.8735529830810327, + "grad_norm": 3.1103439331054688, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8930380344390869, + "num_tokens": 876737176.0, + "step": 22589 + }, + { + "epoch": 2.8736801933596237, + "grad_norm": 3.090918779373169, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.8991817235946655, + "num_tokens": 876778926.0, + "step": 22590 + }, + { + "epoch": 2.8738074036382137, + "grad_norm": 2.91029691696167, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.8984055519104004, + "num_tokens": 876817808.0, + "step": 22591 + }, + { + "epoch": 2.8739346139168047, + "grad_norm": 2.703481912612915, + "learning_rate": 1e-06, + "loss": 0.2243, + "mean_token_accuracy": 0.9160661101341248, + "num_tokens": 876855921.0, + "step": 22592 + }, + { + "epoch": 2.874061824195395, + "grad_norm": 2.810258626937866, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9058592319488525, + "num_tokens": 876896663.0, + "step": 22593 + }, + { + "epoch": 2.8741890344739858, + "grad_norm": 2.6382806301116943, + "learning_rate": 1e-06, + "loss": 0.2258, + "mean_token_accuracy": 0.9138603806495667, + "num_tokens": 876935919.0, + "step": 22594 + }, + { + "epoch": 2.874316244752576, + "grad_norm": 3.008284568786621, + "learning_rate": 1e-06, + "loss": 0.2597, + "mean_token_accuracy": 0.9039854407310486, + "num_tokens": 876969631.0, + "step": 22595 + }, + { + "epoch": 2.8744434550311664, + "grad_norm": 2.613612174987793, + "learning_rate": 1e-06, + "loss": 0.2178, + "mean_token_accuracy": 0.9184553623199463, + "num_tokens": 877009037.0, + "step": 22596 + }, + { + "epoch": 2.874570665309757, + "grad_norm": 3.043267011642456, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.9030666947364807, + "num_tokens": 877045509.0, + "step": 22597 + }, + { + "epoch": 2.8746978755883474, + "grad_norm": 2.861786127090454, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9056458473205566, + "num_tokens": 877081569.0, + "step": 22598 + }, + { + "epoch": 2.874825085866938, + "grad_norm": 2.6861188411712646, + "learning_rate": 1e-06, + "loss": 0.2202, + "mean_token_accuracy": 0.9187787175178528, + "num_tokens": 877119135.0, + "step": 22599 + }, + { + "epoch": 2.8749522961455285, + "grad_norm": 2.6784796714782715, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9088495969772339, + "num_tokens": 877161165.0, + "step": 22600 + }, + { + "epoch": 2.875079506424119, + "grad_norm": 3.147395133972168, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9086837768554688, + "num_tokens": 877193870.0, + "step": 22601 + }, + { + "epoch": 2.8752067167027096, + "grad_norm": 3.1380999088287354, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9048233032226562, + "num_tokens": 877225029.0, + "step": 22602 + }, + { + "epoch": 2.8753339269813, + "grad_norm": 2.9803848266601562, + "learning_rate": 1e-06, + "loss": 0.234, + "mean_token_accuracy": 0.9142310619354248, + "num_tokens": 877259630.0, + "step": 22603 + }, + { + "epoch": 2.8754611372598906, + "grad_norm": 2.839402198791504, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9034603238105774, + "num_tokens": 877300919.0, + "step": 22604 + }, + { + "epoch": 2.875588347538481, + "grad_norm": 2.7423524856567383, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.9007239937782288, + "num_tokens": 877347846.0, + "step": 22605 + }, + { + "epoch": 2.8757155578170717, + "grad_norm": 2.832751512527466, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9170530438423157, + "num_tokens": 877383348.0, + "step": 22606 + }, + { + "epoch": 2.875842768095662, + "grad_norm": 2.719789981842041, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9075639247894287, + "num_tokens": 877423455.0, + "step": 22607 + }, + { + "epoch": 2.8759699783742527, + "grad_norm": 2.894245147705078, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.8996428847312927, + "num_tokens": 877462371.0, + "step": 22608 + }, + { + "epoch": 2.8760971886528433, + "grad_norm": 3.200688600540161, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.902369499206543, + "num_tokens": 877492751.0, + "step": 22609 + }, + { + "epoch": 2.876224398931434, + "grad_norm": 2.793290376663208, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9003773927688599, + "num_tokens": 877532646.0, + "step": 22610 + }, + { + "epoch": 2.8763516092100243, + "grad_norm": 3.182643413543701, + "learning_rate": 1e-06, + "loss": 0.2938, + "mean_token_accuracy": 0.8897222280502319, + "num_tokens": 877566643.0, + "step": 22611 + }, + { + "epoch": 2.876478819488615, + "grad_norm": 2.9293220043182373, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9087628126144409, + "num_tokens": 877603708.0, + "step": 22612 + }, + { + "epoch": 2.8766060297672054, + "grad_norm": 2.8738434314727783, + "learning_rate": 1e-06, + "loss": 0.2537, + "mean_token_accuracy": 0.9076597690582275, + "num_tokens": 877640163.0, + "step": 22613 + }, + { + "epoch": 2.8767332400457954, + "grad_norm": 2.6394600868225098, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9108777642250061, + "num_tokens": 877681197.0, + "step": 22614 + }, + { + "epoch": 2.8768604503243864, + "grad_norm": 2.6212270259857178, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9042974710464478, + "num_tokens": 877725641.0, + "step": 22615 + }, + { + "epoch": 2.8769876606029765, + "grad_norm": 3.0068154335021973, + "learning_rate": 1e-06, + "loss": 0.2608, + "mean_token_accuracy": 0.9047670960426331, + "num_tokens": 877763303.0, + "step": 22616 + }, + { + "epoch": 2.8771148708815675, + "grad_norm": 3.0058786869049072, + "learning_rate": 1e-06, + "loss": 0.2971, + "mean_token_accuracy": 0.8896857500076294, + "num_tokens": 877802166.0, + "step": 22617 + }, + { + "epoch": 2.8772420811601576, + "grad_norm": 2.891375780105591, + "learning_rate": 1e-06, + "loss": 0.2558, + "mean_token_accuracy": 0.9026252031326294, + "num_tokens": 877836481.0, + "step": 22618 + }, + { + "epoch": 2.877369291438748, + "grad_norm": 2.797825813293457, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8955318331718445, + "num_tokens": 877878195.0, + "step": 22619 + }, + { + "epoch": 2.8774965017173386, + "grad_norm": 2.7526352405548096, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9024403691291809, + "num_tokens": 877920384.0, + "step": 22620 + }, + { + "epoch": 2.877623711995929, + "grad_norm": 2.872982978820801, + "learning_rate": 1e-06, + "loss": 0.2506, + "mean_token_accuracy": 0.9056830406188965, + "num_tokens": 877957014.0, + "step": 22621 + }, + { + "epoch": 2.8777509222745197, + "grad_norm": 3.1706721782684326, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8905786275863647, + "num_tokens": 877995795.0, + "step": 22622 + }, + { + "epoch": 2.87787813255311, + "grad_norm": 2.9433670043945312, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.897213876247406, + "num_tokens": 878037197.0, + "step": 22623 + }, + { + "epoch": 2.8780053428317007, + "grad_norm": 2.813767194747925, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9073373079299927, + "num_tokens": 878077459.0, + "step": 22624 + }, + { + "epoch": 2.8781325531102913, + "grad_norm": 3.226806402206421, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.902813196182251, + "num_tokens": 878107259.0, + "step": 22625 + }, + { + "epoch": 2.878259763388882, + "grad_norm": 2.9437127113342285, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.9056249856948853, + "num_tokens": 878144778.0, + "step": 22626 + }, + { + "epoch": 2.8783869736674723, + "grad_norm": 2.740375518798828, + "learning_rate": 1e-06, + "loss": 0.2198, + "mean_token_accuracy": 0.9181760549545288, + "num_tokens": 878180697.0, + "step": 22627 + }, + { + "epoch": 2.878514183946063, + "grad_norm": 2.8704402446746826, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9037172794342041, + "num_tokens": 878221405.0, + "step": 22628 + }, + { + "epoch": 2.8786413942246534, + "grad_norm": 2.8125252723693848, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9121750593185425, + "num_tokens": 878258033.0, + "step": 22629 + }, + { + "epoch": 2.878768604503244, + "grad_norm": 3.2698025703430176, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9081371426582336, + "num_tokens": 878286690.0, + "step": 22630 + }, + { + "epoch": 2.8788958147818344, + "grad_norm": 2.8308935165405273, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.908839762210846, + "num_tokens": 878326605.0, + "step": 22631 + }, + { + "epoch": 2.879023025060425, + "grad_norm": 3.0214695930480957, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.9002403020858765, + "num_tokens": 878363401.0, + "step": 22632 + }, + { + "epoch": 2.8791502353390155, + "grad_norm": 2.739551544189453, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9053570032119751, + "num_tokens": 878403418.0, + "step": 22633 + }, + { + "epoch": 2.879277445617606, + "grad_norm": 2.838327407836914, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9032798409461975, + "num_tokens": 878441656.0, + "step": 22634 + }, + { + "epoch": 2.8794046558961965, + "grad_norm": 2.8350865840911865, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.8987418413162231, + "num_tokens": 878483496.0, + "step": 22635 + }, + { + "epoch": 2.879531866174787, + "grad_norm": 3.0904312133789062, + "learning_rate": 1e-06, + "loss": 0.28, + "mean_token_accuracy": 0.8974915742874146, + "num_tokens": 878521356.0, + "step": 22636 + }, + { + "epoch": 2.8796590764533776, + "grad_norm": 2.8331706523895264, + "learning_rate": 1e-06, + "loss": 0.3051, + "mean_token_accuracy": 0.8906862139701843, + "num_tokens": 878570189.0, + "step": 22637 + }, + { + "epoch": 2.879786286731968, + "grad_norm": 2.7266180515289307, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9079352617263794, + "num_tokens": 878609789.0, + "step": 22638 + }, + { + "epoch": 2.879913497010558, + "grad_norm": 2.7999989986419678, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.907524824142456, + "num_tokens": 878647633.0, + "step": 22639 + }, + { + "epoch": 2.880040707289149, + "grad_norm": 3.1023919582366943, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.894356906414032, + "num_tokens": 878686135.0, + "step": 22640 + }, + { + "epoch": 2.8801679175677393, + "grad_norm": 3.1030654907226562, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.8985344171524048, + "num_tokens": 878723573.0, + "step": 22641 + }, + { + "epoch": 2.8802951278463302, + "grad_norm": 2.6614296436309814, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9131458401679993, + "num_tokens": 878764479.0, + "step": 22642 + }, + { + "epoch": 2.8804223381249203, + "grad_norm": 2.8133466243743896, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.8995774984359741, + "num_tokens": 878805353.0, + "step": 22643 + }, + { + "epoch": 2.880549548403511, + "grad_norm": 2.891763210296631, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9065236449241638, + "num_tokens": 878844765.0, + "step": 22644 + }, + { + "epoch": 2.8806767586821014, + "grad_norm": 2.9101483821868896, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8987109065055847, + "num_tokens": 878882063.0, + "step": 22645 + }, + { + "epoch": 2.880803968960692, + "grad_norm": 2.7318966388702393, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9071381092071533, + "num_tokens": 878922083.0, + "step": 22646 + }, + { + "epoch": 2.8809311792392824, + "grad_norm": 2.8691296577453613, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9080449938774109, + "num_tokens": 878959251.0, + "step": 22647 + }, + { + "epoch": 2.881058389517873, + "grad_norm": 2.6675493717193604, + "learning_rate": 1e-06, + "loss": 0.2228, + "mean_token_accuracy": 0.9167453050613403, + "num_tokens": 878995134.0, + "step": 22648 + }, + { + "epoch": 2.8811855997964635, + "grad_norm": 2.9355292320251465, + "learning_rate": 1e-06, + "loss": 0.2533, + "mean_token_accuracy": 0.905385434627533, + "num_tokens": 879034069.0, + "step": 22649 + }, + { + "epoch": 2.881312810075054, + "grad_norm": 3.004441976547241, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9063259959220886, + "num_tokens": 879067370.0, + "step": 22650 + }, + { + "epoch": 2.8814400203536445, + "grad_norm": 2.9025843143463135, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9081511497497559, + "num_tokens": 879102682.0, + "step": 22651 + }, + { + "epoch": 2.881567230632235, + "grad_norm": 2.8637161254882812, + "learning_rate": 1e-06, + "loss": 0.2283, + "mean_token_accuracy": 0.9150238633155823, + "num_tokens": 879142813.0, + "step": 22652 + }, + { + "epoch": 2.8816944409108256, + "grad_norm": 2.7233405113220215, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9069639444351196, + "num_tokens": 879185232.0, + "step": 22653 + }, + { + "epoch": 2.881821651189416, + "grad_norm": 2.8614468574523926, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.902525007724762, + "num_tokens": 879222847.0, + "step": 22654 + }, + { + "epoch": 2.8819488614680067, + "grad_norm": 3.0954298973083496, + "learning_rate": 1e-06, + "loss": 0.2738, + "mean_token_accuracy": 0.896863579750061, + "num_tokens": 879262491.0, + "step": 22655 + }, + { + "epoch": 2.882076071746597, + "grad_norm": 2.8472163677215576, + "learning_rate": 1e-06, + "loss": 0.2652, + "mean_token_accuracy": 0.8994990587234497, + "num_tokens": 879299672.0, + "step": 22656 + }, + { + "epoch": 2.8822032820251877, + "grad_norm": 2.6794183254241943, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9053023457527161, + "num_tokens": 879340848.0, + "step": 22657 + }, + { + "epoch": 2.8823304923037782, + "grad_norm": 2.807567596435547, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9129304885864258, + "num_tokens": 879376583.0, + "step": 22658 + }, + { + "epoch": 2.8824577025823688, + "grad_norm": 2.708871841430664, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9089964628219604, + "num_tokens": 879415399.0, + "step": 22659 + }, + { + "epoch": 2.8825849128609593, + "grad_norm": 2.6094741821289062, + "learning_rate": 1e-06, + "loss": 0.2398, + "mean_token_accuracy": 0.9088311791419983, + "num_tokens": 879457161.0, + "step": 22660 + }, + { + "epoch": 2.88271212313955, + "grad_norm": 2.8008675575256348, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.902533233165741, + "num_tokens": 879496633.0, + "step": 22661 + }, + { + "epoch": 2.88283933341814, + "grad_norm": 2.8860621452331543, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9000563621520996, + "num_tokens": 879538824.0, + "step": 22662 + }, + { + "epoch": 2.882966543696731, + "grad_norm": 3.1930556297302246, + "learning_rate": 1e-06, + "loss": 0.2947, + "mean_token_accuracy": 0.8947182893753052, + "num_tokens": 879575895.0, + "step": 22663 + }, + { + "epoch": 2.883093753975321, + "grad_norm": 2.6936328411102295, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9046494960784912, + "num_tokens": 879620480.0, + "step": 22664 + }, + { + "epoch": 2.883220964253912, + "grad_norm": 2.827761650085449, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9112703204154968, + "num_tokens": 879659517.0, + "step": 22665 + }, + { + "epoch": 2.883348174532502, + "grad_norm": 2.794693946838379, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.8977729082107544, + "num_tokens": 879702214.0, + "step": 22666 + }, + { + "epoch": 2.883475384811093, + "grad_norm": 2.544992446899414, + "learning_rate": 1e-06, + "loss": 0.2089, + "mean_token_accuracy": 0.9217054843902588, + "num_tokens": 879740641.0, + "step": 22667 + }, + { + "epoch": 2.883602595089683, + "grad_norm": 3.010160207748413, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9027669429779053, + "num_tokens": 879777303.0, + "step": 22668 + }, + { + "epoch": 2.8837298053682736, + "grad_norm": 3.4064433574676514, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9090265035629272, + "num_tokens": 879804523.0, + "step": 22669 + }, + { + "epoch": 2.883857015646864, + "grad_norm": 2.539149761199951, + "learning_rate": 1e-06, + "loss": 0.216, + "mean_token_accuracy": 0.9200775623321533, + "num_tokens": 879847006.0, + "step": 22670 + }, + { + "epoch": 2.8839842259254547, + "grad_norm": 2.7519760131835938, + "learning_rate": 1e-06, + "loss": 0.2632, + "mean_token_accuracy": 0.903531014919281, + "num_tokens": 879891022.0, + "step": 22671 + }, + { + "epoch": 2.884111436204045, + "grad_norm": 2.913421869277954, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9030665159225464, + "num_tokens": 879927466.0, + "step": 22672 + }, + { + "epoch": 2.8842386464826357, + "grad_norm": 2.6272826194763184, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9116570353507996, + "num_tokens": 879969554.0, + "step": 22673 + }, + { + "epoch": 2.8843658567612263, + "grad_norm": 2.61183762550354, + "learning_rate": 1e-06, + "loss": 0.2643, + "mean_token_accuracy": 0.9031800627708435, + "num_tokens": 880017123.0, + "step": 22674 + }, + { + "epoch": 2.884493067039817, + "grad_norm": 2.6247408390045166, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9047200679779053, + "num_tokens": 880061918.0, + "step": 22675 + }, + { + "epoch": 2.8846202773184073, + "grad_norm": 2.751206398010254, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9113085269927979, + "num_tokens": 880103143.0, + "step": 22676 + }, + { + "epoch": 2.884747487596998, + "grad_norm": 2.894911527633667, + "learning_rate": 1e-06, + "loss": 0.2774, + "mean_token_accuracy": 0.8989794254302979, + "num_tokens": 880142868.0, + "step": 22677 + }, + { + "epoch": 2.8848746978755884, + "grad_norm": 3.119582176208496, + "learning_rate": 1e-06, + "loss": 0.2736, + "mean_token_accuracy": 0.8988391160964966, + "num_tokens": 880179413.0, + "step": 22678 + }, + { + "epoch": 2.885001908154179, + "grad_norm": 2.620095729827881, + "learning_rate": 1e-06, + "loss": 0.239, + "mean_token_accuracy": 0.9110337495803833, + "num_tokens": 880220541.0, + "step": 22679 + }, + { + "epoch": 2.8851291184327694, + "grad_norm": 2.7024857997894287, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9119570255279541, + "num_tokens": 880260151.0, + "step": 22680 + }, + { + "epoch": 2.88525632871136, + "grad_norm": 3.2060203552246094, + "learning_rate": 1e-06, + "loss": 0.2487, + "mean_token_accuracy": 0.9070065021514893, + "num_tokens": 880292032.0, + "step": 22681 + }, + { + "epoch": 2.8853835389899505, + "grad_norm": 2.7738282680511475, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9114162921905518, + "num_tokens": 880331823.0, + "step": 22682 + }, + { + "epoch": 2.885510749268541, + "grad_norm": 2.731168508529663, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9033702611923218, + "num_tokens": 880370941.0, + "step": 22683 + }, + { + "epoch": 2.8856379595471315, + "grad_norm": 3.0200612545013428, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9068618416786194, + "num_tokens": 880404403.0, + "step": 22684 + }, + { + "epoch": 2.885765169825722, + "grad_norm": 2.602442979812622, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9103058576583862, + "num_tokens": 880443785.0, + "step": 22685 + }, + { + "epoch": 2.8858923801043126, + "grad_norm": 2.797961950302124, + "learning_rate": 1e-06, + "loss": 0.2324, + "mean_token_accuracy": 0.9117791056632996, + "num_tokens": 880478717.0, + "step": 22686 + }, + { + "epoch": 2.8860195903829027, + "grad_norm": 3.0159406661987305, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9048827886581421, + "num_tokens": 880510440.0, + "step": 22687 + }, + { + "epoch": 2.8861468006614936, + "grad_norm": 2.806964874267578, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9063812494277954, + "num_tokens": 880548259.0, + "step": 22688 + }, + { + "epoch": 2.8862740109400837, + "grad_norm": 3.026186227798462, + "learning_rate": 1e-06, + "loss": 0.3029, + "mean_token_accuracy": 0.8892638683319092, + "num_tokens": 880589396.0, + "step": 22689 + }, + { + "epoch": 2.8864012212186747, + "grad_norm": 2.7923669815063477, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9080450534820557, + "num_tokens": 880625878.0, + "step": 22690 + }, + { + "epoch": 2.886528431497265, + "grad_norm": 2.895867109298706, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9090967774391174, + "num_tokens": 880661412.0, + "step": 22691 + }, + { + "epoch": 2.8866556417758558, + "grad_norm": 2.8046531677246094, + "learning_rate": 1e-06, + "loss": 0.2317, + "mean_token_accuracy": 0.9139165878295898, + "num_tokens": 880700838.0, + "step": 22692 + }, + { + "epoch": 2.886782852054446, + "grad_norm": 2.7768023014068604, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.9088716506958008, + "num_tokens": 880738187.0, + "step": 22693 + }, + { + "epoch": 2.8869100623330364, + "grad_norm": 2.9845705032348633, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9087998270988464, + "num_tokens": 880772252.0, + "step": 22694 + }, + { + "epoch": 2.887037272611627, + "grad_norm": 3.28478741645813, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8864881992340088, + "num_tokens": 880806248.0, + "step": 22695 + }, + { + "epoch": 2.8871644828902174, + "grad_norm": 3.11828875541687, + "learning_rate": 1e-06, + "loss": 0.2951, + "mean_token_accuracy": 0.8922182321548462, + "num_tokens": 880844171.0, + "step": 22696 + }, + { + "epoch": 2.887291693168808, + "grad_norm": 2.7141478061676025, + "learning_rate": 1e-06, + "loss": 0.2454, + "mean_token_accuracy": 0.9087260961532593, + "num_tokens": 880886082.0, + "step": 22697 + }, + { + "epoch": 2.8874189034473985, + "grad_norm": 2.8745434284210205, + "learning_rate": 1e-06, + "loss": 0.2527, + "mean_token_accuracy": 0.9084474444389343, + "num_tokens": 880922627.0, + "step": 22698 + }, + { + "epoch": 2.887546113725989, + "grad_norm": 2.911850929260254, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9073238372802734, + "num_tokens": 880964036.0, + "step": 22699 + }, + { + "epoch": 2.8876733240045795, + "grad_norm": 3.1453638076782227, + "learning_rate": 1e-06, + "loss": 0.3063, + "mean_token_accuracy": 0.8880211710929871, + "num_tokens": 881001708.0, + "step": 22700 + }, + { + "epoch": 2.88780053428317, + "grad_norm": 2.854682445526123, + "learning_rate": 1e-06, + "loss": 0.2438, + "mean_token_accuracy": 0.9103220105171204, + "num_tokens": 881038447.0, + "step": 22701 + }, + { + "epoch": 2.8879277445617606, + "grad_norm": 2.8409695625305176, + "learning_rate": 1e-06, + "loss": 0.2492, + "mean_token_accuracy": 0.9093018770217896, + "num_tokens": 881079524.0, + "step": 22702 + }, + { + "epoch": 2.888054954840351, + "grad_norm": 3.023771286010742, + "learning_rate": 1e-06, + "loss": 0.2856, + "mean_token_accuracy": 0.8960732221603394, + "num_tokens": 881117931.0, + "step": 22703 + }, + { + "epoch": 2.8881821651189417, + "grad_norm": 2.6346564292907715, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9169172048568726, + "num_tokens": 881157892.0, + "step": 22704 + }, + { + "epoch": 2.888309375397532, + "grad_norm": 2.959951162338257, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.903660774230957, + "num_tokens": 881194401.0, + "step": 22705 + }, + { + "epoch": 2.8884365856761227, + "grad_norm": 3.5354881286621094, + "learning_rate": 1e-06, + "loss": 0.2861, + "mean_token_accuracy": 0.894162654876709, + "num_tokens": 881227902.0, + "step": 22706 + }, + { + "epoch": 2.8885637959547132, + "grad_norm": 2.7794597148895264, + "learning_rate": 1e-06, + "loss": 0.2763, + "mean_token_accuracy": 0.8949847221374512, + "num_tokens": 881267951.0, + "step": 22707 + }, + { + "epoch": 2.8886910062333038, + "grad_norm": 2.7949841022491455, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.9100561141967773, + "num_tokens": 881308915.0, + "step": 22708 + }, + { + "epoch": 2.8888182165118943, + "grad_norm": 2.7828269004821777, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9093657732009888, + "num_tokens": 881354832.0, + "step": 22709 + }, + { + "epoch": 2.888945426790485, + "grad_norm": 3.048901081085205, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.8990561366081238, + "num_tokens": 881389688.0, + "step": 22710 + }, + { + "epoch": 2.8890726370690754, + "grad_norm": 2.7736876010894775, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.9104185104370117, + "num_tokens": 881424522.0, + "step": 22711 + }, + { + "epoch": 2.8891998473476654, + "grad_norm": 2.9629619121551514, + "learning_rate": 1e-06, + "loss": 0.2692, + "mean_token_accuracy": 0.9002878665924072, + "num_tokens": 881461255.0, + "step": 22712 + }, + { + "epoch": 2.8893270576262564, + "grad_norm": 2.8230690956115723, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9045013189315796, + "num_tokens": 881503302.0, + "step": 22713 + }, + { + "epoch": 2.8894542679048465, + "grad_norm": 2.9638471603393555, + "learning_rate": 1e-06, + "loss": 0.2414, + "mean_token_accuracy": 0.9105019569396973, + "num_tokens": 881537694.0, + "step": 22714 + }, + { + "epoch": 2.8895814781834375, + "grad_norm": 2.8411295413970947, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9120492935180664, + "num_tokens": 881576078.0, + "step": 22715 + }, + { + "epoch": 2.8897086884620276, + "grad_norm": 2.8853752613067627, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9075236320495605, + "num_tokens": 881610986.0, + "step": 22716 + }, + { + "epoch": 2.889835898740618, + "grad_norm": 2.995262861251831, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9109659194946289, + "num_tokens": 881644580.0, + "step": 22717 + }, + { + "epoch": 2.8899631090192086, + "grad_norm": 2.9905035495758057, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9149215221405029, + "num_tokens": 881675985.0, + "step": 22718 + }, + { + "epoch": 2.890090319297799, + "grad_norm": 2.829669713973999, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9051529169082642, + "num_tokens": 881716268.0, + "step": 22719 + }, + { + "epoch": 2.8902175295763897, + "grad_norm": 2.882899761199951, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.8985888957977295, + "num_tokens": 881758024.0, + "step": 22720 + }, + { + "epoch": 2.89034473985498, + "grad_norm": 2.953270435333252, + "learning_rate": 1e-06, + "loss": 0.2859, + "mean_token_accuracy": 0.8975374698638916, + "num_tokens": 881795181.0, + "step": 22721 + }, + { + "epoch": 2.8904719501335707, + "grad_norm": 2.8327956199645996, + "learning_rate": 1e-06, + "loss": 0.2256, + "mean_token_accuracy": 0.9144813418388367, + "num_tokens": 881830033.0, + "step": 22722 + }, + { + "epoch": 2.8905991604121613, + "grad_norm": 2.6270840167999268, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8982895016670227, + "num_tokens": 881877640.0, + "step": 22723 + }, + { + "epoch": 2.890726370690752, + "grad_norm": 3.175771713256836, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9075946807861328, + "num_tokens": 881908957.0, + "step": 22724 + }, + { + "epoch": 2.8908535809693423, + "grad_norm": 2.9010775089263916, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9115391969680786, + "num_tokens": 881943971.0, + "step": 22725 + }, + { + "epoch": 2.890980791247933, + "grad_norm": 2.6017274856567383, + "learning_rate": 1e-06, + "loss": 0.2483, + "mean_token_accuracy": 0.9077023267745972, + "num_tokens": 881988285.0, + "step": 22726 + }, + { + "epoch": 2.8911080015265234, + "grad_norm": 2.6701059341430664, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.911257266998291, + "num_tokens": 882031382.0, + "step": 22727 + }, + { + "epoch": 2.891235211805114, + "grad_norm": 2.703887462615967, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.9022030234336853, + "num_tokens": 882075684.0, + "step": 22728 + }, + { + "epoch": 2.8913624220837044, + "grad_norm": 2.962070941925049, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9045248031616211, + "num_tokens": 882112227.0, + "step": 22729 + }, + { + "epoch": 2.891489632362295, + "grad_norm": 2.6379594802856445, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9031755328178406, + "num_tokens": 882156327.0, + "step": 22730 + }, + { + "epoch": 2.8916168426408855, + "grad_norm": 2.810309648513794, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9010841846466064, + "num_tokens": 882199680.0, + "step": 22731 + }, + { + "epoch": 2.891744052919476, + "grad_norm": 2.761183738708496, + "learning_rate": 1e-06, + "loss": 0.2229, + "mean_token_accuracy": 0.915366530418396, + "num_tokens": 882235703.0, + "step": 22732 + }, + { + "epoch": 2.8918712631980665, + "grad_norm": 2.7961299419403076, + "learning_rate": 1e-06, + "loss": 0.2526, + "mean_token_accuracy": 0.9080850481987, + "num_tokens": 882279747.0, + "step": 22733 + }, + { + "epoch": 2.891998473476657, + "grad_norm": 2.6833534240722656, + "learning_rate": 1e-06, + "loss": 0.2237, + "mean_token_accuracy": 0.9152815341949463, + "num_tokens": 882317244.0, + "step": 22734 + }, + { + "epoch": 2.8921256837552476, + "grad_norm": 2.872244358062744, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9109787940979004, + "num_tokens": 882356325.0, + "step": 22735 + }, + { + "epoch": 2.892252894033838, + "grad_norm": 2.5721418857574463, + "learning_rate": 1e-06, + "loss": 0.2234, + "mean_token_accuracy": 0.9169561266899109, + "num_tokens": 882399998.0, + "step": 22736 + }, + { + "epoch": 2.892380104312428, + "grad_norm": 2.9080870151519775, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9102270603179932, + "num_tokens": 882436401.0, + "step": 22737 + }, + { + "epoch": 2.892507314591019, + "grad_norm": 2.8026649951934814, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8989756107330322, + "num_tokens": 882483797.0, + "step": 22738 + }, + { + "epoch": 2.8926345248696093, + "grad_norm": 3.0713951587677, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8968538641929626, + "num_tokens": 882523238.0, + "step": 22739 + }, + { + "epoch": 2.8927617351482002, + "grad_norm": 2.8857581615448, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9002209901809692, + "num_tokens": 882563976.0, + "step": 22740 + }, + { + "epoch": 2.8928889454267903, + "grad_norm": 2.945220947265625, + "learning_rate": 1e-06, + "loss": 0.2901, + "mean_token_accuracy": 0.8940838575363159, + "num_tokens": 882605301.0, + "step": 22741 + }, + { + "epoch": 2.893016155705381, + "grad_norm": 2.7126624584198, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9085771441459656, + "num_tokens": 882644146.0, + "step": 22742 + }, + { + "epoch": 2.8931433659839714, + "grad_norm": 2.797482967376709, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9058237075805664, + "num_tokens": 882679566.0, + "step": 22743 + }, + { + "epoch": 2.893270576262562, + "grad_norm": 2.659015417098999, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9068636894226074, + "num_tokens": 882723681.0, + "step": 22744 + }, + { + "epoch": 2.8933977865411524, + "grad_norm": 3.0787930488586426, + "learning_rate": 1e-06, + "loss": 0.2488, + "mean_token_accuracy": 0.9044657945632935, + "num_tokens": 882758282.0, + "step": 22745 + }, + { + "epoch": 2.893524996819743, + "grad_norm": 2.819957733154297, + "learning_rate": 1e-06, + "loss": 0.2674, + "mean_token_accuracy": 0.9021039009094238, + "num_tokens": 882798901.0, + "step": 22746 + }, + { + "epoch": 2.8936522070983335, + "grad_norm": 2.8774077892303467, + "learning_rate": 1e-06, + "loss": 0.2878, + "mean_token_accuracy": 0.8933323621749878, + "num_tokens": 882843043.0, + "step": 22747 + }, + { + "epoch": 2.893779417376924, + "grad_norm": 2.9242587089538574, + "learning_rate": 1e-06, + "loss": 0.308, + "mean_token_accuracy": 0.8922326564788818, + "num_tokens": 882884967.0, + "step": 22748 + }, + { + "epoch": 2.8939066276555145, + "grad_norm": 2.6284468173980713, + "learning_rate": 1e-06, + "loss": 0.2445, + "mean_token_accuracy": 0.9097527265548706, + "num_tokens": 882924153.0, + "step": 22749 + }, + { + "epoch": 2.894033837934105, + "grad_norm": 2.8761377334594727, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9032558798789978, + "num_tokens": 882962411.0, + "step": 22750 + }, + { + "epoch": 2.8941610482126956, + "grad_norm": 2.8182718753814697, + "learning_rate": 1e-06, + "loss": 0.2369, + "mean_token_accuracy": 0.912253201007843, + "num_tokens": 882999879.0, + "step": 22751 + }, + { + "epoch": 2.894288258491286, + "grad_norm": 3.1281137466430664, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.9042778015136719, + "num_tokens": 883034136.0, + "step": 22752 + }, + { + "epoch": 2.8944154687698767, + "grad_norm": 3.165585994720459, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.8984034061431885, + "num_tokens": 883068935.0, + "step": 22753 + }, + { + "epoch": 2.894542679048467, + "grad_norm": 2.9624898433685303, + "learning_rate": 1e-06, + "loss": 0.2831, + "mean_token_accuracy": 0.8955157995223999, + "num_tokens": 883110251.0, + "step": 22754 + }, + { + "epoch": 2.8946698893270577, + "grad_norm": 2.981238603591919, + "learning_rate": 1e-06, + "loss": 0.2974, + "mean_token_accuracy": 0.8907979726791382, + "num_tokens": 883151708.0, + "step": 22755 + }, + { + "epoch": 2.8947970996056482, + "grad_norm": 2.963520050048828, + "learning_rate": 1e-06, + "loss": 0.3119, + "mean_token_accuracy": 0.8834353089332581, + "num_tokens": 883193549.0, + "step": 22756 + }, + { + "epoch": 2.8949243098842388, + "grad_norm": 2.4705018997192383, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9046915769577026, + "num_tokens": 883242293.0, + "step": 22757 + }, + { + "epoch": 2.8950515201628293, + "grad_norm": 2.8997344970703125, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9053016901016235, + "num_tokens": 883280908.0, + "step": 22758 + }, + { + "epoch": 2.89517873044142, + "grad_norm": 2.641599416732788, + "learning_rate": 1e-06, + "loss": 0.2312, + "mean_token_accuracy": 0.9141290187835693, + "num_tokens": 883321779.0, + "step": 22759 + }, + { + "epoch": 2.89530594072001, + "grad_norm": 2.829174041748047, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9009028673171997, + "num_tokens": 883361459.0, + "step": 22760 + }, + { + "epoch": 2.895433150998601, + "grad_norm": 2.9747958183288574, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9010531902313232, + "num_tokens": 883393478.0, + "step": 22761 + }, + { + "epoch": 2.895560361277191, + "grad_norm": 2.9119298458099365, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9028123021125793, + "num_tokens": 883432248.0, + "step": 22762 + }, + { + "epoch": 2.895687571555782, + "grad_norm": 3.2280099391937256, + "learning_rate": 1e-06, + "loss": 0.3023, + "mean_token_accuracy": 0.8912432193756104, + "num_tokens": 883466081.0, + "step": 22763 + }, + { + "epoch": 2.895814781834372, + "grad_norm": 2.82553768157959, + "learning_rate": 1e-06, + "loss": 0.2512, + "mean_token_accuracy": 0.9054320454597473, + "num_tokens": 883503013.0, + "step": 22764 + }, + { + "epoch": 2.895941992112963, + "grad_norm": 3.1006686687469482, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9110268950462341, + "num_tokens": 883532474.0, + "step": 22765 + }, + { + "epoch": 2.896069202391553, + "grad_norm": 2.758507013320923, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.903363049030304, + "num_tokens": 883574627.0, + "step": 22766 + }, + { + "epoch": 2.8961964126701436, + "grad_norm": 2.97511887550354, + "learning_rate": 1e-06, + "loss": 0.2931, + "mean_token_accuracy": 0.8922426104545593, + "num_tokens": 883613044.0, + "step": 22767 + }, + { + "epoch": 2.896323622948734, + "grad_norm": 2.887939453125, + "learning_rate": 1e-06, + "loss": 0.2419, + "mean_token_accuracy": 0.9111378192901611, + "num_tokens": 883649081.0, + "step": 22768 + }, + { + "epoch": 2.8964508332273247, + "grad_norm": 2.99238657951355, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8978089094161987, + "num_tokens": 883685524.0, + "step": 22769 + }, + { + "epoch": 2.896578043505915, + "grad_norm": 2.7672202587127686, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9102990627288818, + "num_tokens": 883726811.0, + "step": 22770 + }, + { + "epoch": 2.8967052537845057, + "grad_norm": 3.153042793273926, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.8999526500701904, + "num_tokens": 883758332.0, + "step": 22771 + }, + { + "epoch": 2.8968324640630962, + "grad_norm": 2.744528293609619, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.9008745551109314, + "num_tokens": 883800566.0, + "step": 22772 + }, + { + "epoch": 2.8969596743416868, + "grad_norm": 2.5791916847229004, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.908497154712677, + "num_tokens": 883841622.0, + "step": 22773 + }, + { + "epoch": 2.8970868846202773, + "grad_norm": 2.777838945388794, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.9094848036766052, + "num_tokens": 883882690.0, + "step": 22774 + }, + { + "epoch": 2.897214094898868, + "grad_norm": 2.8189549446105957, + "learning_rate": 1e-06, + "loss": 0.2517, + "mean_token_accuracy": 0.9065502882003784, + "num_tokens": 883920109.0, + "step": 22775 + }, + { + "epoch": 2.8973413051774584, + "grad_norm": 2.686044216156006, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8967800140380859, + "num_tokens": 883966600.0, + "step": 22776 + }, + { + "epoch": 2.897468515456049, + "grad_norm": 2.634427785873413, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9004822373390198, + "num_tokens": 884013026.0, + "step": 22777 + }, + { + "epoch": 2.8975957257346394, + "grad_norm": 2.732009172439575, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.9074296951293945, + "num_tokens": 884053958.0, + "step": 22778 + }, + { + "epoch": 2.89772293601323, + "grad_norm": 2.811328411102295, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9009588956832886, + "num_tokens": 884094867.0, + "step": 22779 + }, + { + "epoch": 2.8978501462918205, + "grad_norm": 2.971897840499878, + "learning_rate": 1e-06, + "loss": 0.2972, + "mean_token_accuracy": 0.8918776512145996, + "num_tokens": 884135014.0, + "step": 22780 + }, + { + "epoch": 2.897977356570411, + "grad_norm": 2.956921339035034, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.906379759311676, + "num_tokens": 884170652.0, + "step": 22781 + }, + { + "epoch": 2.8981045668490015, + "grad_norm": 2.6187405586242676, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.8999191522598267, + "num_tokens": 884213814.0, + "step": 22782 + }, + { + "epoch": 2.898231777127592, + "grad_norm": 2.8693959712982178, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.9009865522384644, + "num_tokens": 884252695.0, + "step": 22783 + }, + { + "epoch": 2.8983589874061826, + "grad_norm": 3.048912525177002, + "learning_rate": 1e-06, + "loss": 0.2599, + "mean_token_accuracy": 0.9052932858467102, + "num_tokens": 884288638.0, + "step": 22784 + }, + { + "epoch": 2.8984861976847727, + "grad_norm": 2.9229884147644043, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8975527286529541, + "num_tokens": 884329341.0, + "step": 22785 + }, + { + "epoch": 2.8986134079633636, + "grad_norm": 2.9284260272979736, + "learning_rate": 1e-06, + "loss": 0.2846, + "mean_token_accuracy": 0.8980211019515991, + "num_tokens": 884369331.0, + "step": 22786 + }, + { + "epoch": 2.8987406182419537, + "grad_norm": 2.705920934677124, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9025574922561646, + "num_tokens": 884411327.0, + "step": 22787 + }, + { + "epoch": 2.8988678285205447, + "grad_norm": 3.041142225265503, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9041404724121094, + "num_tokens": 884450814.0, + "step": 22788 + }, + { + "epoch": 2.898995038799135, + "grad_norm": 3.069214344024658, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8933384418487549, + "num_tokens": 884491112.0, + "step": 22789 + }, + { + "epoch": 2.8991222490777258, + "grad_norm": 2.9005656242370605, + "learning_rate": 1e-06, + "loss": 0.2704, + "mean_token_accuracy": 0.8990653157234192, + "num_tokens": 884527598.0, + "step": 22790 + }, + { + "epoch": 2.899249459356316, + "grad_norm": 3.0904793739318848, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8971601724624634, + "num_tokens": 884564285.0, + "step": 22791 + }, + { + "epoch": 2.8993766696349064, + "grad_norm": 2.7986907958984375, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9080977439880371, + "num_tokens": 884608589.0, + "step": 22792 + }, + { + "epoch": 2.899503879913497, + "grad_norm": 2.9800031185150146, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9009256958961487, + "num_tokens": 884645044.0, + "step": 22793 + }, + { + "epoch": 2.8996310901920874, + "grad_norm": 2.848334312438965, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9101265668869019, + "num_tokens": 884680038.0, + "step": 22794 + }, + { + "epoch": 2.899758300470678, + "grad_norm": 2.956238031387329, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9031033515930176, + "num_tokens": 884717461.0, + "step": 22795 + }, + { + "epoch": 2.8998855107492685, + "grad_norm": 2.6618058681488037, + "learning_rate": 1e-06, + "loss": 0.2522, + "mean_token_accuracy": 0.9082317352294922, + "num_tokens": 884760003.0, + "step": 22796 + }, + { + "epoch": 2.900012721027859, + "grad_norm": 2.939990520477295, + "learning_rate": 1e-06, + "loss": 0.2703, + "mean_token_accuracy": 0.9011014699935913, + "num_tokens": 884796689.0, + "step": 22797 + }, + { + "epoch": 2.9001399313064495, + "grad_norm": 3.0213677883148193, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8923546075820923, + "num_tokens": 884837951.0, + "step": 22798 + }, + { + "epoch": 2.90026714158504, + "grad_norm": 2.8119468688964844, + "learning_rate": 1e-06, + "loss": 0.2611, + "mean_token_accuracy": 0.9041162133216858, + "num_tokens": 884877130.0, + "step": 22799 + }, + { + "epoch": 2.9003943518636306, + "grad_norm": 2.8540337085723877, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9078696370124817, + "num_tokens": 884915277.0, + "step": 22800 + }, + { + "epoch": 2.900521562142221, + "grad_norm": 2.997269630432129, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9029765129089355, + "num_tokens": 884952223.0, + "step": 22801 + }, + { + "epoch": 2.9006487724208116, + "grad_norm": 2.6061599254608154, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9091307520866394, + "num_tokens": 884998562.0, + "step": 22802 + }, + { + "epoch": 2.900775982699402, + "grad_norm": 3.0461976528167725, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8946132659912109, + "num_tokens": 885034835.0, + "step": 22803 + }, + { + "epoch": 2.9009031929779927, + "grad_norm": 3.116049289703369, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.9023133516311646, + "num_tokens": 885068391.0, + "step": 22804 + }, + { + "epoch": 2.9010304032565832, + "grad_norm": 2.681580066680908, + "learning_rate": 1e-06, + "loss": 0.2423, + "mean_token_accuracy": 0.9089322090148926, + "num_tokens": 885109692.0, + "step": 22805 + }, + { + "epoch": 2.9011576135351738, + "grad_norm": 2.6998167037963867, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9122011661529541, + "num_tokens": 885149067.0, + "step": 22806 + }, + { + "epoch": 2.9012848238137643, + "grad_norm": 2.8353912830352783, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9049093127250671, + "num_tokens": 885188877.0, + "step": 22807 + }, + { + "epoch": 2.901412034092355, + "grad_norm": 3.2321481704711914, + "learning_rate": 1e-06, + "loss": 0.2598, + "mean_token_accuracy": 0.9090787172317505, + "num_tokens": 885225380.0, + "step": 22808 + }, + { + "epoch": 2.9015392443709453, + "grad_norm": 2.883230686187744, + "learning_rate": 1e-06, + "loss": 0.2375, + "mean_token_accuracy": 0.9103361964225769, + "num_tokens": 885259059.0, + "step": 22809 + }, + { + "epoch": 2.9016664546495354, + "grad_norm": 2.704450845718384, + "learning_rate": 1e-06, + "loss": 0.2265, + "mean_token_accuracy": 0.9144482612609863, + "num_tokens": 885300895.0, + "step": 22810 + }, + { + "epoch": 2.9017936649281264, + "grad_norm": 3.0171518325805664, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.8992067575454712, + "num_tokens": 885340963.0, + "step": 22811 + }, + { + "epoch": 2.9019208752067165, + "grad_norm": 2.4254422187805176, + "learning_rate": 1e-06, + "loss": 0.2211, + "mean_token_accuracy": 0.9171052575111389, + "num_tokens": 885387462.0, + "step": 22812 + }, + { + "epoch": 2.9020480854853075, + "grad_norm": 2.784440279006958, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.8995351791381836, + "num_tokens": 885431196.0, + "step": 22813 + }, + { + "epoch": 2.9021752957638975, + "grad_norm": 2.5885684490203857, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9047924280166626, + "num_tokens": 885476396.0, + "step": 22814 + }, + { + "epoch": 2.902302506042488, + "grad_norm": 3.104421854019165, + "learning_rate": 1e-06, + "loss": 0.3022, + "mean_token_accuracy": 0.8900416493415833, + "num_tokens": 885519893.0, + "step": 22815 + }, + { + "epoch": 2.9024297163210786, + "grad_norm": 2.806119441986084, + "learning_rate": 1e-06, + "loss": 0.2164, + "mean_token_accuracy": 0.9182933568954468, + "num_tokens": 885554559.0, + "step": 22816 + }, + { + "epoch": 2.902556926599669, + "grad_norm": 2.684692621231079, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.89870285987854, + "num_tokens": 885600144.0, + "step": 22817 + }, + { + "epoch": 2.9026841368782597, + "grad_norm": 2.7747151851654053, + "learning_rate": 1e-06, + "loss": 0.2491, + "mean_token_accuracy": 0.9095280170440674, + "num_tokens": 885642560.0, + "step": 22818 + }, + { + "epoch": 2.90281134715685, + "grad_norm": 2.7341959476470947, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.8997020721435547, + "num_tokens": 885683952.0, + "step": 22819 + }, + { + "epoch": 2.9029385574354407, + "grad_norm": 2.7898929119110107, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.9020626544952393, + "num_tokens": 885724194.0, + "step": 22820 + }, + { + "epoch": 2.9030657677140312, + "grad_norm": 2.8567254543304443, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9021150469779968, + "num_tokens": 885762898.0, + "step": 22821 + }, + { + "epoch": 2.9031929779926218, + "grad_norm": 2.8372998237609863, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9065796732902527, + "num_tokens": 885802194.0, + "step": 22822 + }, + { + "epoch": 2.9033201882712123, + "grad_norm": 2.829737424850464, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9044617414474487, + "num_tokens": 885843280.0, + "step": 22823 + }, + { + "epoch": 2.903447398549803, + "grad_norm": 2.688680410385132, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.9083185195922852, + "num_tokens": 885885264.0, + "step": 22824 + }, + { + "epoch": 2.9035746088283934, + "grad_norm": 2.8921689987182617, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.9040476679801941, + "num_tokens": 885925864.0, + "step": 22825 + }, + { + "epoch": 2.903701819106984, + "grad_norm": 3.391140937805176, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8939914107322693, + "num_tokens": 885954816.0, + "step": 22826 + }, + { + "epoch": 2.9038290293855744, + "grad_norm": 2.736365556716919, + "learning_rate": 1e-06, + "loss": 0.2233, + "mean_token_accuracy": 0.9134639501571655, + "num_tokens": 885993315.0, + "step": 22827 + }, + { + "epoch": 2.903956239664165, + "grad_norm": 3.0505690574645996, + "learning_rate": 1e-06, + "loss": 0.2354, + "mean_token_accuracy": 0.9100449085235596, + "num_tokens": 886028390.0, + "step": 22828 + }, + { + "epoch": 2.9040834499427555, + "grad_norm": 3.121760845184326, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.8987444043159485, + "num_tokens": 886063142.0, + "step": 22829 + }, + { + "epoch": 2.904210660221346, + "grad_norm": 2.9258689880371094, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9109586477279663, + "num_tokens": 886099555.0, + "step": 22830 + }, + { + "epoch": 2.9043378704999365, + "grad_norm": 3.0284104347229004, + "learning_rate": 1e-06, + "loss": 0.2484, + "mean_token_accuracy": 0.9047729969024658, + "num_tokens": 886131619.0, + "step": 22831 + }, + { + "epoch": 2.904465080778527, + "grad_norm": 3.2853589057922363, + "learning_rate": 1e-06, + "loss": 0.288, + "mean_token_accuracy": 0.8939772248268127, + "num_tokens": 886166042.0, + "step": 22832 + }, + { + "epoch": 2.9045922910571176, + "grad_norm": 3.2601826190948486, + "learning_rate": 1e-06, + "loss": 0.3335, + "mean_token_accuracy": 0.8795318007469177, + "num_tokens": 886202318.0, + "step": 22833 + }, + { + "epoch": 2.904719501335708, + "grad_norm": 2.620257616043091, + "learning_rate": 1e-06, + "loss": 0.2393, + "mean_token_accuracy": 0.9112765789031982, + "num_tokens": 886243690.0, + "step": 22834 + }, + { + "epoch": 2.904846711614298, + "grad_norm": 2.6687960624694824, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9103907346725464, + "num_tokens": 886285043.0, + "step": 22835 + }, + { + "epoch": 2.904973921892889, + "grad_norm": 2.7125823497772217, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9046616554260254, + "num_tokens": 886328540.0, + "step": 22836 + }, + { + "epoch": 2.9051011321714793, + "grad_norm": 3.313760757446289, + "learning_rate": 1e-06, + "loss": 0.2797, + "mean_token_accuracy": 0.8983314037322998, + "num_tokens": 886362566.0, + "step": 22837 + }, + { + "epoch": 2.9052283424500702, + "grad_norm": 3.450728416442871, + "learning_rate": 1e-06, + "loss": 0.2657, + "mean_token_accuracy": 0.9013506174087524, + "num_tokens": 886392033.0, + "step": 22838 + }, + { + "epoch": 2.9053555527286603, + "grad_norm": 2.6827425956726074, + "learning_rate": 1e-06, + "loss": 0.2888, + "mean_token_accuracy": 0.8939608931541443, + "num_tokens": 886438434.0, + "step": 22839 + }, + { + "epoch": 2.905482763007251, + "grad_norm": 2.8166162967681885, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9128358364105225, + "num_tokens": 886476244.0, + "step": 22840 + }, + { + "epoch": 2.9056099732858414, + "grad_norm": 3.0585949420928955, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9050920009613037, + "num_tokens": 886509207.0, + "step": 22841 + }, + { + "epoch": 2.905737183564432, + "grad_norm": 3.0156333446502686, + "learning_rate": 1e-06, + "loss": 0.2648, + "mean_token_accuracy": 0.8973281383514404, + "num_tokens": 886545978.0, + "step": 22842 + }, + { + "epoch": 2.9058643938430224, + "grad_norm": 2.788383960723877, + "learning_rate": 1e-06, + "loss": 0.2518, + "mean_token_accuracy": 0.9063766002655029, + "num_tokens": 886584491.0, + "step": 22843 + }, + { + "epoch": 2.905991604121613, + "grad_norm": 2.6864445209503174, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.8987185955047607, + "num_tokens": 886632805.0, + "step": 22844 + }, + { + "epoch": 2.9061188144002035, + "grad_norm": 2.813502073287964, + "learning_rate": 1e-06, + "loss": 0.2213, + "mean_token_accuracy": 0.9142468571662903, + "num_tokens": 886667972.0, + "step": 22845 + }, + { + "epoch": 2.906246024678794, + "grad_norm": 2.665540933609009, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9032244682312012, + "num_tokens": 886710031.0, + "step": 22846 + }, + { + "epoch": 2.9063732349573845, + "grad_norm": 2.8396151065826416, + "learning_rate": 1e-06, + "loss": 0.2658, + "mean_token_accuracy": 0.9006808996200562, + "num_tokens": 886747002.0, + "step": 22847 + }, + { + "epoch": 2.906500445235975, + "grad_norm": 2.7048356533050537, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9126703143119812, + "num_tokens": 886787932.0, + "step": 22848 + }, + { + "epoch": 2.9066276555145656, + "grad_norm": 2.6232340335845947, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9101049304008484, + "num_tokens": 886829158.0, + "step": 22849 + }, + { + "epoch": 2.906754865793156, + "grad_norm": 3.0014536380767822, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8983573913574219, + "num_tokens": 886871912.0, + "step": 22850 + }, + { + "epoch": 2.9068820760717466, + "grad_norm": 2.949657678604126, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9084243774414062, + "num_tokens": 886907279.0, + "step": 22851 + }, + { + "epoch": 2.907009286350337, + "grad_norm": 2.750748634338379, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.9015399217605591, + "num_tokens": 886948794.0, + "step": 22852 + }, + { + "epoch": 2.9071364966289277, + "grad_norm": 2.6353800296783447, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9042618870735168, + "num_tokens": 886991339.0, + "step": 22853 + }, + { + "epoch": 2.9072637069075182, + "grad_norm": 2.9040849208831787, + "learning_rate": 1e-06, + "loss": 0.2411, + "mean_token_accuracy": 0.9102314710617065, + "num_tokens": 887023698.0, + "step": 22854 + }, + { + "epoch": 2.9073909171861088, + "grad_norm": 2.911874532699585, + "learning_rate": 1e-06, + "loss": 0.2795, + "mean_token_accuracy": 0.8972378373146057, + "num_tokens": 887062162.0, + "step": 22855 + }, + { + "epoch": 2.9075181274646993, + "grad_norm": 2.8546011447906494, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9032056927680969, + "num_tokens": 887099346.0, + "step": 22856 + }, + { + "epoch": 2.90764533774329, + "grad_norm": 2.745882749557495, + "learning_rate": 1e-06, + "loss": 0.2266, + "mean_token_accuracy": 0.9142316579818726, + "num_tokens": 887137552.0, + "step": 22857 + }, + { + "epoch": 2.90777254802188, + "grad_norm": 2.9211792945861816, + "learning_rate": 1e-06, + "loss": 0.2985, + "mean_token_accuracy": 0.8920990824699402, + "num_tokens": 887180960.0, + "step": 22858 + }, + { + "epoch": 2.907899758300471, + "grad_norm": 2.758662700653076, + "learning_rate": 1e-06, + "loss": 0.2321, + "mean_token_accuracy": 0.9137812852859497, + "num_tokens": 887220626.0, + "step": 22859 + }, + { + "epoch": 2.908026968579061, + "grad_norm": 3.0202715396881104, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9086166620254517, + "num_tokens": 887254584.0, + "step": 22860 + }, + { + "epoch": 2.908154178857652, + "grad_norm": 2.863135814666748, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9028573632240295, + "num_tokens": 887294866.0, + "step": 22861 + }, + { + "epoch": 2.908281389136242, + "grad_norm": 2.574215888977051, + "learning_rate": 1e-06, + "loss": 0.2292, + "mean_token_accuracy": 0.9156160354614258, + "num_tokens": 887338571.0, + "step": 22862 + }, + { + "epoch": 2.908408599414833, + "grad_norm": 2.790872097015381, + "learning_rate": 1e-06, + "loss": 0.2407, + "mean_token_accuracy": 0.9079508185386658, + "num_tokens": 887375654.0, + "step": 22863 + }, + { + "epoch": 2.908535809693423, + "grad_norm": 2.8930232524871826, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9065176248550415, + "num_tokens": 887412981.0, + "step": 22864 + }, + { + "epoch": 2.9086630199720136, + "grad_norm": 3.310549259185791, + "learning_rate": 1e-06, + "loss": 0.2666, + "mean_token_accuracy": 0.903445303440094, + "num_tokens": 887443170.0, + "step": 22865 + }, + { + "epoch": 2.908790230250604, + "grad_norm": 2.6993889808654785, + "learning_rate": 1e-06, + "loss": 0.2468, + "mean_token_accuracy": 0.9071063995361328, + "num_tokens": 887483222.0, + "step": 22866 + }, + { + "epoch": 2.9089174405291947, + "grad_norm": 3.0733280181884766, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9018922448158264, + "num_tokens": 887520230.0, + "step": 22867 + }, + { + "epoch": 2.909044650807785, + "grad_norm": 2.670152425765991, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9113689661026001, + "num_tokens": 887560778.0, + "step": 22868 + }, + { + "epoch": 2.9091718610863757, + "grad_norm": 2.8886163234710693, + "learning_rate": 1e-06, + "loss": 0.2224, + "mean_token_accuracy": 0.9136541485786438, + "num_tokens": 887592125.0, + "step": 22869 + }, + { + "epoch": 2.9092990713649662, + "grad_norm": 2.9408931732177734, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.9053950309753418, + "num_tokens": 887629463.0, + "step": 22870 + }, + { + "epoch": 2.9094262816435568, + "grad_norm": 2.796508312225342, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9005758762359619, + "num_tokens": 887671377.0, + "step": 22871 + }, + { + "epoch": 2.9095534919221473, + "grad_norm": 2.8698227405548096, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9003370404243469, + "num_tokens": 887713721.0, + "step": 22872 + }, + { + "epoch": 2.909680702200738, + "grad_norm": 2.7949154376983643, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9024538397789001, + "num_tokens": 887754047.0, + "step": 22873 + }, + { + "epoch": 2.9098079124793284, + "grad_norm": 2.817051649093628, + "learning_rate": 1e-06, + "loss": 0.2852, + "mean_token_accuracy": 0.8940672874450684, + "num_tokens": 887796643.0, + "step": 22874 + }, + { + "epoch": 2.909935122757919, + "grad_norm": 2.8231518268585205, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9073020219802856, + "num_tokens": 887834784.0, + "step": 22875 + }, + { + "epoch": 2.9100623330365094, + "grad_norm": 2.9412102699279785, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9129208922386169, + "num_tokens": 887869998.0, + "step": 22876 + }, + { + "epoch": 2.9101895433151, + "grad_norm": 3.0383310317993164, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9045184254646301, + "num_tokens": 887906930.0, + "step": 22877 + }, + { + "epoch": 2.9103167535936905, + "grad_norm": 2.7034432888031006, + "learning_rate": 1e-06, + "loss": 0.2284, + "mean_token_accuracy": 0.9154548048973083, + "num_tokens": 887945909.0, + "step": 22878 + }, + { + "epoch": 2.910443963872281, + "grad_norm": 3.372012138366699, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.897310733795166, + "num_tokens": 887978487.0, + "step": 22879 + }, + { + "epoch": 2.9105711741508715, + "grad_norm": 2.7433767318725586, + "learning_rate": 1e-06, + "loss": 0.2326, + "mean_token_accuracy": 0.9153152704238892, + "num_tokens": 888016294.0, + "step": 22880 + }, + { + "epoch": 2.910698384429462, + "grad_norm": 2.8783702850341797, + "learning_rate": 1e-06, + "loss": 0.26, + "mean_token_accuracy": 0.903694748878479, + "num_tokens": 888060530.0, + "step": 22881 + }, + { + "epoch": 2.9108255947080526, + "grad_norm": 2.8133482933044434, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9013671875, + "num_tokens": 888098735.0, + "step": 22882 + }, + { + "epoch": 2.9109528049866427, + "grad_norm": 2.7809231281280518, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9075085520744324, + "num_tokens": 888138933.0, + "step": 22883 + }, + { + "epoch": 2.9110800152652336, + "grad_norm": 2.6557888984680176, + "learning_rate": 1e-06, + "loss": 0.2629, + "mean_token_accuracy": 0.9031330347061157, + "num_tokens": 888180811.0, + "step": 22884 + }, + { + "epoch": 2.9112072255438237, + "grad_norm": 2.846519947052002, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.9029486179351807, + "num_tokens": 888219339.0, + "step": 22885 + }, + { + "epoch": 2.9113344358224147, + "grad_norm": 2.8889589309692383, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8972381353378296, + "num_tokens": 888258961.0, + "step": 22886 + }, + { + "epoch": 2.9114616461010048, + "grad_norm": 2.8711705207824707, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.906188428401947, + "num_tokens": 888296524.0, + "step": 22887 + }, + { + "epoch": 2.9115888563795957, + "grad_norm": 2.9420177936553955, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9097688794136047, + "num_tokens": 888330808.0, + "step": 22888 + }, + { + "epoch": 2.911716066658186, + "grad_norm": 2.8567402362823486, + "learning_rate": 1e-06, + "loss": 0.2395, + "mean_token_accuracy": 0.9104345440864563, + "num_tokens": 888364694.0, + "step": 22889 + }, + { + "epoch": 2.9118432769367764, + "grad_norm": 3.097858428955078, + "learning_rate": 1e-06, + "loss": 0.2677, + "mean_token_accuracy": 0.9018411040306091, + "num_tokens": 888397659.0, + "step": 22890 + }, + { + "epoch": 2.911970487215367, + "grad_norm": 2.9487497806549072, + "learning_rate": 1e-06, + "loss": 0.2262, + "mean_token_accuracy": 0.9135468006134033, + "num_tokens": 888431042.0, + "step": 22891 + }, + { + "epoch": 2.9120976974939574, + "grad_norm": 3.007899284362793, + "learning_rate": 1e-06, + "loss": 0.2549, + "mean_token_accuracy": 0.9072262048721313, + "num_tokens": 888466017.0, + "step": 22892 + }, + { + "epoch": 2.912224907772548, + "grad_norm": 2.816413402557373, + "learning_rate": 1e-06, + "loss": 0.2753, + "mean_token_accuracy": 0.9016115665435791, + "num_tokens": 888507730.0, + "step": 22893 + }, + { + "epoch": 2.9123521180511385, + "grad_norm": 2.6258342266082764, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9097663164138794, + "num_tokens": 888551881.0, + "step": 22894 + }, + { + "epoch": 2.912479328329729, + "grad_norm": 3.0142393112182617, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9126778841018677, + "num_tokens": 888587477.0, + "step": 22895 + }, + { + "epoch": 2.9126065386083195, + "grad_norm": 2.61555814743042, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9149110317230225, + "num_tokens": 888625442.0, + "step": 22896 + }, + { + "epoch": 2.91273374888691, + "grad_norm": 2.788809061050415, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8905938863754272, + "num_tokens": 888671588.0, + "step": 22897 + }, + { + "epoch": 2.9128609591655006, + "grad_norm": 2.616389274597168, + "learning_rate": 1e-06, + "loss": 0.2584, + "mean_token_accuracy": 0.9063443541526794, + "num_tokens": 888712726.0, + "step": 22898 + }, + { + "epoch": 2.912988169444091, + "grad_norm": 2.6583478450775146, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.9050137996673584, + "num_tokens": 888753258.0, + "step": 22899 + }, + { + "epoch": 2.9131153797226816, + "grad_norm": 2.9568464756011963, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8983652591705322, + "num_tokens": 888793803.0, + "step": 22900 + }, + { + "epoch": 2.913242590001272, + "grad_norm": 2.898681402206421, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9009208679199219, + "num_tokens": 888832769.0, + "step": 22901 + }, + { + "epoch": 2.9133698002798627, + "grad_norm": 2.6336374282836914, + "learning_rate": 1e-06, + "loss": 0.228, + "mean_token_accuracy": 0.917595386505127, + "num_tokens": 888871809.0, + "step": 22902 + }, + { + "epoch": 2.9134970105584532, + "grad_norm": 2.633369207382202, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9064865112304688, + "num_tokens": 888920327.0, + "step": 22903 + }, + { + "epoch": 2.9136242208370438, + "grad_norm": 2.7749879360198975, + "learning_rate": 1e-06, + "loss": 0.232, + "mean_token_accuracy": 0.9159195423126221, + "num_tokens": 888959021.0, + "step": 22904 + }, + { + "epoch": 2.9137514311156343, + "grad_norm": 2.7351555824279785, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9057416915893555, + "num_tokens": 889003608.0, + "step": 22905 + }, + { + "epoch": 2.913878641394225, + "grad_norm": 3.076225757598877, + "learning_rate": 1e-06, + "loss": 0.2961, + "mean_token_accuracy": 0.8906155824661255, + "num_tokens": 889039544.0, + "step": 22906 + }, + { + "epoch": 2.9140058516728153, + "grad_norm": 3.2213218212127686, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8967607021331787, + "num_tokens": 889070152.0, + "step": 22907 + }, + { + "epoch": 2.9141330619514054, + "grad_norm": 2.926624298095703, + "learning_rate": 1e-06, + "loss": 0.2293, + "mean_token_accuracy": 0.9132708311080933, + "num_tokens": 889103295.0, + "step": 22908 + }, + { + "epoch": 2.9142602722299964, + "grad_norm": 3.159254789352417, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8909491300582886, + "num_tokens": 889141018.0, + "step": 22909 + }, + { + "epoch": 2.9143874825085865, + "grad_norm": 2.6859829425811768, + "learning_rate": 1e-06, + "loss": 0.2418, + "mean_token_accuracy": 0.9088581204414368, + "num_tokens": 889179311.0, + "step": 22910 + }, + { + "epoch": 2.9145146927871775, + "grad_norm": 2.9524197578430176, + "learning_rate": 1e-06, + "loss": 0.254, + "mean_token_accuracy": 0.9101541042327881, + "num_tokens": 889220284.0, + "step": 22911 + }, + { + "epoch": 2.9146419030657675, + "grad_norm": 2.7704250812530518, + "learning_rate": 1e-06, + "loss": 0.2898, + "mean_token_accuracy": 0.8939870595932007, + "num_tokens": 889264893.0, + "step": 22912 + }, + { + "epoch": 2.914769113344358, + "grad_norm": 2.7821409702301025, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8984736800193787, + "num_tokens": 889308894.0, + "step": 22913 + }, + { + "epoch": 2.9148963236229486, + "grad_norm": 2.7057547569274902, + "learning_rate": 1e-06, + "loss": 0.2664, + "mean_token_accuracy": 0.9020265936851501, + "num_tokens": 889352682.0, + "step": 22914 + }, + { + "epoch": 2.915023533901539, + "grad_norm": 2.832507610321045, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9080590009689331, + "num_tokens": 889392593.0, + "step": 22915 + }, + { + "epoch": 2.9151507441801296, + "grad_norm": 2.7639355659484863, + "learning_rate": 1e-06, + "loss": 0.2462, + "mean_token_accuracy": 0.908199667930603, + "num_tokens": 889432012.0, + "step": 22916 + }, + { + "epoch": 2.91527795445872, + "grad_norm": 2.5855038166046143, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9066499471664429, + "num_tokens": 889475621.0, + "step": 22917 + }, + { + "epoch": 2.9154051647373107, + "grad_norm": 2.9876158237457275, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.900351881980896, + "num_tokens": 889510493.0, + "step": 22918 + }, + { + "epoch": 2.9155323750159012, + "grad_norm": 2.863922357559204, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9042196869850159, + "num_tokens": 889548888.0, + "step": 22919 + }, + { + "epoch": 2.9156595852944918, + "grad_norm": 3.2303781509399414, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9113782644271851, + "num_tokens": 889585526.0, + "step": 22920 + }, + { + "epoch": 2.9157867955730823, + "grad_norm": 2.8564040660858154, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9075683355331421, + "num_tokens": 889622151.0, + "step": 22921 + }, + { + "epoch": 2.915914005851673, + "grad_norm": 2.986109495162964, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9111315011978149, + "num_tokens": 889657122.0, + "step": 22922 + }, + { + "epoch": 2.9160412161302633, + "grad_norm": 2.8500235080718994, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9033744931221008, + "num_tokens": 889700405.0, + "step": 22923 + }, + { + "epoch": 2.916168426408854, + "grad_norm": 2.6966381072998047, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9030477404594421, + "num_tokens": 889741395.0, + "step": 22924 + }, + { + "epoch": 2.9162956366874444, + "grad_norm": 3.190565824508667, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.9092369079589844, + "num_tokens": 889774836.0, + "step": 22925 + }, + { + "epoch": 2.916422846966035, + "grad_norm": 2.6324820518493652, + "learning_rate": 1e-06, + "loss": 0.2587, + "mean_token_accuracy": 0.9060182571411133, + "num_tokens": 889822579.0, + "step": 22926 + }, + { + "epoch": 2.9165500572446255, + "grad_norm": 3.03776216506958, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9142894744873047, + "num_tokens": 889852892.0, + "step": 22927 + }, + { + "epoch": 2.916677267523216, + "grad_norm": 2.885502815246582, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9100009202957153, + "num_tokens": 889886122.0, + "step": 22928 + }, + { + "epoch": 2.9168044778018065, + "grad_norm": 2.5861570835113525, + "learning_rate": 1e-06, + "loss": 0.221, + "mean_token_accuracy": 0.9187924861907959, + "num_tokens": 889927003.0, + "step": 22929 + }, + { + "epoch": 2.916931688080397, + "grad_norm": 2.5901989936828613, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9099360108375549, + "num_tokens": 889971422.0, + "step": 22930 + }, + { + "epoch": 2.917058898358987, + "grad_norm": 2.663611888885498, + "learning_rate": 1e-06, + "loss": 0.2357, + "mean_token_accuracy": 0.9123817682266235, + "num_tokens": 890012145.0, + "step": 22931 + }, + { + "epoch": 2.917186108637578, + "grad_norm": 2.7393102645874023, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9002400636672974, + "num_tokens": 890053445.0, + "step": 22932 + }, + { + "epoch": 2.917313318916168, + "grad_norm": 2.7958059310913086, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9017032384872437, + "num_tokens": 890093155.0, + "step": 22933 + }, + { + "epoch": 2.917440529194759, + "grad_norm": 2.751710891723633, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.8998967409133911, + "num_tokens": 890134779.0, + "step": 22934 + }, + { + "epoch": 2.9175677394733492, + "grad_norm": 2.747415781021118, + "learning_rate": 1e-06, + "loss": 0.2604, + "mean_token_accuracy": 0.904830276966095, + "num_tokens": 890177014.0, + "step": 22935 + }, + { + "epoch": 2.91769494975194, + "grad_norm": 2.681922197341919, + "learning_rate": 1e-06, + "loss": 0.2166, + "mean_token_accuracy": 0.9209215641021729, + "num_tokens": 890213483.0, + "step": 22936 + }, + { + "epoch": 2.9178221600305303, + "grad_norm": 3.2401697635650635, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9069740772247314, + "num_tokens": 890246803.0, + "step": 22937 + }, + { + "epoch": 2.917949370309121, + "grad_norm": 3.058027982711792, + "learning_rate": 1e-06, + "loss": 0.2617, + "mean_token_accuracy": 0.9017574787139893, + "num_tokens": 890282239.0, + "step": 22938 + }, + { + "epoch": 2.9180765805877114, + "grad_norm": 2.823957681655884, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.9007167220115662, + "num_tokens": 890321812.0, + "step": 22939 + }, + { + "epoch": 2.918203790866302, + "grad_norm": 2.80826735496521, + "learning_rate": 1e-06, + "loss": 0.2319, + "mean_token_accuracy": 0.9121876955032349, + "num_tokens": 890356556.0, + "step": 22940 + }, + { + "epoch": 2.9183310011448924, + "grad_norm": 2.8833200931549072, + "learning_rate": 1e-06, + "loss": 0.2817, + "mean_token_accuracy": 0.898114025592804, + "num_tokens": 890398216.0, + "step": 22941 + }, + { + "epoch": 2.918458211423483, + "grad_norm": 3.18509840965271, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9013558030128479, + "num_tokens": 890434611.0, + "step": 22942 + }, + { + "epoch": 2.9185854217020735, + "grad_norm": 2.5416676998138428, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9054535031318665, + "num_tokens": 890480478.0, + "step": 22943 + }, + { + "epoch": 2.918712631980664, + "grad_norm": 3.0642552375793457, + "learning_rate": 1e-06, + "loss": 0.2444, + "mean_token_accuracy": 0.911933183670044, + "num_tokens": 890515326.0, + "step": 22944 + }, + { + "epoch": 2.9188398422592545, + "grad_norm": 2.7967820167541504, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8953627347946167, + "num_tokens": 890557134.0, + "step": 22945 + }, + { + "epoch": 2.918967052537845, + "grad_norm": 2.917572021484375, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8934094905853271, + "num_tokens": 890596129.0, + "step": 22946 + }, + { + "epoch": 2.9190942628164356, + "grad_norm": 3.1283445358276367, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9115985035896301, + "num_tokens": 890627550.0, + "step": 22947 + }, + { + "epoch": 2.919221473095026, + "grad_norm": 2.842578649520874, + "learning_rate": 1e-06, + "loss": 0.2902, + "mean_token_accuracy": 0.8941271901130676, + "num_tokens": 890669117.0, + "step": 22948 + }, + { + "epoch": 2.9193486833736166, + "grad_norm": 2.956164598464966, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9024706482887268, + "num_tokens": 890705406.0, + "step": 22949 + }, + { + "epoch": 2.919475893652207, + "grad_norm": 2.841538906097412, + "learning_rate": 1e-06, + "loss": 0.2705, + "mean_token_accuracy": 0.8987845182418823, + "num_tokens": 890745143.0, + "step": 22950 + }, + { + "epoch": 2.9196031039307977, + "grad_norm": 2.9040024280548096, + "learning_rate": 1e-06, + "loss": 0.3213, + "mean_token_accuracy": 0.8839056491851807, + "num_tokens": 890790631.0, + "step": 22951 + }, + { + "epoch": 2.9197303142093882, + "grad_norm": 2.703972578048706, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9089353680610657, + "num_tokens": 890829511.0, + "step": 22952 + }, + { + "epoch": 2.9198575244879788, + "grad_norm": 2.811051607131958, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9032984972000122, + "num_tokens": 890868130.0, + "step": 22953 + }, + { + "epoch": 2.9199847347665693, + "grad_norm": 2.8240854740142822, + "learning_rate": 1e-06, + "loss": 0.2394, + "mean_token_accuracy": 0.906680703163147, + "num_tokens": 890906878.0, + "step": 22954 + }, + { + "epoch": 2.92011194504516, + "grad_norm": 3.2165112495422363, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9031914472579956, + "num_tokens": 890939035.0, + "step": 22955 + }, + { + "epoch": 2.92023915532375, + "grad_norm": 2.9128918647766113, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8940646052360535, + "num_tokens": 890978632.0, + "step": 22956 + }, + { + "epoch": 2.920366365602341, + "grad_norm": 3.003293752670288, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.896811842918396, + "num_tokens": 891016704.0, + "step": 22957 + }, + { + "epoch": 2.920493575880931, + "grad_norm": 3.054206609725952, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.9020710587501526, + "num_tokens": 891053873.0, + "step": 22958 + }, + { + "epoch": 2.920620786159522, + "grad_norm": 2.8746848106384277, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9107308387756348, + "num_tokens": 891089384.0, + "step": 22959 + }, + { + "epoch": 2.920747996438112, + "grad_norm": 3.139902353286743, + "learning_rate": 1e-06, + "loss": 0.3021, + "mean_token_accuracy": 0.8906892538070679, + "num_tokens": 891126830.0, + "step": 22960 + }, + { + "epoch": 2.920875206716703, + "grad_norm": 3.0460922718048096, + "learning_rate": 1e-06, + "loss": 0.2816, + "mean_token_accuracy": 0.8950755596160889, + "num_tokens": 891164698.0, + "step": 22961 + }, + { + "epoch": 2.921002416995293, + "grad_norm": 3.262345314025879, + "learning_rate": 1e-06, + "loss": 0.2854, + "mean_token_accuracy": 0.8969125747680664, + "num_tokens": 891198485.0, + "step": 22962 + }, + { + "epoch": 2.9211296272738836, + "grad_norm": 2.8714804649353027, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9100052714347839, + "num_tokens": 891239779.0, + "step": 22963 + }, + { + "epoch": 2.921256837552474, + "grad_norm": 2.8030076026916504, + "learning_rate": 1e-06, + "loss": 0.2475, + "mean_token_accuracy": 0.9082753658294678, + "num_tokens": 891278604.0, + "step": 22964 + }, + { + "epoch": 2.9213840478310646, + "grad_norm": 2.866381883621216, + "learning_rate": 1e-06, + "loss": 0.2691, + "mean_token_accuracy": 0.9010922312736511, + "num_tokens": 891320309.0, + "step": 22965 + }, + { + "epoch": 2.921511258109655, + "grad_norm": 2.71675705909729, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.9161918759346008, + "num_tokens": 891359880.0, + "step": 22966 + }, + { + "epoch": 2.9216384683882457, + "grad_norm": 3.169787883758545, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.9025678634643555, + "num_tokens": 891396744.0, + "step": 22967 + }, + { + "epoch": 2.9217656786668362, + "grad_norm": 2.76452374458313, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9059446454048157, + "num_tokens": 891438271.0, + "step": 22968 + }, + { + "epoch": 2.9218928889454268, + "grad_norm": 2.8532533645629883, + "learning_rate": 1e-06, + "loss": 0.2552, + "mean_token_accuracy": 0.9047356843948364, + "num_tokens": 891479652.0, + "step": 22969 + }, + { + "epoch": 2.9220200992240173, + "grad_norm": 3.1269326210021973, + "learning_rate": 1e-06, + "loss": 0.2786, + "mean_token_accuracy": 0.8969048261642456, + "num_tokens": 891516198.0, + "step": 22970 + }, + { + "epoch": 2.922147309502608, + "grad_norm": 3.033796548843384, + "learning_rate": 1e-06, + "loss": 0.2543, + "mean_token_accuracy": 0.901860237121582, + "num_tokens": 891548037.0, + "step": 22971 + }, + { + "epoch": 2.9222745197811983, + "grad_norm": 2.9069645404815674, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.910041093826294, + "num_tokens": 891585762.0, + "step": 22972 + }, + { + "epoch": 2.922401730059789, + "grad_norm": 3.253714084625244, + "learning_rate": 1e-06, + "loss": 0.2845, + "mean_token_accuracy": 0.8947290182113647, + "num_tokens": 891619366.0, + "step": 22973 + }, + { + "epoch": 2.9225289403383794, + "grad_norm": 3.0004703998565674, + "learning_rate": 1e-06, + "loss": 0.3093, + "mean_token_accuracy": 0.8892626762390137, + "num_tokens": 891663559.0, + "step": 22974 + }, + { + "epoch": 2.92265615061697, + "grad_norm": 2.921656847000122, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.902580976486206, + "num_tokens": 891701097.0, + "step": 22975 + }, + { + "epoch": 2.9227833608955605, + "grad_norm": 2.830920696258545, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9054470062255859, + "num_tokens": 891742590.0, + "step": 22976 + }, + { + "epoch": 2.922910571174151, + "grad_norm": 2.6389429569244385, + "learning_rate": 1e-06, + "loss": 0.2389, + "mean_token_accuracy": 0.9131327271461487, + "num_tokens": 891783953.0, + "step": 22977 + }, + { + "epoch": 2.9230377814527415, + "grad_norm": 3.137507677078247, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8985838890075684, + "num_tokens": 891816851.0, + "step": 22978 + }, + { + "epoch": 2.923164991731332, + "grad_norm": 3.0928304195404053, + "learning_rate": 1e-06, + "loss": 0.3164, + "mean_token_accuracy": 0.886000394821167, + "num_tokens": 891857010.0, + "step": 22979 + }, + { + "epoch": 2.9232922020099226, + "grad_norm": 2.7619028091430664, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9033434987068176, + "num_tokens": 891899973.0, + "step": 22980 + }, + { + "epoch": 2.9234194122885127, + "grad_norm": 2.950654983520508, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.8975299000740051, + "num_tokens": 891938335.0, + "step": 22981 + }, + { + "epoch": 2.9235466225671036, + "grad_norm": 2.838146448135376, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9090051651000977, + "num_tokens": 891974805.0, + "step": 22982 + }, + { + "epoch": 2.9236738328456937, + "grad_norm": 2.742629051208496, + "learning_rate": 1e-06, + "loss": 0.2448, + "mean_token_accuracy": 0.9083927869796753, + "num_tokens": 892012146.0, + "step": 22983 + }, + { + "epoch": 2.9238010431242847, + "grad_norm": 3.1291160583496094, + "learning_rate": 1e-06, + "loss": 0.2909, + "mean_token_accuracy": 0.8940305709838867, + "num_tokens": 892048890.0, + "step": 22984 + }, + { + "epoch": 2.9239282534028748, + "grad_norm": 3.086073637008667, + "learning_rate": 1e-06, + "loss": 0.2639, + "mean_token_accuracy": 0.9051458239555359, + "num_tokens": 892086917.0, + "step": 22985 + }, + { + "epoch": 2.9240554636814657, + "grad_norm": 3.1450259685516357, + "learning_rate": 1e-06, + "loss": 0.2834, + "mean_token_accuracy": 0.8978676199913025, + "num_tokens": 892120649.0, + "step": 22986 + }, + { + "epoch": 2.924182673960056, + "grad_norm": 2.852757453918457, + "learning_rate": 1e-06, + "loss": 0.2244, + "mean_token_accuracy": 0.9136620163917542, + "num_tokens": 892154917.0, + "step": 22987 + }, + { + "epoch": 2.9243098842386464, + "grad_norm": 2.7747840881347656, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9044104218482971, + "num_tokens": 892198377.0, + "step": 22988 + }, + { + "epoch": 2.924437094517237, + "grad_norm": 3.142082452774048, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9052774906158447, + "num_tokens": 892231135.0, + "step": 22989 + }, + { + "epoch": 2.9245643047958274, + "grad_norm": 3.3003122806549072, + "learning_rate": 1e-06, + "loss": 0.2373, + "mean_token_accuracy": 0.9108266234397888, + "num_tokens": 892264684.0, + "step": 22990 + }, + { + "epoch": 2.924691515074418, + "grad_norm": 2.7344858646392822, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.8950663805007935, + "num_tokens": 892305261.0, + "step": 22991 + }, + { + "epoch": 2.9248187253530085, + "grad_norm": 3.1431145668029785, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9002335071563721, + "num_tokens": 892339822.0, + "step": 22992 + }, + { + "epoch": 2.924945935631599, + "grad_norm": 2.6294260025024414, + "learning_rate": 1e-06, + "loss": 0.2303, + "mean_token_accuracy": 0.9139779806137085, + "num_tokens": 892383478.0, + "step": 22993 + }, + { + "epoch": 2.9250731459101895, + "grad_norm": 2.611853837966919, + "learning_rate": 1e-06, + "loss": 0.2749, + "mean_token_accuracy": 0.89946448802948, + "num_tokens": 892428538.0, + "step": 22994 + }, + { + "epoch": 2.92520035618878, + "grad_norm": 2.921173572540283, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.898391604423523, + "num_tokens": 892467011.0, + "step": 22995 + }, + { + "epoch": 2.9253275664673706, + "grad_norm": 3.2163970470428467, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9010263085365295, + "num_tokens": 892501470.0, + "step": 22996 + }, + { + "epoch": 2.925454776745961, + "grad_norm": 2.8831570148468018, + "learning_rate": 1e-06, + "loss": 0.2615, + "mean_token_accuracy": 0.9041476845741272, + "num_tokens": 892539705.0, + "step": 22997 + }, + { + "epoch": 2.9255819870245516, + "grad_norm": 2.858830451965332, + "learning_rate": 1e-06, + "loss": 0.2566, + "mean_token_accuracy": 0.9055672883987427, + "num_tokens": 892575850.0, + "step": 22998 + }, + { + "epoch": 2.925709197303142, + "grad_norm": 2.7054595947265625, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9166556000709534, + "num_tokens": 892615947.0, + "step": 22999 + }, + { + "epoch": 2.9258364075817327, + "grad_norm": 3.1298046112060547, + "learning_rate": 1e-06, + "loss": 0.249, + "mean_token_accuracy": 0.9067533016204834, + "num_tokens": 892652663.0, + "step": 23000 + }, + { + "epoch": 2.925963617860323, + "grad_norm": 2.789250135421753, + "learning_rate": 1e-06, + "loss": 0.2645, + "mean_token_accuracy": 0.9025155901908875, + "num_tokens": 892692478.0, + "step": 23001 + }, + { + "epoch": 2.9260908281389137, + "grad_norm": 2.7688772678375244, + "learning_rate": 1e-06, + "loss": 0.2872, + "mean_token_accuracy": 0.8943716287612915, + "num_tokens": 892735789.0, + "step": 23002 + }, + { + "epoch": 2.9262180384175043, + "grad_norm": 2.964186668395996, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9050310850143433, + "num_tokens": 892773746.0, + "step": 23003 + }, + { + "epoch": 2.926345248696095, + "grad_norm": 2.922879695892334, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9061170816421509, + "num_tokens": 892812374.0, + "step": 23004 + }, + { + "epoch": 2.9264724589746853, + "grad_norm": 2.725470781326294, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.905092716217041, + "num_tokens": 892858410.0, + "step": 23005 + }, + { + "epoch": 2.9265996692532754, + "grad_norm": 2.7682178020477295, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9110244512557983, + "num_tokens": 892899456.0, + "step": 23006 + }, + { + "epoch": 2.9267268795318664, + "grad_norm": 2.954116106033325, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9052896499633789, + "num_tokens": 892933269.0, + "step": 23007 + }, + { + "epoch": 2.9268540898104565, + "grad_norm": 3.139737367630005, + "learning_rate": 1e-06, + "loss": 0.2747, + "mean_token_accuracy": 0.8987042903900146, + "num_tokens": 892967207.0, + "step": 23008 + }, + { + "epoch": 2.9269813000890474, + "grad_norm": 2.7247426509857178, + "learning_rate": 1e-06, + "loss": 0.2742, + "mean_token_accuracy": 0.8989166617393494, + "num_tokens": 893010421.0, + "step": 23009 + }, + { + "epoch": 2.9271085103676375, + "grad_norm": 2.7181053161621094, + "learning_rate": 1e-06, + "loss": 0.2297, + "mean_token_accuracy": 0.9141203761100769, + "num_tokens": 893050959.0, + "step": 23010 + }, + { + "epoch": 2.927235720646228, + "grad_norm": 2.8888628482818604, + "learning_rate": 1e-06, + "loss": 0.261, + "mean_token_accuracy": 0.9038471579551697, + "num_tokens": 893090661.0, + "step": 23011 + }, + { + "epoch": 2.9273629309248186, + "grad_norm": 2.7863235473632812, + "learning_rate": 1e-06, + "loss": 0.2583, + "mean_token_accuracy": 0.9025377035140991, + "num_tokens": 893133239.0, + "step": 23012 + }, + { + "epoch": 2.927490141203409, + "grad_norm": 2.8765783309936523, + "learning_rate": 1e-06, + "loss": 0.3064, + "mean_token_accuracy": 0.887590765953064, + "num_tokens": 893179128.0, + "step": 23013 + }, + { + "epoch": 2.9276173514819996, + "grad_norm": 2.9773635864257812, + "learning_rate": 1e-06, + "loss": 0.252, + "mean_token_accuracy": 0.9068987965583801, + "num_tokens": 893213895.0, + "step": 23014 + }, + { + "epoch": 2.92774456176059, + "grad_norm": 2.8831286430358887, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9118165373802185, + "num_tokens": 893248339.0, + "step": 23015 + }, + { + "epoch": 2.9278717720391807, + "grad_norm": 2.754650592803955, + "learning_rate": 1e-06, + "loss": 0.2847, + "mean_token_accuracy": 0.8950397968292236, + "num_tokens": 893295808.0, + "step": 23016 + }, + { + "epoch": 2.9279989823177712, + "grad_norm": 3.04621958732605, + "learning_rate": 1e-06, + "loss": 0.3245, + "mean_token_accuracy": 0.8796569108963013, + "num_tokens": 893338366.0, + "step": 23017 + }, + { + "epoch": 2.9281261925963618, + "grad_norm": 3.0586085319519043, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8876975774765015, + "num_tokens": 893380117.0, + "step": 23018 + }, + { + "epoch": 2.9282534028749523, + "grad_norm": 2.9271459579467773, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8962225914001465, + "num_tokens": 893421752.0, + "step": 23019 + }, + { + "epoch": 2.928380613153543, + "grad_norm": 2.75396990776062, + "learning_rate": 1e-06, + "loss": 0.2509, + "mean_token_accuracy": 0.9081283211708069, + "num_tokens": 893461804.0, + "step": 23020 + }, + { + "epoch": 2.9285078234321333, + "grad_norm": 2.9195644855499268, + "learning_rate": 1e-06, + "loss": 0.285, + "mean_token_accuracy": 0.8955414891242981, + "num_tokens": 893503256.0, + "step": 23021 + }, + { + "epoch": 2.928635033710724, + "grad_norm": 2.826646327972412, + "learning_rate": 1e-06, + "loss": 0.2775, + "mean_token_accuracy": 0.9025404453277588, + "num_tokens": 893545897.0, + "step": 23022 + }, + { + "epoch": 2.9287622439893144, + "grad_norm": 2.6283206939697266, + "learning_rate": 1e-06, + "loss": 0.2262, + "mean_token_accuracy": 0.9144660234451294, + "num_tokens": 893584550.0, + "step": 23023 + }, + { + "epoch": 2.928889454267905, + "grad_norm": 2.7214767932891846, + "learning_rate": 1e-06, + "loss": 0.2402, + "mean_token_accuracy": 0.9106447696685791, + "num_tokens": 893625101.0, + "step": 23024 + }, + { + "epoch": 2.9290166645464955, + "grad_norm": 2.8646655082702637, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9037272334098816, + "num_tokens": 893663096.0, + "step": 23025 + }, + { + "epoch": 2.929143874825086, + "grad_norm": 2.8945789337158203, + "learning_rate": 1e-06, + "loss": 0.3125, + "mean_token_accuracy": 0.885908842086792, + "num_tokens": 893705719.0, + "step": 23026 + }, + { + "epoch": 2.9292710851036765, + "grad_norm": 2.9399430751800537, + "learning_rate": 1e-06, + "loss": 0.2949, + "mean_token_accuracy": 0.892376184463501, + "num_tokens": 893747835.0, + "step": 23027 + }, + { + "epoch": 2.929398295382267, + "grad_norm": 3.1078295707702637, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9046531915664673, + "num_tokens": 893781558.0, + "step": 23028 + }, + { + "epoch": 2.929525505660857, + "grad_norm": 3.0116124153137207, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9049065113067627, + "num_tokens": 893818337.0, + "step": 23029 + }, + { + "epoch": 2.929652715939448, + "grad_norm": 2.9394631385803223, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.908515214920044, + "num_tokens": 893851844.0, + "step": 23030 + }, + { + "epoch": 2.929779926218038, + "grad_norm": 2.655921459197998, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.8987991809844971, + "num_tokens": 893899695.0, + "step": 23031 + }, + { + "epoch": 2.929907136496629, + "grad_norm": 3.0588178634643555, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9044114351272583, + "num_tokens": 893934747.0, + "step": 23032 + }, + { + "epoch": 2.9300343467752192, + "grad_norm": 3.050145149230957, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.9028113484382629, + "num_tokens": 893972617.0, + "step": 23033 + }, + { + "epoch": 2.93016155705381, + "grad_norm": 2.729351282119751, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9058068990707397, + "num_tokens": 894014266.0, + "step": 23034 + }, + { + "epoch": 2.9302887673324003, + "grad_norm": 2.8428359031677246, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8960723876953125, + "num_tokens": 894056250.0, + "step": 23035 + }, + { + "epoch": 2.930415977610991, + "grad_norm": 2.9637725353240967, + "learning_rate": 1e-06, + "loss": 0.2694, + "mean_token_accuracy": 0.8988395929336548, + "num_tokens": 894092575.0, + "step": 23036 + }, + { + "epoch": 2.9305431878895813, + "grad_norm": 3.0367414951324463, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.9013102650642395, + "num_tokens": 894131872.0, + "step": 23037 + }, + { + "epoch": 2.930670398168172, + "grad_norm": 2.9633610248565674, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.8970249891281128, + "num_tokens": 894173992.0, + "step": 23038 + }, + { + "epoch": 2.9307976084467624, + "grad_norm": 2.892744302749634, + "learning_rate": 1e-06, + "loss": 0.2725, + "mean_token_accuracy": 0.8992871046066284, + "num_tokens": 894213342.0, + "step": 23039 + }, + { + "epoch": 2.930924818725353, + "grad_norm": 3.015754222869873, + "learning_rate": 1e-06, + "loss": 0.2358, + "mean_token_accuracy": 0.9113690853118896, + "num_tokens": 894246309.0, + "step": 23040 + }, + { + "epoch": 2.9310520290039435, + "grad_norm": 2.881549835205078, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9027494788169861, + "num_tokens": 894284342.0, + "step": 23041 + }, + { + "epoch": 2.931179239282534, + "grad_norm": 2.8420028686523438, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.9029775261878967, + "num_tokens": 894324368.0, + "step": 23042 + }, + { + "epoch": 2.9313064495611245, + "grad_norm": 2.9734222888946533, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9093500375747681, + "num_tokens": 894361981.0, + "step": 23043 + }, + { + "epoch": 2.931433659839715, + "grad_norm": 2.9275059700012207, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8947474956512451, + "num_tokens": 894399838.0, + "step": 23044 + }, + { + "epoch": 2.9315608701183056, + "grad_norm": 2.933431386947632, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9050174355506897, + "num_tokens": 894436981.0, + "step": 23045 + }, + { + "epoch": 2.931688080396896, + "grad_norm": 3.6471941471099854, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8929029107093811, + "num_tokens": 894468528.0, + "step": 23046 + }, + { + "epoch": 2.9318152906754866, + "grad_norm": 2.999913454055786, + "learning_rate": 1e-06, + "loss": 0.2631, + "mean_token_accuracy": 0.9023421406745911, + "num_tokens": 894504726.0, + "step": 23047 + }, + { + "epoch": 2.931942500954077, + "grad_norm": 3.0531809329986572, + "learning_rate": 1e-06, + "loss": 0.2622, + "mean_token_accuracy": 0.9021733999252319, + "num_tokens": 894541804.0, + "step": 23048 + }, + { + "epoch": 2.9320697112326677, + "grad_norm": 2.9615368843078613, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.9031404852867126, + "num_tokens": 894575073.0, + "step": 23049 + }, + { + "epoch": 2.932196921511258, + "grad_norm": 3.2067201137542725, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9027535915374756, + "num_tokens": 894607240.0, + "step": 23050 + }, + { + "epoch": 2.9323241317898487, + "grad_norm": 2.5487003326416016, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9111009836196899, + "num_tokens": 894649216.0, + "step": 23051 + }, + { + "epoch": 2.9324513420684393, + "grad_norm": 2.7631874084472656, + "learning_rate": 1e-06, + "loss": 0.2689, + "mean_token_accuracy": 0.9015191793441772, + "num_tokens": 894692422.0, + "step": 23052 + }, + { + "epoch": 2.93257855234703, + "grad_norm": 2.8147788047790527, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.896859884262085, + "num_tokens": 894732479.0, + "step": 23053 + }, + { + "epoch": 2.93270576262562, + "grad_norm": 3.01808762550354, + "learning_rate": 1e-06, + "loss": 0.2892, + "mean_token_accuracy": 0.8917106986045837, + "num_tokens": 894772327.0, + "step": 23054 + }, + { + "epoch": 2.932832972904211, + "grad_norm": 2.743159532546997, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.901195764541626, + "num_tokens": 894816825.0, + "step": 23055 + }, + { + "epoch": 2.932960183182801, + "grad_norm": 2.994982957839966, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.9007717370986938, + "num_tokens": 894854706.0, + "step": 23056 + }, + { + "epoch": 2.933087393461392, + "grad_norm": 2.703226327896118, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9095558524131775, + "num_tokens": 894895690.0, + "step": 23057 + }, + { + "epoch": 2.933214603739982, + "grad_norm": 2.5516984462738037, + "learning_rate": 1e-06, + "loss": 0.2299, + "mean_token_accuracy": 0.9167467355728149, + "num_tokens": 894934841.0, + "step": 23058 + }, + { + "epoch": 2.933341814018573, + "grad_norm": 2.9627864360809326, + "learning_rate": 1e-06, + "loss": 0.2607, + "mean_token_accuracy": 0.9039013981819153, + "num_tokens": 894974068.0, + "step": 23059 + }, + { + "epoch": 2.933469024297163, + "grad_norm": 2.6374683380126953, + "learning_rate": 1e-06, + "loss": 0.2384, + "mean_token_accuracy": 0.9100275039672852, + "num_tokens": 895016500.0, + "step": 23060 + }, + { + "epoch": 2.9335962345757536, + "grad_norm": 2.857015609741211, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9020411968231201, + "num_tokens": 895058317.0, + "step": 23061 + }, + { + "epoch": 2.933723444854344, + "grad_norm": 2.7067949771881104, + "learning_rate": 1e-06, + "loss": 0.2699, + "mean_token_accuracy": 0.9017099738121033, + "num_tokens": 895103202.0, + "step": 23062 + }, + { + "epoch": 2.9338506551329346, + "grad_norm": 2.7842159271240234, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9082506895065308, + "num_tokens": 895143291.0, + "step": 23063 + }, + { + "epoch": 2.933977865411525, + "grad_norm": 2.688654661178589, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.9069772958755493, + "num_tokens": 895182652.0, + "step": 23064 + }, + { + "epoch": 2.9341050756901157, + "grad_norm": 2.7338457107543945, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9087185859680176, + "num_tokens": 895223107.0, + "step": 23065 + }, + { + "epoch": 2.934232285968706, + "grad_norm": 2.95074200630188, + "learning_rate": 1e-06, + "loss": 0.2344, + "mean_token_accuracy": 0.911662220954895, + "num_tokens": 895257527.0, + "step": 23066 + }, + { + "epoch": 2.9343594962472968, + "grad_norm": 2.848109006881714, + "learning_rate": 1e-06, + "loss": 0.2519, + "mean_token_accuracy": 0.9075658917427063, + "num_tokens": 895294642.0, + "step": 23067 + }, + { + "epoch": 2.9344867065258873, + "grad_norm": 2.761129856109619, + "learning_rate": 1e-06, + "loss": 0.2455, + "mean_token_accuracy": 0.9098485112190247, + "num_tokens": 895336674.0, + "step": 23068 + }, + { + "epoch": 2.934613916804478, + "grad_norm": 3.290595769882202, + "learning_rate": 1e-06, + "loss": 0.2912, + "mean_token_accuracy": 0.8948891162872314, + "num_tokens": 895370829.0, + "step": 23069 + }, + { + "epoch": 2.9347411270830683, + "grad_norm": 2.7155957221984863, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9054521322250366, + "num_tokens": 895411886.0, + "step": 23070 + }, + { + "epoch": 2.934868337361659, + "grad_norm": 2.456707239151001, + "learning_rate": 1e-06, + "loss": 0.2196, + "mean_token_accuracy": 0.9178633689880371, + "num_tokens": 895457076.0, + "step": 23071 + }, + { + "epoch": 2.9349955476402494, + "grad_norm": 2.8029158115386963, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9040156602859497, + "num_tokens": 895498557.0, + "step": 23072 + }, + { + "epoch": 2.93512275791884, + "grad_norm": 2.6792917251586914, + "learning_rate": 1e-06, + "loss": 0.2531, + "mean_token_accuracy": 0.9064092636108398, + "num_tokens": 895541258.0, + "step": 23073 + }, + { + "epoch": 2.9352499681974304, + "grad_norm": 2.8955788612365723, + "learning_rate": 1e-06, + "loss": 0.2497, + "mean_token_accuracy": 0.9062731862068176, + "num_tokens": 895580314.0, + "step": 23074 + }, + { + "epoch": 2.935377178476021, + "grad_norm": 2.9962854385375977, + "learning_rate": 1e-06, + "loss": 0.3015, + "mean_token_accuracy": 0.8885754942893982, + "num_tokens": 895621576.0, + "step": 23075 + }, + { + "epoch": 2.9355043887546115, + "grad_norm": 2.619304656982422, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.8987730145454407, + "num_tokens": 895669636.0, + "step": 23076 + }, + { + "epoch": 2.935631599033202, + "grad_norm": 3.0658652782440186, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.904937744140625, + "num_tokens": 895708562.0, + "step": 23077 + }, + { + "epoch": 2.9357588093117926, + "grad_norm": 3.080487012863159, + "learning_rate": 1e-06, + "loss": 0.2919, + "mean_token_accuracy": 0.8947746753692627, + "num_tokens": 895747052.0, + "step": 23078 + }, + { + "epoch": 2.9358860195903826, + "grad_norm": 3.061739444732666, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8973973989486694, + "num_tokens": 895787367.0, + "step": 23079 + }, + { + "epoch": 2.9360132298689736, + "grad_norm": 2.908207893371582, + "learning_rate": 1e-06, + "loss": 0.2478, + "mean_token_accuracy": 0.9060592651367188, + "num_tokens": 895823418.0, + "step": 23080 + }, + { + "epoch": 2.9361404401475637, + "grad_norm": 2.9810307025909424, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8998485803604126, + "num_tokens": 895859625.0, + "step": 23081 + }, + { + "epoch": 2.9362676504261547, + "grad_norm": 2.9154512882232666, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.910137414932251, + "num_tokens": 895896337.0, + "step": 23082 + }, + { + "epoch": 2.9363948607047448, + "grad_norm": 3.1379854679107666, + "learning_rate": 1e-06, + "loss": 0.2717, + "mean_token_accuracy": 0.8991808891296387, + "num_tokens": 895929797.0, + "step": 23083 + }, + { + "epoch": 2.9365220709833357, + "grad_norm": 2.6090145111083984, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.911657989025116, + "num_tokens": 895974948.0, + "step": 23084 + }, + { + "epoch": 2.936649281261926, + "grad_norm": 2.9944565296173096, + "learning_rate": 1e-06, + "loss": 0.282, + "mean_token_accuracy": 0.8978357315063477, + "num_tokens": 896015382.0, + "step": 23085 + }, + { + "epoch": 2.9367764915405163, + "grad_norm": 2.631739377975464, + "learning_rate": 1e-06, + "loss": 0.2257, + "mean_token_accuracy": 0.9150862097740173, + "num_tokens": 896054267.0, + "step": 23086 + }, + { + "epoch": 2.936903701819107, + "grad_norm": 2.9561376571655273, + "learning_rate": 1e-06, + "loss": 0.271, + "mean_token_accuracy": 0.9037778377532959, + "num_tokens": 896091149.0, + "step": 23087 + }, + { + "epoch": 2.9370309120976974, + "grad_norm": 2.6911609172821045, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9019859433174133, + "num_tokens": 896135218.0, + "step": 23088 + }, + { + "epoch": 2.937158122376288, + "grad_norm": 2.931933641433716, + "learning_rate": 1e-06, + "loss": 0.2499, + "mean_token_accuracy": 0.9075609445571899, + "num_tokens": 896169392.0, + "step": 23089 + }, + { + "epoch": 2.9372853326548785, + "grad_norm": 3.364041566848755, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.90129154920578, + "num_tokens": 896199346.0, + "step": 23090 + }, + { + "epoch": 2.937412542933469, + "grad_norm": 2.7840514183044434, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9072766304016113, + "num_tokens": 896241325.0, + "step": 23091 + }, + { + "epoch": 2.9375397532120595, + "grad_norm": 2.812917470932007, + "learning_rate": 1e-06, + "loss": 0.2449, + "mean_token_accuracy": 0.9087178111076355, + "num_tokens": 896281871.0, + "step": 23092 + }, + { + "epoch": 2.93766696349065, + "grad_norm": 2.871819496154785, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.901946485042572, + "num_tokens": 896319639.0, + "step": 23093 + }, + { + "epoch": 2.9377941737692406, + "grad_norm": 2.7317965030670166, + "learning_rate": 1e-06, + "loss": 0.2154, + "mean_token_accuracy": 0.920617938041687, + "num_tokens": 896358255.0, + "step": 23094 + }, + { + "epoch": 2.937921384047831, + "grad_norm": 2.712263584136963, + "learning_rate": 1e-06, + "loss": 0.2758, + "mean_token_accuracy": 0.8983260989189148, + "num_tokens": 896401849.0, + "step": 23095 + }, + { + "epoch": 2.9380485943264216, + "grad_norm": 2.738208770751953, + "learning_rate": 1e-06, + "loss": 0.2653, + "mean_token_accuracy": 0.9032632112503052, + "num_tokens": 896442657.0, + "step": 23096 + }, + { + "epoch": 2.938175804605012, + "grad_norm": 2.9138946533203125, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9101515412330627, + "num_tokens": 896480709.0, + "step": 23097 + }, + { + "epoch": 2.9383030148836027, + "grad_norm": 2.5473411083221436, + "learning_rate": 1e-06, + "loss": 0.2362, + "mean_token_accuracy": 0.9117496609687805, + "num_tokens": 896523577.0, + "step": 23098 + }, + { + "epoch": 2.938430225162193, + "grad_norm": 2.8717381954193115, + "learning_rate": 1e-06, + "loss": 0.2465, + "mean_token_accuracy": 0.910443902015686, + "num_tokens": 896558199.0, + "step": 23099 + }, + { + "epoch": 2.9385574354407837, + "grad_norm": 2.814422130584717, + "learning_rate": 1e-06, + "loss": 0.3138, + "mean_token_accuracy": 0.8838043808937073, + "num_tokens": 896602247.0, + "step": 23100 + }, + { + "epoch": 2.9386846457193743, + "grad_norm": 3.054478883743286, + "learning_rate": 1e-06, + "loss": 0.2832, + "mean_token_accuracy": 0.8969007730484009, + "num_tokens": 896638657.0, + "step": 23101 + }, + { + "epoch": 2.938811855997965, + "grad_norm": 2.8489391803741455, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9075313806533813, + "num_tokens": 896680432.0, + "step": 23102 + }, + { + "epoch": 2.9389390662765553, + "grad_norm": 3.055209159851074, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8860475420951843, + "num_tokens": 896721391.0, + "step": 23103 + }, + { + "epoch": 2.9390662765551454, + "grad_norm": 2.7865774631500244, + "learning_rate": 1e-06, + "loss": 0.2243, + "mean_token_accuracy": 0.9164804816246033, + "num_tokens": 896758695.0, + "step": 23104 + }, + { + "epoch": 2.9391934868337364, + "grad_norm": 2.7917585372924805, + "learning_rate": 1e-06, + "loss": 0.2251, + "mean_token_accuracy": 0.9149734973907471, + "num_tokens": 896797220.0, + "step": 23105 + }, + { + "epoch": 2.9393206971123265, + "grad_norm": 2.6002213954925537, + "learning_rate": 1e-06, + "loss": 0.2351, + "mean_token_accuracy": 0.9132241606712341, + "num_tokens": 896841409.0, + "step": 23106 + }, + { + "epoch": 2.9394479073909174, + "grad_norm": 2.7681713104248047, + "learning_rate": 1e-06, + "loss": 0.2367, + "mean_token_accuracy": 0.9077746868133545, + "num_tokens": 896880101.0, + "step": 23107 + }, + { + "epoch": 2.9395751176695075, + "grad_norm": 3.1144890785217285, + "learning_rate": 1e-06, + "loss": 0.2721, + "mean_token_accuracy": 0.8984407186508179, + "num_tokens": 896913763.0, + "step": 23108 + }, + { + "epoch": 2.939702327948098, + "grad_norm": 2.86614727973938, + "learning_rate": 1e-06, + "loss": 0.265, + "mean_token_accuracy": 0.9032071232795715, + "num_tokens": 896951840.0, + "step": 23109 + }, + { + "epoch": 2.9398295382266886, + "grad_norm": 2.967262029647827, + "learning_rate": 1e-06, + "loss": 0.2798, + "mean_token_accuracy": 0.8944395780563354, + "num_tokens": 896993664.0, + "step": 23110 + }, + { + "epoch": 2.939956748505279, + "grad_norm": 2.602231025695801, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9122813940048218, + "num_tokens": 897035408.0, + "step": 23111 + }, + { + "epoch": 2.9400839587838696, + "grad_norm": 2.910977363586426, + "learning_rate": 1e-06, + "loss": 0.2309, + "mean_token_accuracy": 0.9125628471374512, + "num_tokens": 897071158.0, + "step": 23112 + }, + { + "epoch": 2.94021116906246, + "grad_norm": 2.825965404510498, + "learning_rate": 1e-06, + "loss": 0.2785, + "mean_token_accuracy": 0.8980816602706909, + "num_tokens": 897111944.0, + "step": 23113 + }, + { + "epoch": 2.9403383793410507, + "grad_norm": 2.7884809970855713, + "learning_rate": 1e-06, + "loss": 0.2702, + "mean_token_accuracy": 0.9012966156005859, + "num_tokens": 897156601.0, + "step": 23114 + }, + { + "epoch": 2.940465589619641, + "grad_norm": 3.1458497047424316, + "learning_rate": 1e-06, + "loss": 0.3316, + "mean_token_accuracy": 0.879349946975708, + "num_tokens": 897198185.0, + "step": 23115 + }, + { + "epoch": 2.9405927998982317, + "grad_norm": 2.8141868114471436, + "learning_rate": 1e-06, + "loss": 0.3024, + "mean_token_accuracy": 0.8851262927055359, + "num_tokens": 897241211.0, + "step": 23116 + }, + { + "epoch": 2.9407200101768223, + "grad_norm": 3.130714178085327, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8974518775939941, + "num_tokens": 897279580.0, + "step": 23117 + }, + { + "epoch": 2.940847220455413, + "grad_norm": 2.9164323806762695, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.9116206765174866, + "num_tokens": 897316559.0, + "step": 23118 + }, + { + "epoch": 2.9409744307340033, + "grad_norm": 2.8986034393310547, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.908155083656311, + "num_tokens": 897353427.0, + "step": 23119 + }, + { + "epoch": 2.941101641012594, + "grad_norm": 3.100496768951416, + "learning_rate": 1e-06, + "loss": 0.3039, + "mean_token_accuracy": 0.8908379077911377, + "num_tokens": 897389836.0, + "step": 23120 + }, + { + "epoch": 2.9412288512911844, + "grad_norm": 3.057194709777832, + "learning_rate": 1e-06, + "loss": 0.251, + "mean_token_accuracy": 0.905775785446167, + "num_tokens": 897421485.0, + "step": 23121 + }, + { + "epoch": 2.941356061569775, + "grad_norm": 2.8518691062927246, + "learning_rate": 1e-06, + "loss": 0.2783, + "mean_token_accuracy": 0.8983196020126343, + "num_tokens": 897461248.0, + "step": 23122 + }, + { + "epoch": 2.9414832718483654, + "grad_norm": 2.74969220161438, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9115456342697144, + "num_tokens": 897498935.0, + "step": 23123 + }, + { + "epoch": 2.941610482126956, + "grad_norm": 2.910665512084961, + "learning_rate": 1e-06, + "loss": 0.2486, + "mean_token_accuracy": 0.906907856464386, + "num_tokens": 897533797.0, + "step": 23124 + }, + { + "epoch": 2.9417376924055465, + "grad_norm": 2.9069457054138184, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9048361778259277, + "num_tokens": 897570597.0, + "step": 23125 + }, + { + "epoch": 2.941864902684137, + "grad_norm": 2.9696919918060303, + "learning_rate": 1e-06, + "loss": 0.2771, + "mean_token_accuracy": 0.9000564813613892, + "num_tokens": 897611314.0, + "step": 23126 + }, + { + "epoch": 2.941992112962727, + "grad_norm": 2.7994015216827393, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8937540054321289, + "num_tokens": 897655984.0, + "step": 23127 + }, + { + "epoch": 2.942119323241318, + "grad_norm": 2.993159294128418, + "learning_rate": 1e-06, + "loss": 0.2713, + "mean_token_accuracy": 0.9006980061531067, + "num_tokens": 897696219.0, + "step": 23128 + }, + { + "epoch": 2.942246533519908, + "grad_norm": 2.9099349975585938, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.9005225300788879, + "num_tokens": 897735596.0, + "step": 23129 + }, + { + "epoch": 2.942373743798499, + "grad_norm": 2.888632297515869, + "learning_rate": 1e-06, + "loss": 0.2581, + "mean_token_accuracy": 0.9043383598327637, + "num_tokens": 897775023.0, + "step": 23130 + }, + { + "epoch": 2.9425009540770892, + "grad_norm": 2.80192232131958, + "learning_rate": 1e-06, + "loss": 0.276, + "mean_token_accuracy": 0.8969318866729736, + "num_tokens": 897815053.0, + "step": 23131 + }, + { + "epoch": 2.94262816435568, + "grad_norm": 3.1074676513671875, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8967949151992798, + "num_tokens": 897850821.0, + "step": 23132 + }, + { + "epoch": 2.9427553746342703, + "grad_norm": 2.5430524349212646, + "learning_rate": 1e-06, + "loss": 0.2393, + "mean_token_accuracy": 0.9106268286705017, + "num_tokens": 897894482.0, + "step": 23133 + }, + { + "epoch": 2.942882584912861, + "grad_norm": 2.6040496826171875, + "learning_rate": 1e-06, + "loss": 0.2424, + "mean_token_accuracy": 0.9105643033981323, + "num_tokens": 897934098.0, + "step": 23134 + }, + { + "epoch": 2.9430097951914513, + "grad_norm": 2.682260751724243, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9046973586082458, + "num_tokens": 897977569.0, + "step": 23135 + }, + { + "epoch": 2.943137005470042, + "grad_norm": 2.6959221363067627, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9073845148086548, + "num_tokens": 898021085.0, + "step": 23136 + }, + { + "epoch": 2.9432642157486324, + "grad_norm": 3.2157487869262695, + "learning_rate": 1e-06, + "loss": 0.3004, + "mean_token_accuracy": 0.8919433355331421, + "num_tokens": 898057522.0, + "step": 23137 + }, + { + "epoch": 2.943391426027223, + "grad_norm": 2.76651930809021, + "learning_rate": 1e-06, + "loss": 0.2824, + "mean_token_accuracy": 0.8960169553756714, + "num_tokens": 898099531.0, + "step": 23138 + }, + { + "epoch": 2.9435186363058135, + "grad_norm": 2.9069976806640625, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9074761867523193, + "num_tokens": 898136154.0, + "step": 23139 + }, + { + "epoch": 2.943645846584404, + "grad_norm": 2.868856906890869, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.8977242708206177, + "num_tokens": 898175454.0, + "step": 23140 + }, + { + "epoch": 2.9437730568629945, + "grad_norm": 2.9849767684936523, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9098949432373047, + "num_tokens": 898213701.0, + "step": 23141 + }, + { + "epoch": 2.943900267141585, + "grad_norm": 3.1457011699676514, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.8992164731025696, + "num_tokens": 898249165.0, + "step": 23142 + }, + { + "epoch": 2.9440274774201756, + "grad_norm": 2.7761878967285156, + "learning_rate": 1e-06, + "loss": 0.2467, + "mean_token_accuracy": 0.9105128049850464, + "num_tokens": 898287550.0, + "step": 23143 + }, + { + "epoch": 2.944154687698766, + "grad_norm": 2.9974524974823, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.9063566327095032, + "num_tokens": 898322484.0, + "step": 23144 + }, + { + "epoch": 2.9442818979773566, + "grad_norm": 2.846193790435791, + "learning_rate": 1e-06, + "loss": 0.255, + "mean_token_accuracy": 0.9083389639854431, + "num_tokens": 898361953.0, + "step": 23145 + }, + { + "epoch": 2.944409108255947, + "grad_norm": 2.99481201171875, + "learning_rate": 1e-06, + "loss": 0.2748, + "mean_token_accuracy": 0.9020282030105591, + "num_tokens": 898400175.0, + "step": 23146 + }, + { + "epoch": 2.9445363185345377, + "grad_norm": 2.7209856510162354, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.905359148979187, + "num_tokens": 898442504.0, + "step": 23147 + }, + { + "epoch": 2.944663528813128, + "grad_norm": 2.654204845428467, + "learning_rate": 1e-06, + "loss": 0.2322, + "mean_token_accuracy": 0.9142666459083557, + "num_tokens": 898483291.0, + "step": 23148 + }, + { + "epoch": 2.9447907390917187, + "grad_norm": 2.893308162689209, + "learning_rate": 1e-06, + "loss": 0.2532, + "mean_token_accuracy": 0.9055216312408447, + "num_tokens": 898520731.0, + "step": 23149 + }, + { + "epoch": 2.9449179493703093, + "grad_norm": 2.742655038833618, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8974780440330505, + "num_tokens": 898568521.0, + "step": 23150 + }, + { + "epoch": 2.9450451596489, + "grad_norm": 2.9445760250091553, + "learning_rate": 1e-06, + "loss": 0.2762, + "mean_token_accuracy": 0.8993893265724182, + "num_tokens": 898605134.0, + "step": 23151 + }, + { + "epoch": 2.94517236992749, + "grad_norm": 2.5816073417663574, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9044365882873535, + "num_tokens": 898651005.0, + "step": 23152 + }, + { + "epoch": 2.945299580206081, + "grad_norm": 3.101076126098633, + "learning_rate": 1e-06, + "loss": 0.2328, + "mean_token_accuracy": 0.9109876155853271, + "num_tokens": 898683358.0, + "step": 23153 + }, + { + "epoch": 2.945426790484671, + "grad_norm": 2.8438823223114014, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9092574119567871, + "num_tokens": 898721147.0, + "step": 23154 + }, + { + "epoch": 2.945554000763262, + "grad_norm": 2.9131784439086914, + "learning_rate": 1e-06, + "loss": 0.2804, + "mean_token_accuracy": 0.8973056674003601, + "num_tokens": 898759385.0, + "step": 23155 + }, + { + "epoch": 2.945681211041852, + "grad_norm": 2.795210838317871, + "learning_rate": 1e-06, + "loss": 0.2568, + "mean_token_accuracy": 0.9070727825164795, + "num_tokens": 898798374.0, + "step": 23156 + }, + { + "epoch": 2.945808421320443, + "grad_norm": 3.2895667552948, + "learning_rate": 1e-06, + "loss": 0.2651, + "mean_token_accuracy": 0.9025992155075073, + "num_tokens": 898830489.0, + "step": 23157 + }, + { + "epoch": 2.945935631599033, + "grad_norm": 2.9848506450653076, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9081051349639893, + "num_tokens": 898864967.0, + "step": 23158 + }, + { + "epoch": 2.9460628418776236, + "grad_norm": 2.8488993644714355, + "learning_rate": 1e-06, + "loss": 0.2654, + "mean_token_accuracy": 0.9012980461120605, + "num_tokens": 898906271.0, + "step": 23159 + }, + { + "epoch": 2.946190052156214, + "grad_norm": 3.102181911468506, + "learning_rate": 1e-06, + "loss": 0.2456, + "mean_token_accuracy": 0.9082401990890503, + "num_tokens": 898938953.0, + "step": 23160 + }, + { + "epoch": 2.9463172624348046, + "grad_norm": 2.908623456954956, + "learning_rate": 1e-06, + "loss": 0.2539, + "mean_token_accuracy": 0.9047585725784302, + "num_tokens": 898979502.0, + "step": 23161 + }, + { + "epoch": 2.946444472713395, + "grad_norm": 2.6391921043395996, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9014843106269836, + "num_tokens": 899023929.0, + "step": 23162 + }, + { + "epoch": 2.9465716829919857, + "grad_norm": 2.8770241737365723, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8954848051071167, + "num_tokens": 899062572.0, + "step": 23163 + }, + { + "epoch": 2.946698893270576, + "grad_norm": 2.7942209243774414, + "learning_rate": 1e-06, + "loss": 0.2882, + "mean_token_accuracy": 0.8931941986083984, + "num_tokens": 899107612.0, + "step": 23164 + }, + { + "epoch": 2.9468261035491667, + "grad_norm": 3.0492775440216064, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.902097761631012, + "num_tokens": 899141468.0, + "step": 23165 + }, + { + "epoch": 2.9469533138277573, + "grad_norm": 2.969515800476074, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9113311767578125, + "num_tokens": 899174462.0, + "step": 23166 + }, + { + "epoch": 2.947080524106348, + "grad_norm": 3.072998046875, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.902206301689148, + "num_tokens": 899213388.0, + "step": 23167 + }, + { + "epoch": 2.9472077343849383, + "grad_norm": 2.9081919193267822, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9074679017066956, + "num_tokens": 899251077.0, + "step": 23168 + }, + { + "epoch": 2.947334944663529, + "grad_norm": 3.0800509452819824, + "learning_rate": 1e-06, + "loss": 0.2992, + "mean_token_accuracy": 0.8898362517356873, + "num_tokens": 899288571.0, + "step": 23169 + }, + { + "epoch": 2.9474621549421194, + "grad_norm": 3.417764663696289, + "learning_rate": 1e-06, + "loss": 0.2688, + "mean_token_accuracy": 0.8977211713790894, + "num_tokens": 899320424.0, + "step": 23170 + }, + { + "epoch": 2.94758936522071, + "grad_norm": 3.0029380321502686, + "learning_rate": 1e-06, + "loss": 0.2701, + "mean_token_accuracy": 0.9001133441925049, + "num_tokens": 899354473.0, + "step": 23171 + }, + { + "epoch": 2.9477165754993004, + "grad_norm": 2.7762091159820557, + "learning_rate": 1e-06, + "loss": 0.2429, + "mean_token_accuracy": 0.9089369177818298, + "num_tokens": 899393852.0, + "step": 23172 + }, + { + "epoch": 2.947843785777891, + "grad_norm": 2.975715160369873, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8976311087608337, + "num_tokens": 899432037.0, + "step": 23173 + }, + { + "epoch": 2.9479709960564815, + "grad_norm": 2.574563980102539, + "learning_rate": 1e-06, + "loss": 0.2673, + "mean_token_accuracy": 0.9040446877479553, + "num_tokens": 899477229.0, + "step": 23174 + }, + { + "epoch": 2.948098206335072, + "grad_norm": 2.6924777030944824, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.913240909576416, + "num_tokens": 899517668.0, + "step": 23175 + }, + { + "epoch": 2.9482254166136626, + "grad_norm": 2.8891162872314453, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.8989181518554688, + "num_tokens": 899557466.0, + "step": 23176 + }, + { + "epoch": 2.9483526268922526, + "grad_norm": 2.871574878692627, + "learning_rate": 1e-06, + "loss": 0.2916, + "mean_token_accuracy": 0.8928677439689636, + "num_tokens": 899598795.0, + "step": 23177 + }, + { + "epoch": 2.9484798371708436, + "grad_norm": 2.8856723308563232, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9037623405456543, + "num_tokens": 899635688.0, + "step": 23178 + }, + { + "epoch": 2.9486070474494337, + "grad_norm": 2.7448811531066895, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9089492559432983, + "num_tokens": 899670504.0, + "step": 23179 + }, + { + "epoch": 2.9487342577280247, + "grad_norm": 2.78778338432312, + "learning_rate": 1e-06, + "loss": 0.2387, + "mean_token_accuracy": 0.9103176593780518, + "num_tokens": 899710052.0, + "step": 23180 + }, + { + "epoch": 2.9488614680066147, + "grad_norm": 3.016247034072876, + "learning_rate": 1e-06, + "loss": 0.2602, + "mean_token_accuracy": 0.9030401706695557, + "num_tokens": 899746256.0, + "step": 23181 + }, + { + "epoch": 2.9489886782852053, + "grad_norm": 2.7511558532714844, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9109615087509155, + "num_tokens": 899783851.0, + "step": 23182 + }, + { + "epoch": 2.949115888563796, + "grad_norm": 2.7144927978515625, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9114089012145996, + "num_tokens": 899825125.0, + "step": 23183 + }, + { + "epoch": 2.9492430988423863, + "grad_norm": 2.613842725753784, + "learning_rate": 1e-06, + "loss": 0.2351, + "mean_token_accuracy": 0.912384033203125, + "num_tokens": 899868876.0, + "step": 23184 + }, + { + "epoch": 2.949370309120977, + "grad_norm": 2.8405168056488037, + "learning_rate": 1e-06, + "loss": 0.2388, + "mean_token_accuracy": 0.9104838371276855, + "num_tokens": 899906356.0, + "step": 23185 + }, + { + "epoch": 2.9494975193995674, + "grad_norm": 3.209970235824585, + "learning_rate": 1e-06, + "loss": 0.2496, + "mean_token_accuracy": 0.9072340726852417, + "num_tokens": 899942163.0, + "step": 23186 + }, + { + "epoch": 2.949624729678158, + "grad_norm": 3.0366098880767822, + "learning_rate": 1e-06, + "loss": 0.2393, + "mean_token_accuracy": 0.9103549718856812, + "num_tokens": 899978400.0, + "step": 23187 + }, + { + "epoch": 2.9497519399567484, + "grad_norm": 2.8569259643554688, + "learning_rate": 1e-06, + "loss": 0.2687, + "mean_token_accuracy": 0.9068775177001953, + "num_tokens": 900020445.0, + "step": 23188 + }, + { + "epoch": 2.949879150235339, + "grad_norm": 2.6301395893096924, + "learning_rate": 1e-06, + "loss": 0.2248, + "mean_token_accuracy": 0.9154795408248901, + "num_tokens": 900062558.0, + "step": 23189 + }, + { + "epoch": 2.9500063605139295, + "grad_norm": 3.0314934253692627, + "learning_rate": 1e-06, + "loss": 0.2557, + "mean_token_accuracy": 0.904091477394104, + "num_tokens": 900098272.0, + "step": 23190 + }, + { + "epoch": 2.95013357079252, + "grad_norm": 2.828171491622925, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.8937422037124634, + "num_tokens": 900140666.0, + "step": 23191 + }, + { + "epoch": 2.9502607810711106, + "grad_norm": 2.8300042152404785, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9093219041824341, + "num_tokens": 900178668.0, + "step": 23192 + }, + { + "epoch": 2.950387991349701, + "grad_norm": 3.161431074142456, + "learning_rate": 1e-06, + "loss": 0.281, + "mean_token_accuracy": 0.8991994857788086, + "num_tokens": 900214197.0, + "step": 23193 + }, + { + "epoch": 2.9505152016282916, + "grad_norm": 2.791299819946289, + "learning_rate": 1e-06, + "loss": 0.275, + "mean_token_accuracy": 0.8971511125564575, + "num_tokens": 900258225.0, + "step": 23194 + }, + { + "epoch": 2.950642411906882, + "grad_norm": 2.81367564201355, + "learning_rate": 1e-06, + "loss": 0.2589, + "mean_token_accuracy": 0.9043095111846924, + "num_tokens": 900294797.0, + "step": 23195 + }, + { + "epoch": 2.9507696221854727, + "grad_norm": 2.6154396533966064, + "learning_rate": 1e-06, + "loss": 0.2306, + "mean_token_accuracy": 0.9139949083328247, + "num_tokens": 900336542.0, + "step": 23196 + }, + { + "epoch": 2.950896832464063, + "grad_norm": 3.187767505645752, + "learning_rate": 1e-06, + "loss": 0.3002, + "mean_token_accuracy": 0.891651451587677, + "num_tokens": 900374771.0, + "step": 23197 + }, + { + "epoch": 2.9510240427426537, + "grad_norm": 3.00848650932312, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8935120105743408, + "num_tokens": 900414384.0, + "step": 23198 + }, + { + "epoch": 2.9511512530212443, + "grad_norm": 3.0386390686035156, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9052579998970032, + "num_tokens": 900449832.0, + "step": 23199 + }, + { + "epoch": 2.951278463299835, + "grad_norm": 2.9188873767852783, + "learning_rate": 1e-06, + "loss": 0.2874, + "mean_token_accuracy": 0.8943177461624146, + "num_tokens": 900495410.0, + "step": 23200 + }, + { + "epoch": 2.9514056735784253, + "grad_norm": 2.807103395462036, + "learning_rate": 1e-06, + "loss": 0.2515, + "mean_token_accuracy": 0.9062069058418274, + "num_tokens": 900536390.0, + "step": 23201 + }, + { + "epoch": 2.9515328838570154, + "grad_norm": 3.02664852142334, + "learning_rate": 1e-06, + "loss": 0.2799, + "mean_token_accuracy": 0.8970885276794434, + "num_tokens": 900571560.0, + "step": 23202 + }, + { + "epoch": 2.9516600941356064, + "grad_norm": 2.5919675827026367, + "learning_rate": 1e-06, + "loss": 0.227, + "mean_token_accuracy": 0.9149522185325623, + "num_tokens": 900613423.0, + "step": 23203 + }, + { + "epoch": 2.9517873044141965, + "grad_norm": 2.9188954830169678, + "learning_rate": 1e-06, + "loss": 0.2231, + "mean_token_accuracy": 0.9181298613548279, + "num_tokens": 900644590.0, + "step": 23204 + }, + { + "epoch": 2.9519145146927874, + "grad_norm": 2.639319896697998, + "learning_rate": 1e-06, + "loss": 0.24, + "mean_token_accuracy": 0.9104887247085571, + "num_tokens": 900685499.0, + "step": 23205 + }, + { + "epoch": 2.9520417249713775, + "grad_norm": 3.2336599826812744, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.8994001746177673, + "num_tokens": 900718453.0, + "step": 23206 + }, + { + "epoch": 2.952168935249968, + "grad_norm": 3.27079439163208, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.893905520439148, + "num_tokens": 900750974.0, + "step": 23207 + }, + { + "epoch": 2.9522961455285586, + "grad_norm": 2.8012280464172363, + "learning_rate": 1e-06, + "loss": 0.2743, + "mean_token_accuracy": 0.900762677192688, + "num_tokens": 900793665.0, + "step": 23208 + }, + { + "epoch": 2.952423355807149, + "grad_norm": 2.5528624057769775, + "learning_rate": 1e-06, + "loss": 0.2359, + "mean_token_accuracy": 0.9129191040992737, + "num_tokens": 900838590.0, + "step": 23209 + }, + { + "epoch": 2.9525505660857396, + "grad_norm": 3.3711464405059814, + "learning_rate": 1e-06, + "loss": 0.3036, + "mean_token_accuracy": 0.8893148899078369, + "num_tokens": 900870230.0, + "step": 23210 + }, + { + "epoch": 2.95267777636433, + "grad_norm": 2.78946852684021, + "learning_rate": 1e-06, + "loss": 0.2839, + "mean_token_accuracy": 0.898173987865448, + "num_tokens": 900914612.0, + "step": 23211 + }, + { + "epoch": 2.9528049866429207, + "grad_norm": 2.660689353942871, + "learning_rate": 1e-06, + "loss": 0.2553, + "mean_token_accuracy": 0.9069935083389282, + "num_tokens": 900959317.0, + "step": 23212 + }, + { + "epoch": 2.952932196921511, + "grad_norm": 2.9028432369232178, + "learning_rate": 1e-06, + "loss": 0.2907, + "mean_token_accuracy": 0.8959582448005676, + "num_tokens": 900998250.0, + "step": 23213 + }, + { + "epoch": 2.9530594072001017, + "grad_norm": 2.733983278274536, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.902280330657959, + "num_tokens": 901040928.0, + "step": 23214 + }, + { + "epoch": 2.9531866174786923, + "grad_norm": 2.729448080062866, + "learning_rate": 1e-06, + "loss": 0.2481, + "mean_token_accuracy": 0.9071381092071533, + "num_tokens": 901081175.0, + "step": 23215 + }, + { + "epoch": 2.953313827757283, + "grad_norm": 2.8862993717193604, + "learning_rate": 1e-06, + "loss": 0.2887, + "mean_token_accuracy": 0.8941488862037659, + "num_tokens": 901117572.0, + "step": 23216 + }, + { + "epoch": 2.9534410380358733, + "grad_norm": 2.722105026245117, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.9106068015098572, + "num_tokens": 901158991.0, + "step": 23217 + }, + { + "epoch": 2.953568248314464, + "grad_norm": 2.993478536605835, + "learning_rate": 1e-06, + "loss": 0.2544, + "mean_token_accuracy": 0.9070076942443848, + "num_tokens": 901194640.0, + "step": 23218 + }, + { + "epoch": 2.9536954585930544, + "grad_norm": 2.8774571418762207, + "learning_rate": 1e-06, + "loss": 0.2548, + "mean_token_accuracy": 0.9065080881118774, + "num_tokens": 901232478.0, + "step": 23219 + }, + { + "epoch": 2.953822668871645, + "grad_norm": 2.718345880508423, + "learning_rate": 1e-06, + "loss": 0.2928, + "mean_token_accuracy": 0.8924979567527771, + "num_tokens": 901279376.0, + "step": 23220 + }, + { + "epoch": 2.9539498791502354, + "grad_norm": 2.9739420413970947, + "learning_rate": 1e-06, + "loss": 0.2934, + "mean_token_accuracy": 0.8915418386459351, + "num_tokens": 901321888.0, + "step": 23221 + }, + { + "epoch": 2.954077089428826, + "grad_norm": 2.8866844177246094, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9062456488609314, + "num_tokens": 901360343.0, + "step": 23222 + }, + { + "epoch": 2.9542042997074165, + "grad_norm": 2.9777512550354004, + "learning_rate": 1e-06, + "loss": 0.2575, + "mean_token_accuracy": 0.9042525887489319, + "num_tokens": 901394769.0, + "step": 23223 + }, + { + "epoch": 2.954331509986007, + "grad_norm": 2.9143550395965576, + "learning_rate": 1e-06, + "loss": 0.2501, + "mean_token_accuracy": 0.9050021171569824, + "num_tokens": 901431025.0, + "step": 23224 + }, + { + "epoch": 2.954458720264597, + "grad_norm": 3.0857954025268555, + "learning_rate": 1e-06, + "loss": 0.2695, + "mean_token_accuracy": 0.8999390602111816, + "num_tokens": 901465895.0, + "step": 23225 + }, + { + "epoch": 2.954585930543188, + "grad_norm": 2.846085548400879, + "learning_rate": 1e-06, + "loss": 0.2891, + "mean_token_accuracy": 0.8943596482276917, + "num_tokens": 901506266.0, + "step": 23226 + }, + { + "epoch": 2.954713140821778, + "grad_norm": 2.7470858097076416, + "learning_rate": 1e-06, + "loss": 0.2815, + "mean_token_accuracy": 0.8998286724090576, + "num_tokens": 901553165.0, + "step": 23227 + }, + { + "epoch": 2.954840351100369, + "grad_norm": 2.835808277130127, + "learning_rate": 1e-06, + "loss": 0.2554, + "mean_token_accuracy": 0.9036232233047485, + "num_tokens": 901591907.0, + "step": 23228 + }, + { + "epoch": 2.954967561378959, + "grad_norm": 2.677644968032837, + "learning_rate": 1e-06, + "loss": 0.2555, + "mean_token_accuracy": 0.9045664072036743, + "num_tokens": 901632570.0, + "step": 23229 + }, + { + "epoch": 2.95509477165755, + "grad_norm": 2.647204875946045, + "learning_rate": 1e-06, + "loss": 0.257, + "mean_token_accuracy": 0.9031037092208862, + "num_tokens": 901676499.0, + "step": 23230 + }, + { + "epoch": 2.9552219819361403, + "grad_norm": 2.996736764907837, + "learning_rate": 1e-06, + "loss": 0.2627, + "mean_token_accuracy": 0.9024665355682373, + "num_tokens": 901716936.0, + "step": 23231 + }, + { + "epoch": 2.955349192214731, + "grad_norm": 2.8860864639282227, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8910971283912659, + "num_tokens": 901760124.0, + "step": 23232 + }, + { + "epoch": 2.9554764024933213, + "grad_norm": 2.8063576221466064, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9064539670944214, + "num_tokens": 901795933.0, + "step": 23233 + }, + { + "epoch": 2.955603612771912, + "grad_norm": 3.0742440223693848, + "learning_rate": 1e-06, + "loss": 0.2744, + "mean_token_accuracy": 0.896266520023346, + "num_tokens": 901831298.0, + "step": 23234 + }, + { + "epoch": 2.9557308230505024, + "grad_norm": 2.862689256668091, + "learning_rate": 1e-06, + "loss": 0.2685, + "mean_token_accuracy": 0.8990412950515747, + "num_tokens": 901872345.0, + "step": 23235 + }, + { + "epoch": 2.955858033329093, + "grad_norm": 3.3032021522521973, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8933871984481812, + "num_tokens": 901906439.0, + "step": 23236 + }, + { + "epoch": 2.9559852436076834, + "grad_norm": 2.694232702255249, + "learning_rate": 1e-06, + "loss": 0.2297, + "mean_token_accuracy": 0.9139638543128967, + "num_tokens": 901947450.0, + "step": 23237 + }, + { + "epoch": 2.956112453886274, + "grad_norm": 3.0262248516082764, + "learning_rate": 1e-06, + "loss": 0.2982, + "mean_token_accuracy": 0.8941488862037659, + "num_tokens": 901991048.0, + "step": 23238 + }, + { + "epoch": 2.9562396641648645, + "grad_norm": 3.079180955886841, + "learning_rate": 1e-06, + "loss": 0.2576, + "mean_token_accuracy": 0.9049034118652344, + "num_tokens": 902027358.0, + "step": 23239 + }, + { + "epoch": 2.956366874443455, + "grad_norm": 2.6063265800476074, + "learning_rate": 1e-06, + "loss": 0.2216, + "mean_token_accuracy": 0.9191080927848816, + "num_tokens": 902070408.0, + "step": 23240 + }, + { + "epoch": 2.9564940847220456, + "grad_norm": 2.7732975482940674, + "learning_rate": 1e-06, + "loss": 0.2431, + "mean_token_accuracy": 0.9082749485969543, + "num_tokens": 902108884.0, + "step": 23241 + }, + { + "epoch": 2.956621295000636, + "grad_norm": 2.7614238262176514, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9042484164237976, + "num_tokens": 902151670.0, + "step": 23242 + }, + { + "epoch": 2.9567485052792266, + "grad_norm": 2.826493501663208, + "learning_rate": 1e-06, + "loss": 0.2768, + "mean_token_accuracy": 0.8978471755981445, + "num_tokens": 902193222.0, + "step": 23243 + }, + { + "epoch": 2.956875715557817, + "grad_norm": 2.7867512702941895, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9051650762557983, + "num_tokens": 902234807.0, + "step": 23244 + }, + { + "epoch": 2.9570029258364077, + "grad_norm": 2.598771095275879, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.904787540435791, + "num_tokens": 902277861.0, + "step": 23245 + }, + { + "epoch": 2.957130136114998, + "grad_norm": 2.7350950241088867, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.905386745929718, + "num_tokens": 902318872.0, + "step": 23246 + }, + { + "epoch": 2.9572573463935887, + "grad_norm": 2.9073100090026855, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9081324934959412, + "num_tokens": 902358610.0, + "step": 23247 + }, + { + "epoch": 2.9573845566721793, + "grad_norm": 2.9169411659240723, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9045586585998535, + "num_tokens": 902394874.0, + "step": 23248 + }, + { + "epoch": 2.95751176695077, + "grad_norm": 2.614464521408081, + "learning_rate": 1e-06, + "loss": 0.2399, + "mean_token_accuracy": 0.9094387888908386, + "num_tokens": 902436999.0, + "step": 23249 + }, + { + "epoch": 2.95763897722936, + "grad_norm": 2.8807199001312256, + "learning_rate": 1e-06, + "loss": 0.2377, + "mean_token_accuracy": 0.9105191230773926, + "num_tokens": 902470753.0, + "step": 23250 + }, + { + "epoch": 2.957766187507951, + "grad_norm": 3.15501070022583, + "learning_rate": 1e-06, + "loss": 0.3165, + "mean_token_accuracy": 0.8848263025283813, + "num_tokens": 902510249.0, + "step": 23251 + }, + { + "epoch": 2.957893397786541, + "grad_norm": 3.0531504154205322, + "learning_rate": 1e-06, + "loss": 0.2772, + "mean_token_accuracy": 0.8974390029907227, + "num_tokens": 902549128.0, + "step": 23252 + }, + { + "epoch": 2.958020608065132, + "grad_norm": 3.015956163406372, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8943817615509033, + "num_tokens": 902590025.0, + "step": 23253 + }, + { + "epoch": 2.958147818343722, + "grad_norm": 2.912386178970337, + "learning_rate": 1e-06, + "loss": 0.3012, + "mean_token_accuracy": 0.8902562856674194, + "num_tokens": 902634533.0, + "step": 23254 + }, + { + "epoch": 2.958275028622313, + "grad_norm": 2.662142276763916, + "learning_rate": 1e-06, + "loss": 0.2313, + "mean_token_accuracy": 0.9152877330780029, + "num_tokens": 902674533.0, + "step": 23255 + }, + { + "epoch": 2.958402238900903, + "grad_norm": 3.340075731277466, + "learning_rate": 1e-06, + "loss": 0.2569, + "mean_token_accuracy": 0.9058426022529602, + "num_tokens": 902718896.0, + "step": 23256 + }, + { + "epoch": 2.9585294491794936, + "grad_norm": 3.0723230838775635, + "learning_rate": 1e-06, + "loss": 0.262, + "mean_token_accuracy": 0.9009369611740112, + "num_tokens": 902753847.0, + "step": 23257 + }, + { + "epoch": 2.958656659458084, + "grad_norm": 2.6802332401275635, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9045296907424927, + "num_tokens": 902796193.0, + "step": 23258 + }, + { + "epoch": 2.9587838697366746, + "grad_norm": 2.613351345062256, + "learning_rate": 1e-06, + "loss": 0.2229, + "mean_token_accuracy": 0.9173115491867065, + "num_tokens": 902834451.0, + "step": 23259 + }, + { + "epoch": 2.958911080015265, + "grad_norm": 2.6729345321655273, + "learning_rate": 1e-06, + "loss": 0.2723, + "mean_token_accuracy": 0.8994712829589844, + "num_tokens": 902879179.0, + "step": 23260 + }, + { + "epoch": 2.9590382902938557, + "grad_norm": 3.138836145401001, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.89835125207901, + "num_tokens": 902915724.0, + "step": 23261 + }, + { + "epoch": 2.959165500572446, + "grad_norm": 3.0409457683563232, + "learning_rate": 1e-06, + "loss": 0.2662, + "mean_token_accuracy": 0.901458740234375, + "num_tokens": 902952544.0, + "step": 23262 + }, + { + "epoch": 2.9592927108510367, + "grad_norm": 3.014300584793091, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8973768949508667, + "num_tokens": 902986969.0, + "step": 23263 + }, + { + "epoch": 2.9594199211296273, + "grad_norm": 2.7218501567840576, + "learning_rate": 1e-06, + "loss": 0.2614, + "mean_token_accuracy": 0.9037691354751587, + "num_tokens": 903031871.0, + "step": 23264 + }, + { + "epoch": 2.959547131408218, + "grad_norm": 2.9192135334014893, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.9011416435241699, + "num_tokens": 903070645.0, + "step": 23265 + }, + { + "epoch": 2.9596743416868083, + "grad_norm": 2.610607385635376, + "learning_rate": 1e-06, + "loss": 0.2221, + "mean_token_accuracy": 0.9193366765975952, + "num_tokens": 903107885.0, + "step": 23266 + }, + { + "epoch": 2.959801551965399, + "grad_norm": 2.9011216163635254, + "learning_rate": 1e-06, + "loss": 0.2853, + "mean_token_accuracy": 0.896257758140564, + "num_tokens": 903151880.0, + "step": 23267 + }, + { + "epoch": 2.9599287622439894, + "grad_norm": 2.6655287742614746, + "learning_rate": 1e-06, + "loss": 0.2561, + "mean_token_accuracy": 0.903741717338562, + "num_tokens": 903195372.0, + "step": 23268 + }, + { + "epoch": 2.96005597252258, + "grad_norm": 2.8390142917633057, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8984042406082153, + "num_tokens": 903234170.0, + "step": 23269 + }, + { + "epoch": 2.9601831828011704, + "grad_norm": 3.052015542984009, + "learning_rate": 1e-06, + "loss": 0.283, + "mean_token_accuracy": 0.8964775800704956, + "num_tokens": 903273355.0, + "step": 23270 + }, + { + "epoch": 2.960310393079761, + "grad_norm": 2.8272407054901123, + "learning_rate": 1e-06, + "loss": 0.2693, + "mean_token_accuracy": 0.9006675481796265, + "num_tokens": 903311316.0, + "step": 23271 + }, + { + "epoch": 2.9604376033583515, + "grad_norm": 2.7698745727539062, + "learning_rate": 1e-06, + "loss": 0.2605, + "mean_token_accuracy": 0.9047126770019531, + "num_tokens": 903355630.0, + "step": 23272 + }, + { + "epoch": 2.960564813636942, + "grad_norm": 2.7748374938964844, + "learning_rate": 1e-06, + "loss": 0.273, + "mean_token_accuracy": 0.900141179561615, + "num_tokens": 903398398.0, + "step": 23273 + }, + { + "epoch": 2.9606920239155325, + "grad_norm": 2.521596908569336, + "learning_rate": 1e-06, + "loss": 0.2348, + "mean_token_accuracy": 0.9127053022384644, + "num_tokens": 903443195.0, + "step": 23274 + }, + { + "epoch": 2.9608192341941226, + "grad_norm": 2.925307273864746, + "learning_rate": 1e-06, + "loss": 0.3038, + "mean_token_accuracy": 0.8911941647529602, + "num_tokens": 903482484.0, + "step": 23275 + }, + { + "epoch": 2.9609464444727136, + "grad_norm": 3.2854976654052734, + "learning_rate": 1e-06, + "loss": 0.2908, + "mean_token_accuracy": 0.894845724105835, + "num_tokens": 903520172.0, + "step": 23276 + }, + { + "epoch": 2.9610736547513037, + "grad_norm": 2.8052899837493896, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.902317464351654, + "num_tokens": 903557921.0, + "step": 23277 + }, + { + "epoch": 2.9612008650298947, + "grad_norm": 3.025305986404419, + "learning_rate": 1e-06, + "loss": 0.2586, + "mean_token_accuracy": 0.9037115573883057, + "num_tokens": 903591764.0, + "step": 23278 + }, + { + "epoch": 2.9613280753084847, + "grad_norm": 3.0663094520568848, + "learning_rate": 1e-06, + "loss": 0.2428, + "mean_token_accuracy": 0.9100586175918579, + "num_tokens": 903624516.0, + "step": 23279 + }, + { + "epoch": 2.9614552855870753, + "grad_norm": 2.7793540954589844, + "learning_rate": 1e-06, + "loss": 0.2684, + "mean_token_accuracy": 0.902539849281311, + "num_tokens": 903663080.0, + "step": 23280 + }, + { + "epoch": 2.961582495865666, + "grad_norm": 2.6895930767059326, + "learning_rate": 1e-06, + "loss": 0.2923, + "mean_token_accuracy": 0.8918635845184326, + "num_tokens": 903710617.0, + "step": 23281 + }, + { + "epoch": 2.9617097061442563, + "grad_norm": 2.784855604171753, + "learning_rate": 1e-06, + "loss": 0.2303, + "mean_token_accuracy": 0.913077175617218, + "num_tokens": 903748394.0, + "step": 23282 + }, + { + "epoch": 2.961836916422847, + "grad_norm": 2.6048972606658936, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9136837720870972, + "num_tokens": 903791441.0, + "step": 23283 + }, + { + "epoch": 2.9619641267014374, + "grad_norm": 2.9219753742218018, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.8988101482391357, + "num_tokens": 903833909.0, + "step": 23284 + }, + { + "epoch": 2.962091336980028, + "grad_norm": 3.2313482761383057, + "learning_rate": 1e-06, + "loss": 0.2996, + "mean_token_accuracy": 0.8977102041244507, + "num_tokens": 903868834.0, + "step": 23285 + }, + { + "epoch": 2.9622185472586184, + "grad_norm": 2.8777060508728027, + "learning_rate": 1e-06, + "loss": 0.232, + "mean_token_accuracy": 0.9127497673034668, + "num_tokens": 903903897.0, + "step": 23286 + }, + { + "epoch": 2.962345757537209, + "grad_norm": 2.6553268432617188, + "learning_rate": 1e-06, + "loss": 0.2596, + "mean_token_accuracy": 0.904083251953125, + "num_tokens": 903948750.0, + "step": 23287 + }, + { + "epoch": 2.9624729678157995, + "grad_norm": 3.0263357162475586, + "learning_rate": 1e-06, + "loss": 0.2708, + "mean_token_accuracy": 0.900287926197052, + "num_tokens": 903986554.0, + "step": 23288 + }, + { + "epoch": 2.96260017809439, + "grad_norm": 3.1057214736938477, + "learning_rate": 1e-06, + "loss": 0.2922, + "mean_token_accuracy": 0.8910045623779297, + "num_tokens": 904025747.0, + "step": 23289 + }, + { + "epoch": 2.9627273883729806, + "grad_norm": 2.9705114364624023, + "learning_rate": 1e-06, + "loss": 0.2493, + "mean_token_accuracy": 0.9051800966262817, + "num_tokens": 904061405.0, + "step": 23290 + }, + { + "epoch": 2.962854598651571, + "grad_norm": 2.5741238594055176, + "learning_rate": 1e-06, + "loss": 0.2381, + "mean_token_accuracy": 0.912290096282959, + "num_tokens": 904103403.0, + "step": 23291 + }, + { + "epoch": 2.9629818089301616, + "grad_norm": 2.7606470584869385, + "learning_rate": 1e-06, + "loss": 0.259, + "mean_token_accuracy": 0.9052772521972656, + "num_tokens": 904144441.0, + "step": 23292 + }, + { + "epoch": 2.963109019208752, + "grad_norm": 2.550889492034912, + "learning_rate": 1e-06, + "loss": 0.2495, + "mean_token_accuracy": 0.9081011414527893, + "num_tokens": 904191550.0, + "step": 23293 + }, + { + "epoch": 2.9632362294873427, + "grad_norm": 2.876892566680908, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9095218181610107, + "num_tokens": 904228469.0, + "step": 23294 + }, + { + "epoch": 2.963363439765933, + "grad_norm": 2.940870523452759, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.9017190933227539, + "num_tokens": 904269087.0, + "step": 23295 + }, + { + "epoch": 2.9634906500445237, + "grad_norm": 2.6100914478302, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9024976491928101, + "num_tokens": 904313536.0, + "step": 23296 + }, + { + "epoch": 2.9636178603231143, + "grad_norm": 2.63972544670105, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9144130945205688, + "num_tokens": 904352361.0, + "step": 23297 + }, + { + "epoch": 2.963745070601705, + "grad_norm": 2.5558252334594727, + "learning_rate": 1e-06, + "loss": 0.2536, + "mean_token_accuracy": 0.9052810668945312, + "num_tokens": 904396512.0, + "step": 23298 + }, + { + "epoch": 2.9638722808802953, + "grad_norm": 2.916801691055298, + "learning_rate": 1e-06, + "loss": 0.2681, + "mean_token_accuracy": 0.8991693258285522, + "num_tokens": 904437632.0, + "step": 23299 + }, + { + "epoch": 2.9639994911588854, + "grad_norm": 2.956604480743408, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.8996719121932983, + "num_tokens": 904478127.0, + "step": 23300 + }, + { + "epoch": 2.9641267014374764, + "grad_norm": 3.4642300605773926, + "learning_rate": 1e-06, + "loss": 0.3211, + "mean_token_accuracy": 0.8858146071434021, + "num_tokens": 904513068.0, + "step": 23301 + }, + { + "epoch": 2.9642539117160664, + "grad_norm": 2.8051092624664307, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.9058552384376526, + "num_tokens": 904550737.0, + "step": 23302 + }, + { + "epoch": 2.9643811219946574, + "grad_norm": 2.4352922439575195, + "learning_rate": 1e-06, + "loss": 0.2379, + "mean_token_accuracy": 0.9116547107696533, + "num_tokens": 904595380.0, + "step": 23303 + }, + { + "epoch": 2.9645083322732475, + "grad_norm": 3.142460584640503, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9020888209342957, + "num_tokens": 904629325.0, + "step": 23304 + }, + { + "epoch": 2.964635542551838, + "grad_norm": 2.8499410152435303, + "learning_rate": 1e-06, + "loss": 0.2628, + "mean_token_accuracy": 0.9025261402130127, + "num_tokens": 904669322.0, + "step": 23305 + }, + { + "epoch": 2.9647627528304286, + "grad_norm": 3.0745363235473633, + "learning_rate": 1e-06, + "loss": 0.2619, + "mean_token_accuracy": 0.901042103767395, + "num_tokens": 904704526.0, + "step": 23306 + }, + { + "epoch": 2.964889963109019, + "grad_norm": 2.7056386470794678, + "learning_rate": 1e-06, + "loss": 0.2482, + "mean_token_accuracy": 0.907792866230011, + "num_tokens": 904743777.0, + "step": 23307 + }, + { + "epoch": 2.9650171733876096, + "grad_norm": 2.875558614730835, + "learning_rate": 1e-06, + "loss": 0.2716, + "mean_token_accuracy": 0.8999006152153015, + "num_tokens": 904781892.0, + "step": 23308 + }, + { + "epoch": 2.9651443836662, + "grad_norm": 3.2649946212768555, + "learning_rate": 1e-06, + "loss": 0.2829, + "mean_token_accuracy": 0.8983404040336609, + "num_tokens": 904822798.0, + "step": 23309 + }, + { + "epoch": 2.9652715939447907, + "grad_norm": 2.8491263389587402, + "learning_rate": 1e-06, + "loss": 0.2504, + "mean_token_accuracy": 0.9094388484954834, + "num_tokens": 904859304.0, + "step": 23310 + }, + { + "epoch": 2.965398804223381, + "grad_norm": 2.979384422302246, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9057840704917908, + "num_tokens": 904895836.0, + "step": 23311 + }, + { + "epoch": 2.9655260145019717, + "grad_norm": 2.781677722930908, + "learning_rate": 1e-06, + "loss": 0.2345, + "mean_token_accuracy": 0.9092722535133362, + "num_tokens": 904931846.0, + "step": 23312 + }, + { + "epoch": 2.9656532247805623, + "grad_norm": 2.9259250164031982, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9062072038650513, + "num_tokens": 904969535.0, + "step": 23313 + }, + { + "epoch": 2.965780435059153, + "grad_norm": 2.9488162994384766, + "learning_rate": 1e-06, + "loss": 0.3157, + "mean_token_accuracy": 0.8836153745651245, + "num_tokens": 905012186.0, + "step": 23314 + }, + { + "epoch": 2.9659076453377433, + "grad_norm": 3.101997137069702, + "learning_rate": 1e-06, + "loss": 0.2796, + "mean_token_accuracy": 0.8977027535438538, + "num_tokens": 905050418.0, + "step": 23315 + }, + { + "epoch": 2.966034855616334, + "grad_norm": 2.888507843017578, + "learning_rate": 1e-06, + "loss": 0.2534, + "mean_token_accuracy": 0.9076307415962219, + "num_tokens": 905090317.0, + "step": 23316 + }, + { + "epoch": 2.9661620658949244, + "grad_norm": 3.1332974433898926, + "learning_rate": 1e-06, + "loss": 0.2844, + "mean_token_accuracy": 0.8974906206130981, + "num_tokens": 905126391.0, + "step": 23317 + }, + { + "epoch": 2.966289276173515, + "grad_norm": 3.1693785190582275, + "learning_rate": 1e-06, + "loss": 0.3034, + "mean_token_accuracy": 0.8883268237113953, + "num_tokens": 905164519.0, + "step": 23318 + }, + { + "epoch": 2.9664164864521054, + "grad_norm": 2.889838457107544, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9025775194168091, + "num_tokens": 905201046.0, + "step": 23319 + }, + { + "epoch": 2.966543696730696, + "grad_norm": 2.8450379371643066, + "learning_rate": 1e-06, + "loss": 0.2516, + "mean_token_accuracy": 0.9056839346885681, + "num_tokens": 905239411.0, + "step": 23320 + }, + { + "epoch": 2.9666709070092865, + "grad_norm": 2.812051773071289, + "learning_rate": 1e-06, + "loss": 0.2361, + "mean_token_accuracy": 0.9126298427581787, + "num_tokens": 905278200.0, + "step": 23321 + }, + { + "epoch": 2.966798117287877, + "grad_norm": 2.6429734230041504, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9086155891418457, + "num_tokens": 905321644.0, + "step": 23322 + }, + { + "epoch": 2.966925327566467, + "grad_norm": 2.8972203731536865, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8965541124343872, + "num_tokens": 905362750.0, + "step": 23323 + }, + { + "epoch": 2.967052537845058, + "grad_norm": 2.840787649154663, + "learning_rate": 1e-06, + "loss": 0.2242, + "mean_token_accuracy": 0.9135776162147522, + "num_tokens": 905396588.0, + "step": 23324 + }, + { + "epoch": 2.967179748123648, + "grad_norm": 2.698946237564087, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9012173414230347, + "num_tokens": 905438296.0, + "step": 23325 + }, + { + "epoch": 2.967306958402239, + "grad_norm": 2.755760669708252, + "learning_rate": 1e-06, + "loss": 0.2485, + "mean_token_accuracy": 0.9056859016418457, + "num_tokens": 905476704.0, + "step": 23326 + }, + { + "epoch": 2.967434168680829, + "grad_norm": 2.806159257888794, + "learning_rate": 1e-06, + "loss": 0.2679, + "mean_token_accuracy": 0.9023072123527527, + "num_tokens": 905516174.0, + "step": 23327 + }, + { + "epoch": 2.96756137895942, + "grad_norm": 2.851653575897217, + "learning_rate": 1e-06, + "loss": 0.2609, + "mean_token_accuracy": 0.9013165235519409, + "num_tokens": 905555514.0, + "step": 23328 + }, + { + "epoch": 2.9676885892380103, + "grad_norm": 2.8180503845214844, + "learning_rate": 1e-06, + "loss": 0.2409, + "mean_token_accuracy": 0.9098361730575562, + "num_tokens": 905592163.0, + "step": 23329 + }, + { + "epoch": 2.967815799516601, + "grad_norm": 3.1660962104797363, + "learning_rate": 1e-06, + "loss": 0.2965, + "mean_token_accuracy": 0.8959461450576782, + "num_tokens": 905628597.0, + "step": 23330 + }, + { + "epoch": 2.9679430097951913, + "grad_norm": 2.680421829223633, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8994995355606079, + "num_tokens": 905670470.0, + "step": 23331 + }, + { + "epoch": 2.968070220073782, + "grad_norm": 2.777622699737549, + "learning_rate": 1e-06, + "loss": 0.274, + "mean_token_accuracy": 0.8988004922866821, + "num_tokens": 905711133.0, + "step": 23332 + }, + { + "epoch": 2.9681974303523724, + "grad_norm": 2.8498852252960205, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.908649206161499, + "num_tokens": 905751026.0, + "step": 23333 + }, + { + "epoch": 2.968324640630963, + "grad_norm": 2.912067174911499, + "learning_rate": 1e-06, + "loss": 0.2806, + "mean_token_accuracy": 0.8969986438751221, + "num_tokens": 905789883.0, + "step": 23334 + }, + { + "epoch": 2.9684518509095534, + "grad_norm": 2.7069644927978516, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9072040915489197, + "num_tokens": 905829771.0, + "step": 23335 + }, + { + "epoch": 2.968579061188144, + "grad_norm": 2.7971811294555664, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9054394364356995, + "num_tokens": 905871365.0, + "step": 23336 + }, + { + "epoch": 2.9687062714667345, + "grad_norm": 2.6467490196228027, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9086328744888306, + "num_tokens": 905913087.0, + "step": 23337 + }, + { + "epoch": 2.968833481745325, + "grad_norm": 2.570063352584839, + "learning_rate": 1e-06, + "loss": 0.2376, + "mean_token_accuracy": 0.9107575416564941, + "num_tokens": 905958063.0, + "step": 23338 + }, + { + "epoch": 2.9689606920239155, + "grad_norm": 2.7657277584075928, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.9101623296737671, + "num_tokens": 905996309.0, + "step": 23339 + }, + { + "epoch": 2.969087902302506, + "grad_norm": 3.0140860080718994, + "learning_rate": 1e-06, + "loss": 0.2943, + "mean_token_accuracy": 0.8940315246582031, + "num_tokens": 906036150.0, + "step": 23340 + }, + { + "epoch": 2.9692151125810966, + "grad_norm": 2.6329073905944824, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9075683355331421, + "num_tokens": 906079665.0, + "step": 23341 + }, + { + "epoch": 2.969342322859687, + "grad_norm": 3.0326576232910156, + "learning_rate": 1e-06, + "loss": 0.2282, + "mean_token_accuracy": 0.9118165969848633, + "num_tokens": 906111995.0, + "step": 23342 + }, + { + "epoch": 2.9694695331382777, + "grad_norm": 3.4520130157470703, + "learning_rate": 1e-06, + "loss": 0.2924, + "mean_token_accuracy": 0.8921289443969727, + "num_tokens": 906143238.0, + "step": 23343 + }, + { + "epoch": 2.969596743416868, + "grad_norm": 3.353738307952881, + "learning_rate": 1e-06, + "loss": 0.2623, + "mean_token_accuracy": 0.904263436794281, + "num_tokens": 906173716.0, + "step": 23344 + }, + { + "epoch": 2.9697239536954587, + "grad_norm": 3.0390408039093018, + "learning_rate": 1e-06, + "loss": 0.2675, + "mean_token_accuracy": 0.9039809703826904, + "num_tokens": 906207672.0, + "step": 23345 + }, + { + "epoch": 2.9698511639740492, + "grad_norm": 3.1175365447998047, + "learning_rate": 1e-06, + "loss": 0.2707, + "mean_token_accuracy": 0.9010237455368042, + "num_tokens": 906245750.0, + "step": 23346 + }, + { + "epoch": 2.9699783742526398, + "grad_norm": 2.753406286239624, + "learning_rate": 1e-06, + "loss": 0.2457, + "mean_token_accuracy": 0.9084246754646301, + "num_tokens": 906284240.0, + "step": 23347 + }, + { + "epoch": 2.97010558453123, + "grad_norm": 2.9589462280273438, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8957904577255249, + "num_tokens": 906325712.0, + "step": 23348 + }, + { + "epoch": 2.970232794809821, + "grad_norm": 2.6152052879333496, + "learning_rate": 1e-06, + "loss": 0.238, + "mean_token_accuracy": 0.9133741855621338, + "num_tokens": 906369873.0, + "step": 23349 + }, + { + "epoch": 2.970360005088411, + "grad_norm": 3.0166373252868652, + "learning_rate": 1e-06, + "loss": 0.3028, + "mean_token_accuracy": 0.8899552822113037, + "num_tokens": 906409695.0, + "step": 23350 + }, + { + "epoch": 2.970487215367002, + "grad_norm": 2.961482524871826, + "learning_rate": 1e-06, + "loss": 0.2347, + "mean_token_accuracy": 0.9112927913665771, + "num_tokens": 906444193.0, + "step": 23351 + }, + { + "epoch": 2.970614425645592, + "grad_norm": 3.0507171154022217, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9053281545639038, + "num_tokens": 906477245.0, + "step": 23352 + }, + { + "epoch": 2.970741635924183, + "grad_norm": 2.8415703773498535, + "learning_rate": 1e-06, + "loss": 0.236, + "mean_token_accuracy": 0.9122840166091919, + "num_tokens": 906512475.0, + "step": 23353 + }, + { + "epoch": 2.970868846202773, + "grad_norm": 2.8288230895996094, + "learning_rate": 1e-06, + "loss": 0.2433, + "mean_token_accuracy": 0.9112136363983154, + "num_tokens": 906550612.0, + "step": 23354 + }, + { + "epoch": 2.9709960564813636, + "grad_norm": 2.6434807777404785, + "learning_rate": 1e-06, + "loss": 0.2404, + "mean_token_accuracy": 0.9114401340484619, + "num_tokens": 906595325.0, + "step": 23355 + }, + { + "epoch": 2.971123266759954, + "grad_norm": 2.871206760406494, + "learning_rate": 1e-06, + "loss": 0.2546, + "mean_token_accuracy": 0.9082580804824829, + "num_tokens": 906633846.0, + "step": 23356 + }, + { + "epoch": 2.9712504770385446, + "grad_norm": 2.869699239730835, + "learning_rate": 1e-06, + "loss": 0.2545, + "mean_token_accuracy": 0.9039469957351685, + "num_tokens": 906671380.0, + "step": 23357 + }, + { + "epoch": 2.971377687317135, + "grad_norm": 3.2782304286956787, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.898004412651062, + "num_tokens": 906708147.0, + "step": 23358 + }, + { + "epoch": 2.9715048975957257, + "grad_norm": 2.604116916656494, + "learning_rate": 1e-06, + "loss": 0.2463, + "mean_token_accuracy": 0.9074612855911255, + "num_tokens": 906750933.0, + "step": 23359 + }, + { + "epoch": 2.971632107874316, + "grad_norm": 2.5531680583953857, + "learning_rate": 1e-06, + "loss": 0.2394, + "mean_token_accuracy": 0.9092423915863037, + "num_tokens": 906792504.0, + "step": 23360 + }, + { + "epoch": 2.9717593181529067, + "grad_norm": 2.6237566471099854, + "learning_rate": 1e-06, + "loss": 0.2529, + "mean_token_accuracy": 0.9051014184951782, + "num_tokens": 906838066.0, + "step": 23361 + }, + { + "epoch": 2.9718865284314973, + "grad_norm": 3.094609022140503, + "learning_rate": 1e-06, + "loss": 0.2578, + "mean_token_accuracy": 0.9057132005691528, + "num_tokens": 906870542.0, + "step": 23362 + }, + { + "epoch": 2.972013738710088, + "grad_norm": 2.6381068229675293, + "learning_rate": 1e-06, + "loss": 0.2184, + "mean_token_accuracy": 0.9154519438743591, + "num_tokens": 906909488.0, + "step": 23363 + }, + { + "epoch": 2.9721409489886783, + "grad_norm": 2.8826253414154053, + "learning_rate": 1e-06, + "loss": 0.242, + "mean_token_accuracy": 0.9093378186225891, + "num_tokens": 906945636.0, + "step": 23364 + }, + { + "epoch": 2.972268159267269, + "grad_norm": 2.622838258743286, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9111852645874023, + "num_tokens": 906988659.0, + "step": 23365 + }, + { + "epoch": 2.9723953695458594, + "grad_norm": 2.6728742122650146, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.9140014052391052, + "num_tokens": 907032652.0, + "step": 23366 + }, + { + "epoch": 2.97252257982445, + "grad_norm": 3.0108373165130615, + "learning_rate": 1e-06, + "loss": 0.248, + "mean_token_accuracy": 0.9070738554000854, + "num_tokens": 907065478.0, + "step": 23367 + }, + { + "epoch": 2.9726497901030404, + "grad_norm": 2.9315507411956787, + "learning_rate": 1e-06, + "loss": 0.2714, + "mean_token_accuracy": 0.9006844758987427, + "num_tokens": 907103957.0, + "step": 23368 + }, + { + "epoch": 2.972777000381631, + "grad_norm": 2.8544225692749023, + "learning_rate": 1e-06, + "loss": 0.2275, + "mean_token_accuracy": 0.9162392020225525, + "num_tokens": 907141672.0, + "step": 23369 + }, + { + "epoch": 2.9729042106602215, + "grad_norm": 2.7684757709503174, + "learning_rate": 1e-06, + "loss": 0.2667, + "mean_token_accuracy": 0.9019713997840881, + "num_tokens": 907182261.0, + "step": 23370 + }, + { + "epoch": 2.973031420938812, + "grad_norm": 2.6659414768218994, + "learning_rate": 1e-06, + "loss": 0.246, + "mean_token_accuracy": 0.9079952239990234, + "num_tokens": 907225042.0, + "step": 23371 + }, + { + "epoch": 2.9731586312174025, + "grad_norm": 3.0194151401519775, + "learning_rate": 1e-06, + "loss": 0.2963, + "mean_token_accuracy": 0.8943359851837158, + "num_tokens": 907260526.0, + "step": 23372 + }, + { + "epoch": 2.9732858414959926, + "grad_norm": 2.7587814331054688, + "learning_rate": 1e-06, + "loss": 0.2757, + "mean_token_accuracy": 0.8968711495399475, + "num_tokens": 907304954.0, + "step": 23373 + }, + { + "epoch": 2.9734130517745836, + "grad_norm": 2.941244602203369, + "learning_rate": 1e-06, + "loss": 0.2412, + "mean_token_accuracy": 0.9081406593322754, + "num_tokens": 907339741.0, + "step": 23374 + }, + { + "epoch": 2.9735402620531737, + "grad_norm": 2.797682285308838, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.9109413623809814, + "num_tokens": 907380740.0, + "step": 23375 + }, + { + "epoch": 2.9736674723317646, + "grad_norm": 2.846735954284668, + "learning_rate": 1e-06, + "loss": 0.2507, + "mean_token_accuracy": 0.9078561067581177, + "num_tokens": 907418871.0, + "step": 23376 + }, + { + "epoch": 2.9737946826103547, + "grad_norm": 3.137479543685913, + "learning_rate": 1e-06, + "loss": 0.2863, + "mean_token_accuracy": 0.8936722278594971, + "num_tokens": 907457679.0, + "step": 23377 + }, + { + "epoch": 2.9739218928889453, + "grad_norm": 2.900031805038452, + "learning_rate": 1e-06, + "loss": 0.2813, + "mean_token_accuracy": 0.8963468074798584, + "num_tokens": 907500505.0, + "step": 23378 + }, + { + "epoch": 2.974049103167536, + "grad_norm": 2.9790892601013184, + "learning_rate": 1e-06, + "loss": 0.2342, + "mean_token_accuracy": 0.9107566475868225, + "num_tokens": 907533102.0, + "step": 23379 + }, + { + "epoch": 2.9741763134461263, + "grad_norm": 3.077838897705078, + "learning_rate": 1e-06, + "loss": 0.289, + "mean_token_accuracy": 0.8929933309555054, + "num_tokens": 907570427.0, + "step": 23380 + }, + { + "epoch": 2.974303523724717, + "grad_norm": 2.9955673217773438, + "learning_rate": 1e-06, + "loss": 0.2821, + "mean_token_accuracy": 0.896465539932251, + "num_tokens": 907610247.0, + "step": 23381 + }, + { + "epoch": 2.9744307340033074, + "grad_norm": 2.9475669860839844, + "learning_rate": 1e-06, + "loss": 0.2663, + "mean_token_accuracy": 0.9020484089851379, + "num_tokens": 907646453.0, + "step": 23382 + }, + { + "epoch": 2.974557944281898, + "grad_norm": 3.1901795864105225, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.9040718078613281, + "num_tokens": 907679371.0, + "step": 23383 + }, + { + "epoch": 2.9746851545604884, + "grad_norm": 2.69409441947937, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9095532894134521, + "num_tokens": 907717407.0, + "step": 23384 + }, + { + "epoch": 2.974812364839079, + "grad_norm": 2.831986904144287, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9133245944976807, + "num_tokens": 907752085.0, + "step": 23385 + }, + { + "epoch": 2.9749395751176695, + "grad_norm": 2.8308746814727783, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.90336674451828, + "num_tokens": 907791971.0, + "step": 23386 + }, + { + "epoch": 2.97506678539626, + "grad_norm": 2.550727605819702, + "learning_rate": 1e-06, + "loss": 0.2737, + "mean_token_accuracy": 0.89811110496521, + "num_tokens": 907837742.0, + "step": 23387 + }, + { + "epoch": 2.9751939956748505, + "grad_norm": 3.030251979827881, + "learning_rate": 1e-06, + "loss": 0.2818, + "mean_token_accuracy": 0.8941066265106201, + "num_tokens": 907876218.0, + "step": 23388 + }, + { + "epoch": 2.975321205953441, + "grad_norm": 2.900428533554077, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.8981119990348816, + "num_tokens": 907916204.0, + "step": 23389 + }, + { + "epoch": 2.9754484162320316, + "grad_norm": 2.6743524074554443, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9113966822624207, + "num_tokens": 907953914.0, + "step": 23390 + }, + { + "epoch": 2.975575626510622, + "grad_norm": 2.9080686569213867, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.902589738368988, + "num_tokens": 907991963.0, + "step": 23391 + }, + { + "epoch": 2.9757028367892127, + "grad_norm": 2.659606695175171, + "learning_rate": 1e-06, + "loss": 0.2601, + "mean_token_accuracy": 0.9043465852737427, + "num_tokens": 908036886.0, + "step": 23392 + }, + { + "epoch": 2.975830047067803, + "grad_norm": 3.0749666690826416, + "learning_rate": 1e-06, + "loss": 0.2603, + "mean_token_accuracy": 0.9006086587905884, + "num_tokens": 908072786.0, + "step": 23393 + }, + { + "epoch": 2.9759572573463937, + "grad_norm": 3.0589663982391357, + "learning_rate": 1e-06, + "loss": 0.2647, + "mean_token_accuracy": 0.9004601240158081, + "num_tokens": 908108988.0, + "step": 23394 + }, + { + "epoch": 2.9760844676249842, + "grad_norm": 2.92156720161438, + "learning_rate": 1e-06, + "loss": 0.2436, + "mean_token_accuracy": 0.9084425568580627, + "num_tokens": 908144403.0, + "step": 23395 + }, + { + "epoch": 2.9762116779035748, + "grad_norm": 3.0491883754730225, + "learning_rate": 1e-06, + "loss": 0.2803, + "mean_token_accuracy": 0.8987691402435303, + "num_tokens": 908179849.0, + "step": 23396 + }, + { + "epoch": 2.9763388881821653, + "grad_norm": 2.7634189128875732, + "learning_rate": 1e-06, + "loss": 0.2616, + "mean_token_accuracy": 0.905215859413147, + "num_tokens": 908218889.0, + "step": 23397 + }, + { + "epoch": 2.9764660984607554, + "grad_norm": 2.7426748275756836, + "learning_rate": 1e-06, + "loss": 0.2624, + "mean_token_accuracy": 0.9045439958572388, + "num_tokens": 908260488.0, + "step": 23398 + }, + { + "epoch": 2.9765933087393464, + "grad_norm": 2.7858898639678955, + "learning_rate": 1e-06, + "loss": 0.2678, + "mean_token_accuracy": 0.899978518486023, + "num_tokens": 908302896.0, + "step": 23399 + }, + { + "epoch": 2.9767205190179364, + "grad_norm": 2.5626471042633057, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9121662378311157, + "num_tokens": 908343017.0, + "step": 23400 + }, + { + "epoch": 2.9768477292965274, + "grad_norm": 3.0178937911987305, + "learning_rate": 1e-06, + "loss": 0.2962, + "mean_token_accuracy": 0.8930224776268005, + "num_tokens": 908382666.0, + "step": 23401 + }, + { + "epoch": 2.9769749395751175, + "grad_norm": 2.690727949142456, + "learning_rate": 1e-06, + "loss": 0.2644, + "mean_token_accuracy": 0.9019432663917542, + "num_tokens": 908426434.0, + "step": 23402 + }, + { + "epoch": 2.977102149853708, + "grad_norm": 3.014033555984497, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9030686616897583, + "num_tokens": 908461629.0, + "step": 23403 + }, + { + "epoch": 2.9772293601322986, + "grad_norm": 3.063060998916626, + "learning_rate": 1e-06, + "loss": 0.2406, + "mean_token_accuracy": 0.9146591424942017, + "num_tokens": 908496313.0, + "step": 23404 + }, + { + "epoch": 2.977356570410889, + "grad_norm": 3.0778160095214844, + "learning_rate": 1e-06, + "loss": 0.2385, + "mean_token_accuracy": 0.910973310470581, + "num_tokens": 908528848.0, + "step": 23405 + }, + { + "epoch": 2.9774837806894796, + "grad_norm": 2.812897205352783, + "learning_rate": 1e-06, + "loss": 0.2722, + "mean_token_accuracy": 0.9009637832641602, + "num_tokens": 908569720.0, + "step": 23406 + }, + { + "epoch": 2.97761099096807, + "grad_norm": 3.034827470779419, + "learning_rate": 1e-06, + "loss": 0.2918, + "mean_token_accuracy": 0.8927531242370605, + "num_tokens": 908609046.0, + "step": 23407 + }, + { + "epoch": 2.9777382012466607, + "grad_norm": 2.557856321334839, + "learning_rate": 1e-06, + "loss": 0.2313, + "mean_token_accuracy": 0.9121217727661133, + "num_tokens": 908650384.0, + "step": 23408 + }, + { + "epoch": 2.977865411525251, + "grad_norm": 2.6379261016845703, + "learning_rate": 1e-06, + "loss": 0.23, + "mean_token_accuracy": 0.9139652252197266, + "num_tokens": 908692387.0, + "step": 23409 + }, + { + "epoch": 2.9779926218038417, + "grad_norm": 2.9094622135162354, + "learning_rate": 1e-06, + "loss": 0.2812, + "mean_token_accuracy": 0.8955531120300293, + "num_tokens": 908734630.0, + "step": 23410 + }, + { + "epoch": 2.9781198320824323, + "grad_norm": 2.8829965591430664, + "learning_rate": 1e-06, + "loss": 0.2841, + "mean_token_accuracy": 0.8936293721199036, + "num_tokens": 908773404.0, + "step": 23411 + }, + { + "epoch": 2.978247042361023, + "grad_norm": 2.7661218643188477, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9061335325241089, + "num_tokens": 908814325.0, + "step": 23412 + }, + { + "epoch": 2.9783742526396133, + "grad_norm": 2.914422035217285, + "learning_rate": 1e-06, + "loss": 0.286, + "mean_token_accuracy": 0.8941121101379395, + "num_tokens": 908854104.0, + "step": 23413 + }, + { + "epoch": 2.978501462918204, + "grad_norm": 2.8990304470062256, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.8989955186843872, + "num_tokens": 908893065.0, + "step": 23414 + }, + { + "epoch": 2.9786286731967944, + "grad_norm": 2.975330352783203, + "learning_rate": 1e-06, + "loss": 0.2726, + "mean_token_accuracy": 0.9008511900901794, + "num_tokens": 908931585.0, + "step": 23415 + }, + { + "epoch": 2.978755883475385, + "grad_norm": 2.975783348083496, + "learning_rate": 1e-06, + "loss": 0.2494, + "mean_token_accuracy": 0.9078209400177002, + "num_tokens": 908970882.0, + "step": 23416 + }, + { + "epoch": 2.9788830937539754, + "grad_norm": 2.9876251220703125, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8951720595359802, + "num_tokens": 909011091.0, + "step": 23417 + }, + { + "epoch": 2.979010304032566, + "grad_norm": 3.136859893798828, + "learning_rate": 1e-06, + "loss": 0.2977, + "mean_token_accuracy": 0.8909853100776672, + "num_tokens": 909050884.0, + "step": 23418 + }, + { + "epoch": 2.9791375143111565, + "grad_norm": 2.9592444896698, + "learning_rate": 1e-06, + "loss": 0.2338, + "mean_token_accuracy": 0.9123222231864929, + "num_tokens": 909083164.0, + "step": 23419 + }, + { + "epoch": 2.979264724589747, + "grad_norm": 3.002979040145874, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9074889421463013, + "num_tokens": 909119847.0, + "step": 23420 + }, + { + "epoch": 2.979391934868337, + "grad_norm": 2.9173076152801514, + "learning_rate": 1e-06, + "loss": 0.229, + "mean_token_accuracy": 0.9163593053817749, + "num_tokens": 909154898.0, + "step": 23421 + }, + { + "epoch": 2.979519145146928, + "grad_norm": 2.7293264865875244, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.909636914730072, + "num_tokens": 909194614.0, + "step": 23422 + }, + { + "epoch": 2.979646355425518, + "grad_norm": 3.1432669162750244, + "learning_rate": 1e-06, + "loss": 0.297, + "mean_token_accuracy": 0.8931496143341064, + "num_tokens": 909234348.0, + "step": 23423 + }, + { + "epoch": 2.979773565704109, + "grad_norm": 2.7769479751586914, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.898938775062561, + "num_tokens": 909280293.0, + "step": 23424 + }, + { + "epoch": 2.979900775982699, + "grad_norm": 2.8774592876434326, + "learning_rate": 1e-06, + "loss": 0.2514, + "mean_token_accuracy": 0.9074762463569641, + "num_tokens": 909318983.0, + "step": 23425 + }, + { + "epoch": 2.98002798626129, + "grad_norm": 2.896242618560791, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.9116742610931396, + "num_tokens": 909354522.0, + "step": 23426 + }, + { + "epoch": 2.9801551965398803, + "grad_norm": 2.8899037837982178, + "learning_rate": 1e-06, + "loss": 0.2613, + "mean_token_accuracy": 0.903863787651062, + "num_tokens": 909390670.0, + "step": 23427 + }, + { + "epoch": 2.980282406818471, + "grad_norm": 2.817986249923706, + "learning_rate": 1e-06, + "loss": 0.2332, + "mean_token_accuracy": 0.9123079776763916, + "num_tokens": 909426642.0, + "step": 23428 + }, + { + "epoch": 2.9804096170970613, + "grad_norm": 3.071636199951172, + "learning_rate": 1e-06, + "loss": 0.3251, + "mean_token_accuracy": 0.8852971196174622, + "num_tokens": 909465438.0, + "step": 23429 + }, + { + "epoch": 2.980536827375652, + "grad_norm": 2.8784115314483643, + "learning_rate": 1e-06, + "loss": 0.2873, + "mean_token_accuracy": 0.8969364166259766, + "num_tokens": 909505769.0, + "step": 23430 + }, + { + "epoch": 2.9806640376542424, + "grad_norm": 2.5747201442718506, + "learning_rate": 1e-06, + "loss": 0.2172, + "mean_token_accuracy": 0.9161585569381714, + "num_tokens": 909545674.0, + "step": 23431 + }, + { + "epoch": 2.980791247932833, + "grad_norm": 2.806532621383667, + "learning_rate": 1e-06, + "loss": 0.241, + "mean_token_accuracy": 0.9116166830062866, + "num_tokens": 909582554.0, + "step": 23432 + }, + { + "epoch": 2.9809184582114234, + "grad_norm": 2.836596727371216, + "learning_rate": 1e-06, + "loss": 0.2577, + "mean_token_accuracy": 0.9012505412101746, + "num_tokens": 909621216.0, + "step": 23433 + }, + { + "epoch": 2.981045668490014, + "grad_norm": 2.5003745555877686, + "learning_rate": 1e-06, + "loss": 0.2479, + "mean_token_accuracy": 0.9084720611572266, + "num_tokens": 909666180.0, + "step": 23434 + }, + { + "epoch": 2.9811728787686045, + "grad_norm": 2.7214925289154053, + "learning_rate": 1e-06, + "loss": 0.2335, + "mean_token_accuracy": 0.9131767749786377, + "num_tokens": 909704034.0, + "step": 23435 + }, + { + "epoch": 2.981300089047195, + "grad_norm": 3.275181770324707, + "learning_rate": 1e-06, + "loss": 0.2727, + "mean_token_accuracy": 0.8986557722091675, + "num_tokens": 909737695.0, + "step": 23436 + }, + { + "epoch": 2.9814272993257855, + "grad_norm": 2.745469808578491, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9020037055015564, + "num_tokens": 909777803.0, + "step": 23437 + }, + { + "epoch": 2.981554509604376, + "grad_norm": 2.871490716934204, + "learning_rate": 1e-06, + "loss": 0.2773, + "mean_token_accuracy": 0.8978289365768433, + "num_tokens": 909822086.0, + "step": 23438 + }, + { + "epoch": 2.9816817198829666, + "grad_norm": 3.078728675842285, + "learning_rate": 1e-06, + "loss": 0.258, + "mean_token_accuracy": 0.9052932858467102, + "num_tokens": 909857224.0, + "step": 23439 + }, + { + "epoch": 2.981808930161557, + "grad_norm": 2.9974703788757324, + "learning_rate": 1e-06, + "loss": 0.2811, + "mean_token_accuracy": 0.8970383405685425, + "num_tokens": 909894987.0, + "step": 23440 + }, + { + "epoch": 2.9819361404401477, + "grad_norm": 2.8219900131225586, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.9072791337966919, + "num_tokens": 909934375.0, + "step": 23441 + }, + { + "epoch": 2.982063350718738, + "grad_norm": 2.7578482627868652, + "learning_rate": 1e-06, + "loss": 0.2348, + "mean_token_accuracy": 0.9118824601173401, + "num_tokens": 909971363.0, + "step": 23442 + }, + { + "epoch": 2.9821905609973287, + "grad_norm": 2.722402811050415, + "learning_rate": 1e-06, + "loss": 0.2339, + "mean_token_accuracy": 0.911856472492218, + "num_tokens": 910012392.0, + "step": 23443 + }, + { + "epoch": 2.9823177712759192, + "grad_norm": 3.2468273639678955, + "learning_rate": 1e-06, + "loss": 0.2528, + "mean_token_accuracy": 0.9046201109886169, + "num_tokens": 910044587.0, + "step": 23444 + }, + { + "epoch": 2.9824449815545098, + "grad_norm": 2.699077844619751, + "learning_rate": 1e-06, + "loss": 0.2563, + "mean_token_accuracy": 0.9035249948501587, + "num_tokens": 910087499.0, + "step": 23445 + }, + { + "epoch": 2.9825721918331, + "grad_norm": 2.9579761028289795, + "learning_rate": 1e-06, + "loss": 0.2864, + "mean_token_accuracy": 0.89473557472229, + "num_tokens": 910130097.0, + "step": 23446 + }, + { + "epoch": 2.982699402111691, + "grad_norm": 2.937979221343994, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9010018110275269, + "num_tokens": 910168168.0, + "step": 23447 + }, + { + "epoch": 2.982826612390281, + "grad_norm": 2.6722824573516846, + "learning_rate": 1e-06, + "loss": 0.2582, + "mean_token_accuracy": 0.9047916531562805, + "num_tokens": 910209606.0, + "step": 23448 + }, + { + "epoch": 2.982953822668872, + "grad_norm": 3.0188186168670654, + "learning_rate": 1e-06, + "loss": 0.2764, + "mean_token_accuracy": 0.8964049220085144, + "num_tokens": 910246360.0, + "step": 23449 + }, + { + "epoch": 2.983081032947462, + "grad_norm": 2.7559850215911865, + "learning_rate": 1e-06, + "loss": 0.2636, + "mean_token_accuracy": 0.904245138168335, + "num_tokens": 910288907.0, + "step": 23450 + }, + { + "epoch": 2.983208243226053, + "grad_norm": 2.574716806411743, + "learning_rate": 1e-06, + "loss": 0.2524, + "mean_token_accuracy": 0.9079791307449341, + "num_tokens": 910333703.0, + "step": 23451 + }, + { + "epoch": 2.983335453504643, + "grad_norm": 2.7192673683166504, + "learning_rate": 1e-06, + "loss": 0.2427, + "mean_token_accuracy": 0.9129416346549988, + "num_tokens": 910374371.0, + "step": 23452 + }, + { + "epoch": 2.9834626637832335, + "grad_norm": 2.701021671295166, + "learning_rate": 1e-06, + "loss": 0.2867, + "mean_token_accuracy": 0.8949742317199707, + "num_tokens": 910419923.0, + "step": 23453 + }, + { + "epoch": 2.983589874061824, + "grad_norm": 2.7814245223999023, + "learning_rate": 1e-06, + "loss": 0.2473, + "mean_token_accuracy": 0.9079292416572571, + "num_tokens": 910461350.0, + "step": 23454 + }, + { + "epoch": 2.9837170843404146, + "grad_norm": 2.7238895893096924, + "learning_rate": 1e-06, + "loss": 0.2659, + "mean_token_accuracy": 0.9029213190078735, + "num_tokens": 910505449.0, + "step": 23455 + }, + { + "epoch": 2.983844294619005, + "grad_norm": 2.954681873321533, + "learning_rate": 1e-06, + "loss": 0.2895, + "mean_token_accuracy": 0.8963661193847656, + "num_tokens": 910544279.0, + "step": 23456 + }, + { + "epoch": 2.9839715048975957, + "grad_norm": 2.860802412033081, + "learning_rate": 1e-06, + "loss": 0.2593, + "mean_token_accuracy": 0.9053389430046082, + "num_tokens": 910584633.0, + "step": 23457 + }, + { + "epoch": 2.984098715176186, + "grad_norm": 2.7612829208374023, + "learning_rate": 1e-06, + "loss": 0.2451, + "mean_token_accuracy": 0.9109454154968262, + "num_tokens": 910620788.0, + "step": 23458 + }, + { + "epoch": 2.9842259254547767, + "grad_norm": 2.6399059295654297, + "learning_rate": 1e-06, + "loss": 0.2447, + "mean_token_accuracy": 0.9098962545394897, + "num_tokens": 910664121.0, + "step": 23459 + }, + { + "epoch": 2.9843531357333672, + "grad_norm": 2.7933907508850098, + "learning_rate": 1e-06, + "loss": 0.2696, + "mean_token_accuracy": 0.9003214836120605, + "num_tokens": 910705832.0, + "step": 23460 + }, + { + "epoch": 2.9844803460119578, + "grad_norm": 3.158540964126587, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.8918004631996155, + "num_tokens": 910742850.0, + "step": 23461 + }, + { + "epoch": 2.9846075562905483, + "grad_norm": 2.9228837490081787, + "learning_rate": 1e-06, + "loss": 0.2638, + "mean_token_accuracy": 0.9025689959526062, + "num_tokens": 910781901.0, + "step": 23462 + }, + { + "epoch": 2.984734766569139, + "grad_norm": 3.0422258377075195, + "learning_rate": 1e-06, + "loss": 0.2766, + "mean_token_accuracy": 0.899075984954834, + "num_tokens": 910820815.0, + "step": 23463 + }, + { + "epoch": 2.9848619768477294, + "grad_norm": 2.7926673889160156, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.9012529253959656, + "num_tokens": 910867082.0, + "step": 23464 + }, + { + "epoch": 2.98498918712632, + "grad_norm": 2.794771194458008, + "learning_rate": 1e-06, + "loss": 0.2676, + "mean_token_accuracy": 0.9025203585624695, + "num_tokens": 910910002.0, + "step": 23465 + }, + { + "epoch": 2.9851163974049104, + "grad_norm": 2.8690075874328613, + "learning_rate": 1e-06, + "loss": 0.2538, + "mean_token_accuracy": 0.9051381349563599, + "num_tokens": 910950845.0, + "step": 23466 + }, + { + "epoch": 2.985243607683501, + "grad_norm": 3.0693767070770264, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.9071865081787109, + "num_tokens": 910993335.0, + "step": 23467 + }, + { + "epoch": 2.9853708179620915, + "grad_norm": 2.663768768310547, + "learning_rate": 1e-06, + "loss": 0.2298, + "mean_token_accuracy": 0.9154160022735596, + "num_tokens": 911031490.0, + "step": 23468 + }, + { + "epoch": 2.985498028240682, + "grad_norm": 2.9836032390594482, + "learning_rate": 1e-06, + "loss": 0.2661, + "mean_token_accuracy": 0.9019636511802673, + "num_tokens": 911070703.0, + "step": 23469 + }, + { + "epoch": 2.9856252385192725, + "grad_norm": 2.853750228881836, + "learning_rate": 1e-06, + "loss": 0.2505, + "mean_token_accuracy": 0.9068630933761597, + "num_tokens": 911108379.0, + "step": 23470 + }, + { + "epoch": 2.9857524487978626, + "grad_norm": 2.995181083679199, + "learning_rate": 1e-06, + "loss": 0.2838, + "mean_token_accuracy": 0.894890308380127, + "num_tokens": 911149073.0, + "step": 23471 + }, + { + "epoch": 2.9858796590764536, + "grad_norm": 2.680640935897827, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9106481075286865, + "num_tokens": 911190226.0, + "step": 23472 + }, + { + "epoch": 2.9860068693550437, + "grad_norm": 3.3885741233825684, + "learning_rate": 1e-06, + "loss": 0.2788, + "mean_token_accuracy": 0.895598292350769, + "num_tokens": 911223823.0, + "step": 23473 + }, + { + "epoch": 2.9861340796336346, + "grad_norm": 2.665318489074707, + "learning_rate": 1e-06, + "loss": 0.2733, + "mean_token_accuracy": 0.9012240171432495, + "num_tokens": 911271173.0, + "step": 23474 + }, + { + "epoch": 2.9862612899122247, + "grad_norm": 2.5783822536468506, + "learning_rate": 1e-06, + "loss": 0.2453, + "mean_token_accuracy": 0.909104585647583, + "num_tokens": 911317557.0, + "step": 23475 + }, + { + "epoch": 2.9863885001908153, + "grad_norm": 2.834580421447754, + "learning_rate": 1e-06, + "loss": 0.2279, + "mean_token_accuracy": 0.9125769138336182, + "num_tokens": 911350458.0, + "step": 23476 + }, + { + "epoch": 2.986515710469406, + "grad_norm": 2.7564728260040283, + "learning_rate": 1e-06, + "loss": 0.2336, + "mean_token_accuracy": 0.9133033752441406, + "num_tokens": 911386452.0, + "step": 23477 + }, + { + "epoch": 2.9866429207479963, + "grad_norm": 2.9630329608917236, + "learning_rate": 1e-06, + "loss": 0.2865, + "mean_token_accuracy": 0.8927950859069824, + "num_tokens": 911426047.0, + "step": 23478 + }, + { + "epoch": 2.986770131026587, + "grad_norm": 3.0042471885681152, + "learning_rate": 1e-06, + "loss": 0.2767, + "mean_token_accuracy": 0.8978641629219055, + "num_tokens": 911463405.0, + "step": 23479 + }, + { + "epoch": 2.9868973413051774, + "grad_norm": 2.7770140171051025, + "learning_rate": 1e-06, + "loss": 0.2572, + "mean_token_accuracy": 0.9056696891784668, + "num_tokens": 911502088.0, + "step": 23480 + }, + { + "epoch": 2.987024551583768, + "grad_norm": 2.6755032539367676, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.898505687713623, + "num_tokens": 911547745.0, + "step": 23481 + }, + { + "epoch": 2.9871517618623584, + "grad_norm": 2.9522628784179688, + "learning_rate": 1e-06, + "loss": 0.2579, + "mean_token_accuracy": 0.9047059416770935, + "num_tokens": 911586358.0, + "step": 23482 + }, + { + "epoch": 2.987278972140949, + "grad_norm": 3.0162034034729004, + "learning_rate": 1e-06, + "loss": 0.3008, + "mean_token_accuracy": 0.8895084857940674, + "num_tokens": 911625981.0, + "step": 23483 + }, + { + "epoch": 2.9874061824195395, + "grad_norm": 2.896850824356079, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9059987664222717, + "num_tokens": 911659990.0, + "step": 23484 + }, + { + "epoch": 2.98753339269813, + "grad_norm": 2.6792373657226562, + "learning_rate": 1e-06, + "loss": 0.243, + "mean_token_accuracy": 0.9093211889266968, + "num_tokens": 911698661.0, + "step": 23485 + }, + { + "epoch": 2.9876606029767205, + "grad_norm": 2.8879895210266113, + "learning_rate": 1e-06, + "loss": 0.263, + "mean_token_accuracy": 0.9013643264770508, + "num_tokens": 911736416.0, + "step": 23486 + }, + { + "epoch": 2.987787813255311, + "grad_norm": 2.5235588550567627, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9101181030273438, + "num_tokens": 911782159.0, + "step": 23487 + }, + { + "epoch": 2.9879150235339016, + "grad_norm": 2.6311557292938232, + "learning_rate": 1e-06, + "loss": 0.2352, + "mean_token_accuracy": 0.9131414890289307, + "num_tokens": 911821122.0, + "step": 23488 + }, + { + "epoch": 2.988042233812492, + "grad_norm": 2.888157844543457, + "learning_rate": 1e-06, + "loss": 0.278, + "mean_token_accuracy": 0.8961441516876221, + "num_tokens": 911862395.0, + "step": 23489 + }, + { + "epoch": 2.9881694440910826, + "grad_norm": 2.7976279258728027, + "learning_rate": 1e-06, + "loss": 0.256, + "mean_token_accuracy": 0.9051450490951538, + "num_tokens": 911903124.0, + "step": 23490 + }, + { + "epoch": 2.988296654369673, + "grad_norm": 2.7846779823303223, + "learning_rate": 1e-06, + "loss": 0.2434, + "mean_token_accuracy": 0.910892128944397, + "num_tokens": 911943577.0, + "step": 23491 + }, + { + "epoch": 2.9884238646482637, + "grad_norm": 3.0436389446258545, + "learning_rate": 1e-06, + "loss": 0.2452, + "mean_token_accuracy": 0.9102442860603333, + "num_tokens": 911976820.0, + "step": 23492 + }, + { + "epoch": 2.9885510749268542, + "grad_norm": 2.928041458129883, + "learning_rate": 1e-06, + "loss": 0.2594, + "mean_token_accuracy": 0.9046549201011658, + "num_tokens": 912013146.0, + "step": 23493 + }, + { + "epoch": 2.9886782852054448, + "grad_norm": 2.9697859287261963, + "learning_rate": 1e-06, + "loss": 0.2464, + "mean_token_accuracy": 0.9074368476867676, + "num_tokens": 912051384.0, + "step": 23494 + }, + { + "epoch": 2.9888054954840353, + "grad_norm": 2.9754979610443115, + "learning_rate": 1e-06, + "loss": 0.2801, + "mean_token_accuracy": 0.8977736234664917, + "num_tokens": 912089247.0, + "step": 23495 + }, + { + "epoch": 2.9889327057626254, + "grad_norm": 2.809138536453247, + "learning_rate": 1e-06, + "loss": 0.2719, + "mean_token_accuracy": 0.9007343053817749, + "num_tokens": 912129740.0, + "step": 23496 + }, + { + "epoch": 2.9890599160412163, + "grad_norm": 2.7748446464538574, + "learning_rate": 1e-06, + "loss": 0.2556, + "mean_token_accuracy": 0.9056053161621094, + "num_tokens": 912170490.0, + "step": 23497 + }, + { + "epoch": 2.9891871263198064, + "grad_norm": 2.8404412269592285, + "learning_rate": 1e-06, + "loss": 0.2525, + "mean_token_accuracy": 0.9041922092437744, + "num_tokens": 912208209.0, + "step": 23498 + }, + { + "epoch": 2.9893143365983974, + "grad_norm": 3.021878719329834, + "learning_rate": 1e-06, + "loss": 0.3179, + "mean_token_accuracy": 0.8853417634963989, + "num_tokens": 912248987.0, + "step": 23499 + }, + { + "epoch": 2.9894415468769875, + "grad_norm": 3.0372109413146973, + "learning_rate": 1e-06, + "loss": 0.3048, + "mean_token_accuracy": 0.8867992162704468, + "num_tokens": 912288391.0, + "step": 23500 + }, + { + "epoch": 2.989568757155578, + "grad_norm": 2.669907808303833, + "learning_rate": 1e-06, + "loss": 0.2618, + "mean_token_accuracy": 0.9030172824859619, + "num_tokens": 912330583.0, + "step": 23501 + }, + { + "epoch": 2.9896959674341685, + "grad_norm": 2.709818124771118, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9149760603904724, + "num_tokens": 912368348.0, + "step": 23502 + }, + { + "epoch": 2.989823177712759, + "grad_norm": 2.9451286792755127, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9033271074295044, + "num_tokens": 912406516.0, + "step": 23503 + }, + { + "epoch": 2.9899503879913496, + "grad_norm": 2.8549811840057373, + "learning_rate": 1e-06, + "loss": 0.2642, + "mean_token_accuracy": 0.9029498100280762, + "num_tokens": 912443418.0, + "step": 23504 + }, + { + "epoch": 2.99007759826994, + "grad_norm": 2.7610840797424316, + "learning_rate": 1e-06, + "loss": 0.2756, + "mean_token_accuracy": 0.8982150554656982, + "num_tokens": 912484325.0, + "step": 23505 + }, + { + "epoch": 2.9902048085485307, + "grad_norm": 2.879626750946045, + "learning_rate": 1e-06, + "loss": 0.2728, + "mean_token_accuracy": 0.8998068571090698, + "num_tokens": 912522144.0, + "step": 23506 + }, + { + "epoch": 2.990332018827121, + "grad_norm": 2.835172176361084, + "learning_rate": 1e-06, + "loss": 0.2633, + "mean_token_accuracy": 0.9017181396484375, + "num_tokens": 912560397.0, + "step": 23507 + }, + { + "epoch": 2.9904592291057117, + "grad_norm": 2.9692680835723877, + "learning_rate": 1e-06, + "loss": 0.2588, + "mean_token_accuracy": 0.9045932292938232, + "num_tokens": 912598017.0, + "step": 23508 + }, + { + "epoch": 2.9905864393843022, + "grad_norm": 2.976726531982422, + "learning_rate": 1e-06, + "loss": 0.2712, + "mean_token_accuracy": 0.899933934211731, + "num_tokens": 912637096.0, + "step": 23509 + }, + { + "epoch": 2.9907136496628928, + "grad_norm": 2.7750258445739746, + "learning_rate": 1e-06, + "loss": 0.2324, + "mean_token_accuracy": 0.9141354560852051, + "num_tokens": 912672777.0, + "step": 23510 + }, + { + "epoch": 2.9908408599414833, + "grad_norm": 2.6913323402404785, + "learning_rate": 1e-06, + "loss": 0.2403, + "mean_token_accuracy": 0.9108163714408875, + "num_tokens": 912714948.0, + "step": 23511 + }, + { + "epoch": 2.990968070220074, + "grad_norm": 3.092881441116333, + "learning_rate": 1e-06, + "loss": 0.267, + "mean_token_accuracy": 0.9026453495025635, + "num_tokens": 912745720.0, + "step": 23512 + }, + { + "epoch": 2.9910952804986644, + "grad_norm": 2.629889726638794, + "learning_rate": 1e-06, + "loss": 0.2665, + "mean_token_accuracy": 0.900115966796875, + "num_tokens": 912788637.0, + "step": 23513 + }, + { + "epoch": 2.991222490777255, + "grad_norm": 3.0265326499938965, + "learning_rate": 1e-06, + "loss": 0.3085, + "mean_token_accuracy": 0.8882154226303101, + "num_tokens": 912830336.0, + "step": 23514 + }, + { + "epoch": 2.9913497010558454, + "grad_norm": 3.2990827560424805, + "learning_rate": 1e-06, + "loss": 0.2802, + "mean_token_accuracy": 0.9001543521881104, + "num_tokens": 912860807.0, + "step": 23515 + }, + { + "epoch": 2.991476911334436, + "grad_norm": 2.653043746948242, + "learning_rate": 1e-06, + "loss": 0.2343, + "mean_token_accuracy": 0.9135591983795166, + "num_tokens": 912900538.0, + "step": 23516 + }, + { + "epoch": 2.9916041216130265, + "grad_norm": 3.1426706314086914, + "learning_rate": 1e-06, + "loss": 0.2591, + "mean_token_accuracy": 0.902585506439209, + "num_tokens": 912936622.0, + "step": 23517 + }, + { + "epoch": 2.991731331891617, + "grad_norm": 2.8793931007385254, + "learning_rate": 1e-06, + "loss": 0.3147, + "mean_token_accuracy": 0.8858526945114136, + "num_tokens": 912981355.0, + "step": 23518 + }, + { + "epoch": 2.991858542170207, + "grad_norm": 3.1088457107543945, + "learning_rate": 1e-06, + "loss": 0.2883, + "mean_token_accuracy": 0.8952275514602661, + "num_tokens": 913015038.0, + "step": 23519 + }, + { + "epoch": 2.991985752448798, + "grad_norm": 3.095043420791626, + "learning_rate": 1e-06, + "loss": 0.2787, + "mean_token_accuracy": 0.899592936038971, + "num_tokens": 913050641.0, + "step": 23520 + }, + { + "epoch": 2.992112962727388, + "grad_norm": 2.699753999710083, + "learning_rate": 1e-06, + "loss": 0.2459, + "mean_token_accuracy": 0.9100346565246582, + "num_tokens": 913088158.0, + "step": 23521 + }, + { + "epoch": 2.992240173005979, + "grad_norm": 2.7241876125335693, + "learning_rate": 1e-06, + "loss": 0.2754, + "mean_token_accuracy": 0.898371160030365, + "num_tokens": 913132535.0, + "step": 23522 + }, + { + "epoch": 2.992367383284569, + "grad_norm": 2.9169716835021973, + "learning_rate": 1e-06, + "loss": 0.2397, + "mean_token_accuracy": 0.909538745880127, + "num_tokens": 913167899.0, + "step": 23523 + }, + { + "epoch": 2.99249459356316, + "grad_norm": 2.9143078327178955, + "learning_rate": 1e-06, + "loss": 0.2933, + "mean_token_accuracy": 0.8948789834976196, + "num_tokens": 913209301.0, + "step": 23524 + }, + { + "epoch": 2.9926218038417502, + "grad_norm": 2.512327194213867, + "learning_rate": 1e-06, + "loss": 0.2254, + "mean_token_accuracy": 0.9156380891799927, + "num_tokens": 913254430.0, + "step": 23525 + }, + { + "epoch": 2.992749014120341, + "grad_norm": 2.6995997428894043, + "learning_rate": 1e-06, + "loss": 0.2401, + "mean_token_accuracy": 0.9108880758285522, + "num_tokens": 913293160.0, + "step": 23526 + }, + { + "epoch": 2.9928762243989313, + "grad_norm": 3.0408473014831543, + "learning_rate": 1e-06, + "loss": 0.2341, + "mean_token_accuracy": 0.9130351543426514, + "num_tokens": 913326158.0, + "step": 23527 + }, + { + "epoch": 2.993003434677522, + "grad_norm": 2.628385305404663, + "learning_rate": 1e-06, + "loss": 0.2458, + "mean_token_accuracy": 0.9071816205978394, + "num_tokens": 913366824.0, + "step": 23528 + }, + { + "epoch": 2.9931306449561124, + "grad_norm": 2.9723148345947266, + "learning_rate": 1e-06, + "loss": 0.2641, + "mean_token_accuracy": 0.901346743106842, + "num_tokens": 913403033.0, + "step": 23529 + }, + { + "epoch": 2.993257855234703, + "grad_norm": 2.9340906143188477, + "learning_rate": 1e-06, + "loss": 0.287, + "mean_token_accuracy": 0.8962740898132324, + "num_tokens": 913445969.0, + "step": 23530 + }, + { + "epoch": 2.9933850655132934, + "grad_norm": 3.0449411869049072, + "learning_rate": 1e-06, + "loss": 0.2808, + "mean_token_accuracy": 0.9002472162246704, + "num_tokens": 913483980.0, + "step": 23531 + }, + { + "epoch": 2.993512275791884, + "grad_norm": 2.770214796066284, + "learning_rate": 1e-06, + "loss": 0.2461, + "mean_token_accuracy": 0.9067411422729492, + "num_tokens": 913522962.0, + "step": 23532 + }, + { + "epoch": 2.9936394860704745, + "grad_norm": 2.705063819885254, + "learning_rate": 1e-06, + "loss": 0.2606, + "mean_token_accuracy": 0.9035512804985046, + "num_tokens": 913569449.0, + "step": 23533 + }, + { + "epoch": 2.993766696349065, + "grad_norm": 2.920098066329956, + "learning_rate": 1e-06, + "loss": 0.2396, + "mean_token_accuracy": 0.9087594747543335, + "num_tokens": 913602347.0, + "step": 23534 + }, + { + "epoch": 2.9938939066276555, + "grad_norm": 2.670504093170166, + "learning_rate": 1e-06, + "loss": 0.2474, + "mean_token_accuracy": 0.9086095094680786, + "num_tokens": 913642594.0, + "step": 23535 + }, + { + "epoch": 2.994021116906246, + "grad_norm": 2.746846914291382, + "learning_rate": 1e-06, + "loss": 0.2243, + "mean_token_accuracy": 0.9150569438934326, + "num_tokens": 913676023.0, + "step": 23536 + }, + { + "epoch": 2.9941483271848366, + "grad_norm": 2.878868579864502, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9055049419403076, + "num_tokens": 913717790.0, + "step": 23537 + }, + { + "epoch": 2.994275537463427, + "grad_norm": 2.8985135555267334, + "learning_rate": 1e-06, + "loss": 0.2503, + "mean_token_accuracy": 0.9070345759391785, + "num_tokens": 913755701.0, + "step": 23538 + }, + { + "epoch": 2.9944027477420176, + "grad_norm": 2.9473206996917725, + "learning_rate": 1e-06, + "loss": 0.2415, + "mean_token_accuracy": 0.908856987953186, + "num_tokens": 913796522.0, + "step": 23539 + }, + { + "epoch": 2.994529958020608, + "grad_norm": 3.0569465160369873, + "learning_rate": 1e-06, + "loss": 0.25, + "mean_token_accuracy": 0.9066455364227295, + "num_tokens": 913830358.0, + "step": 23540 + }, + { + "epoch": 2.9946571682991987, + "grad_norm": 2.845578670501709, + "learning_rate": 1e-06, + "loss": 0.2498, + "mean_token_accuracy": 0.9062321186065674, + "num_tokens": 913868656.0, + "step": 23541 + }, + { + "epoch": 2.9947843785777892, + "grad_norm": 2.625391960144043, + "learning_rate": 1e-06, + "loss": 0.2567, + "mean_token_accuracy": 0.9046180844306946, + "num_tokens": 913914397.0, + "step": 23542 + }, + { + "epoch": 2.9949115888563798, + "grad_norm": 3.0015366077423096, + "learning_rate": 1e-06, + "loss": 0.2585, + "mean_token_accuracy": 0.9045780301094055, + "num_tokens": 913952323.0, + "step": 23543 + }, + { + "epoch": 2.99503879913497, + "grad_norm": 3.0571839809417725, + "learning_rate": 1e-06, + "loss": 0.2542, + "mean_token_accuracy": 0.9061734676361084, + "num_tokens": 913986839.0, + "step": 23544 + }, + { + "epoch": 2.995166009413561, + "grad_norm": 3.066561698913574, + "learning_rate": 1e-06, + "loss": 0.2732, + "mean_token_accuracy": 0.8977785706520081, + "num_tokens": 914024807.0, + "step": 23545 + }, + { + "epoch": 2.995293219692151, + "grad_norm": 3.0710792541503906, + "learning_rate": 1e-06, + "loss": 0.2944, + "mean_token_accuracy": 0.8906954526901245, + "num_tokens": 914065766.0, + "step": 23546 + }, + { + "epoch": 2.995420429970742, + "grad_norm": 2.9039905071258545, + "learning_rate": 1e-06, + "loss": 0.2476, + "mean_token_accuracy": 0.9098228216171265, + "num_tokens": 914100607.0, + "step": 23547 + }, + { + "epoch": 2.995547640249332, + "grad_norm": 2.8497366905212402, + "learning_rate": 1e-06, + "loss": 0.2683, + "mean_token_accuracy": 0.901276707649231, + "num_tokens": 914138466.0, + "step": 23548 + }, + { + "epoch": 2.995674850527923, + "grad_norm": 2.784302234649658, + "learning_rate": 1e-06, + "loss": 0.2745, + "mean_token_accuracy": 0.8990913033485413, + "num_tokens": 914183539.0, + "step": 23549 + }, + { + "epoch": 2.995802060806513, + "grad_norm": 2.6565685272216797, + "learning_rate": 1e-06, + "loss": 0.2836, + "mean_token_accuracy": 0.8961426019668579, + "num_tokens": 914228499.0, + "step": 23550 + }, + { + "epoch": 2.9959292710851035, + "grad_norm": 2.8174099922180176, + "learning_rate": 1e-06, + "loss": 0.2731, + "mean_token_accuracy": 0.9023385047912598, + "num_tokens": 914271152.0, + "step": 23551 + }, + { + "epoch": 2.996056481363694, + "grad_norm": 2.7447330951690674, + "learning_rate": 1e-06, + "loss": 0.2711, + "mean_token_accuracy": 0.9023844003677368, + "num_tokens": 914310673.0, + "step": 23552 + }, + { + "epoch": 2.9961836916422846, + "grad_norm": 3.007983922958374, + "learning_rate": 1e-06, + "loss": 0.2698, + "mean_token_accuracy": 0.903508186340332, + "num_tokens": 914347520.0, + "step": 23553 + }, + { + "epoch": 2.996310901920875, + "grad_norm": 3.0708324909210205, + "learning_rate": 1e-06, + "loss": 0.2868, + "mean_token_accuracy": 0.894406259059906, + "num_tokens": 914383509.0, + "step": 23554 + }, + { + "epoch": 2.9964381121994657, + "grad_norm": 2.6807172298431396, + "learning_rate": 1e-06, + "loss": 0.2751, + "mean_token_accuracy": 0.8990782499313354, + "num_tokens": 914425867.0, + "step": 23555 + }, + { + "epoch": 2.996565322478056, + "grad_norm": 3.2171249389648438, + "learning_rate": 1e-06, + "loss": 0.2564, + "mean_token_accuracy": 0.90895676612854, + "num_tokens": 914458988.0, + "step": 23556 + }, + { + "epoch": 2.9966925327566467, + "grad_norm": 2.910531520843506, + "learning_rate": 1e-06, + "loss": 0.2926, + "mean_token_accuracy": 0.8951476216316223, + "num_tokens": 914498239.0, + "step": 23557 + }, + { + "epoch": 2.9968197430352372, + "grad_norm": 2.949512243270874, + "learning_rate": 1e-06, + "loss": 0.2472, + "mean_token_accuracy": 0.9087216854095459, + "num_tokens": 914531638.0, + "step": 23558 + }, + { + "epoch": 2.9969469533138278, + "grad_norm": 2.56184720993042, + "learning_rate": 1e-06, + "loss": 0.2281, + "mean_token_accuracy": 0.9142738580703735, + "num_tokens": 914573508.0, + "step": 23559 + }, + { + "epoch": 2.9970741635924183, + "grad_norm": 2.8003861904144287, + "learning_rate": 1e-06, + "loss": 0.2203, + "mean_token_accuracy": 0.9174621105194092, + "num_tokens": 914610374.0, + "step": 23560 + }, + { + "epoch": 2.997201373871009, + "grad_norm": 2.8597702980041504, + "learning_rate": 1e-06, + "loss": 0.2571, + "mean_token_accuracy": 0.9048904776573181, + "num_tokens": 914650034.0, + "step": 23561 + }, + { + "epoch": 2.9973285841495994, + "grad_norm": 3.3141121864318848, + "learning_rate": 1e-06, + "loss": 0.2782, + "mean_token_accuracy": 0.900719165802002, + "num_tokens": 914680929.0, + "step": 23562 + }, + { + "epoch": 2.99745579442819, + "grad_norm": 3.0456485748291016, + "learning_rate": 1e-06, + "loss": 0.2697, + "mean_token_accuracy": 0.89992755651474, + "num_tokens": 914716709.0, + "step": 23563 + }, + { + "epoch": 2.9975830047067804, + "grad_norm": 3.0072572231292725, + "learning_rate": 1e-06, + "loss": 0.2843, + "mean_token_accuracy": 0.8963687419891357, + "num_tokens": 914758082.0, + "step": 23564 + }, + { + "epoch": 2.997710214985371, + "grad_norm": 2.844125509262085, + "learning_rate": 1e-06, + "loss": 0.2903, + "mean_token_accuracy": 0.8972874283790588, + "num_tokens": 914802533.0, + "step": 23565 + }, + { + "epoch": 2.9978374252639615, + "grad_norm": 2.8040032386779785, + "learning_rate": 1e-06, + "loss": 0.2413, + "mean_token_accuracy": 0.9114036560058594, + "num_tokens": 914840119.0, + "step": 23566 + }, + { + "epoch": 2.997964635542552, + "grad_norm": 2.797175407409668, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9039477109909058, + "num_tokens": 914877880.0, + "step": 23567 + }, + { + "epoch": 2.9980918458211425, + "grad_norm": 2.627549648284912, + "learning_rate": 1e-06, + "loss": 0.2574, + "mean_token_accuracy": 0.903533935546875, + "num_tokens": 914919174.0, + "step": 23568 + }, + { + "epoch": 2.9982190560997326, + "grad_norm": 3.009533166885376, + "learning_rate": 1e-06, + "loss": 0.2573, + "mean_token_accuracy": 0.904678463935852, + "num_tokens": 914959825.0, + "step": 23569 + }, + { + "epoch": 2.9983462663783236, + "grad_norm": 2.7329962253570557, + "learning_rate": 1e-06, + "loss": 0.2212, + "mean_token_accuracy": 0.9153487682342529, + "num_tokens": 914997016.0, + "step": 23570 + }, + { + "epoch": 2.9984734766569137, + "grad_norm": 2.688417434692383, + "learning_rate": 1e-06, + "loss": 0.2309, + "mean_token_accuracy": 0.9127185940742493, + "num_tokens": 915035425.0, + "step": 23571 + }, + { + "epoch": 2.9986006869355046, + "grad_norm": 2.6162028312683105, + "learning_rate": 1e-06, + "loss": 0.2466, + "mean_token_accuracy": 0.908108115196228, + "num_tokens": 915077667.0, + "step": 23572 + }, + { + "epoch": 2.9987278972140947, + "grad_norm": 2.856940984725952, + "learning_rate": 1e-06, + "loss": 0.2502, + "mean_token_accuracy": 0.9063456058502197, + "num_tokens": 915113545.0, + "step": 23573 + }, + { + "epoch": 2.9988551074926852, + "grad_norm": 2.988110303878784, + "learning_rate": 1e-06, + "loss": 0.2686, + "mean_token_accuracy": 0.9030773639678955, + "num_tokens": 915152999.0, + "step": 23574 + }, + { + "epoch": 2.9989823177712758, + "grad_norm": 2.8763272762298584, + "learning_rate": 1e-06, + "loss": 0.2551, + "mean_token_accuracy": 0.9041581153869629, + "num_tokens": 915188322.0, + "step": 23575 + }, + { + "epoch": 2.9991095280498663, + "grad_norm": 2.9131953716278076, + "learning_rate": 1e-06, + "loss": 0.2637, + "mean_token_accuracy": 0.9029244780540466, + "num_tokens": 915224951.0, + "step": 23576 + }, + { + "epoch": 2.999236738328457, + "grad_norm": 2.898937940597534, + "learning_rate": 1e-06, + "loss": 0.2469, + "mean_token_accuracy": 0.907992959022522, + "num_tokens": 915260752.0, + "step": 23577 + }, + { + "epoch": 2.9993639486070474, + "grad_norm": 2.8803629875183105, + "learning_rate": 1e-06, + "loss": 0.291, + "mean_token_accuracy": 0.894351601600647, + "num_tokens": 915303450.0, + "step": 23578 + }, + { + "epoch": 2.999491158885638, + "grad_norm": 2.898926019668579, + "learning_rate": 1e-06, + "loss": 0.2535, + "mean_token_accuracy": 0.9065767526626587, + "num_tokens": 915338551.0, + "step": 23579 + }, + { + "epoch": 2.9996183691642284, + "grad_norm": 3.085141658782959, + "learning_rate": 1e-06, + "loss": 0.2814, + "mean_token_accuracy": 0.897841215133667, + "num_tokens": 915376329.0, + "step": 23580 + }, + { + "epoch": 2.999745579442819, + "grad_norm": 3.0024609565734863, + "learning_rate": 1e-06, + "loss": 0.2386, + "mean_token_accuracy": 0.9125669002532959, + "num_tokens": 915409691.0, + "step": 23581 + }, + { + "epoch": 2.9998727897214095, + "grad_norm": 2.819023609161377, + "learning_rate": 1e-06, + "loss": 0.2398, + "mean_token_accuracy": 0.9109331369400024, + "num_tokens": 915448832.0, + "step": 23582 + }, + { + "epoch": 3.0, + "grad_norm": 2.8904545307159424, + "learning_rate": 1e-06, + "loss": 0.2521, + "mean_token_accuracy": 0.9067351818084717, + "num_tokens": 915483506.0, + "step": 23583 + }, + { + "epoch": 3.0, + "step": 23583, + "total_flos": 5.7509457374759354e+19, + "train_loss": 0.3138526783479623, + "train_runtime": 47000.1043, + "train_samples_per_second": 8.028, + "train_steps_per_second": 0.502 + } + ], + "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": 5.7509457374759354e+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..99b82b8 --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28b7e9e15e9344e32296d84cec6068ba137940c6ca9ad0ffbd29008765f59675 +size 12625